# Why No One Agrees on What Their AI Tokens Cost - Pratyusha Duvvuri, Cisco

Pratyusha Duvvuri, Cisco

[Talk page](https://ai.engineer/talks/KdG9OSOtJOA-why-no-one-agrees-on-what-their) · [Video](https://www.youtube.com/watch?v=KdG9OSOtJOA)

## Why No One Agrees on What Their AI Tokens Cost

An engineer wants to know whether a better model is worth the money. Finance wants to know who owns the bill, even after the provider revises it. Pratyusha Duvvuri walks through the usage data, pricing agreements and two different company hierarchies behind those answers.

### What are you buying with the extra three points?

[0:12](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=12s)

Pratyusha Duvvuri gave the room seven seconds to think about what *tokenomics* meant. Her team had put the question to people at about 100 companies, from engineers and product managers to CFOs. One person wanted to know what a token cost. Another wanted a financial breakdown of the return on investment for every team. Duvvuri had come to the problem through agent observability and distributed tracing at Galileo. Even with that background, counting usage would get her only partway to the second answer.

The meaning of a large bill was changing, too. Duvvuri recalled an executive suggesting that an employee earning $500,000 ought to spend around $250,000 on tokens. High consumption was a sign that someone was making use of the tools. Then came an account of a company spending $1.5 million in a week and the alarm that figure caused. Once efficiency enters the conversation, the interesting decision is which calls merit the expense.

Galileo had an internal task on which a cheaper model scored 96% on the team’s evaluation. Running the evaluation with a more expensive model raised the score to 99%. The team kept the cheaper one. Those three percentage points had too little value on that internal task to justify switching. Duvvuri would weigh the same improvement differently for an agent talking to customers, where a better response could improve someone’s experience. That was a judgment about where she would spend more. The measured gain was on the internal evaluation.

Even the price side of that decision depends on who signed the contract. Customers described agreements at four cents per credit and others at three cents. Another agreement charged nothing up to a $30 threshold, then applied a rate of ten cents per credit after the threshold was crossed. In that last case, knowing how many credits a call used still leaves a question: where was the customer relative to the threshold? A single published rate cannot reproduce all three bills.

### Three people open the dashboard for different reasons

[4:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=282s)

The customer interviews gave the team three audiences to design for. Each had a decision to make while the meter was still running.

- **Engineer.** With a $10,000 budget for the month, how much have I used? Seeing which models and skills account for consumption helps an engineer decide where a cheaper model might do the job.
- **Engineering manager.** The observability team is delivering results and overspending. The product team is spending less and missing its promised KPIs. Ranking the two by cost would put the cheaper team first and leave the manager’s question unanswered. The manager wants to see how usage and adoption are changing in each team, then consider those trends alongside the results.
- **Finance.** A CFO needs budgets and alerts early enough to act, and an account of the return that will make sense to the CEO or board. Duvvuri likened metered usage to a taxi ride. You know the meter is running, but you learn the fare at the end. That makes waiting for the invoice an uncomfortable way to manage a budget.

### A provider total still needs an owner

[6:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=405s)

A provider portal could already show Duvvuri costs, skills used and sometimes a projection. As an engineer using several services, she still had to move between portals to piece together her usage. Adding their totals introduces another problem. In her example, a provider initially reports a monthly charge of $100,000. Its reconciliation process later puts the charge at $120,000. A dashboard that preserved only the first figure would now understate that same month’s cost by $20,000.

Even the corrected total leaves finance looking for an owner. A workforce system such as Workday traces a person to a manager, that manager to another manager, and so on up to the CEO. Finance organizes costs according to its own reporting hierarchy, taking into account categories such as operating expenses (OPEX) and cost of goods sold (COGS). Knowing who manages the person who incurred a charge answers only one of those ownership questions.

Duvvuri’s working definition of tokenomics is “the observability and interpretability of token usage and cost.” Observability gets the usage and charges into view across providers. Interpretability lets someone ask which models contributed to a charge, how to categorize the work for a particular use case and what return might justify it. The distinction matters when designing a dashboard. A total can be accurate and still give a manager too little information to make a decision.

### Daily provider data and two company hierarchies

[9:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=560s)

How does one person’s usage become both a team total and a finance report? Onboarding brings together three inputs. A workforce hierarchy from Workday, ADP or an equivalent system supplies reporting relationships. Provider API keys or account identifiers let the product fetch usage data every day, giving the dashboard a one-day refresh cadence. The customer also supplies its finance hierarchy, since the accounting structure differs from company to company.

The workforce import could be a CSV with six required fields. Customers were willing to share information such as names, IDs and job functions while keeping details such as tenure private. A limited export made it possible to supply the reporting structure without handing over an employee’s entire history. Customers who supplied more attributes could break down usage in more ways.

![The slide lists workforce hierarchy data, provider account details for usage reporting, and an optional finance hierarchy. Three illustrated colleagues review charts beside the list.](https://ai.engineer/api/talk-essay-media/60a566390fcae5f7902b4303cb23b84c202203b98f41fa427a536b0f40a1ad27)

The three inputs have different owners: the company reporting structure, provider accounts, and the finance hierarchy.

How does the same spend reach two different owners?
Both views use the provider figures. The workforce tree determines which leader’s total includes a person’s usage. Department membership places that person’s cost in the finance tree.
- Workforce hierarchy — The customer supplies people and their reporting relationships.
- Provider data — The product retrieves usage and cost across providers on a daily cadence.
- Finance hierarchy — The customer’s cost map has departments at the ends of its branches.
- Leader view — A leader can follow a team total down to an individual’s usage.
- Finance view — Costs attributed to department members contribute to the appropriate financial totals.
- Workforce hierarchy → Leader view (Groups people under managers)
- Provider data → Leader view (Supplies usage and spend)
- Provider data → Finance view (Supplies costs to allocate)
- Finance hierarchy → Finance view (Groups costs by department)

### Reading adoption, cost and the projected range

[10:55](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=655s)

The demo opens in the leader view, designed for engineers and engineering managers and also accessible to finance. Actual company figures are blurred. Selecting a week or month to date sets the period for reading the cost and adoption trends, with a fiscal calendar available for financial reporting. The cost dashboard tracks token consumption and money spent. The adoption dashboard answers a different question about the same period, such as whether the team’s use of a coding tool is growing or declining.

Will the budget last a month, or could it run out in a week? That question from managers and finance teams prompted a forecast preview. Look for the part of the graph that opens out as it projects usage forward. Its width matters: the chart presents a range of possible usage, and Duvvuri tied the wide spread to changes in models and costs. Forecasting was still awaiting public release at the time of the demo.

### Following a leader’s total down to one person

[12:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=767s)

Whose usage is included in an executive’s total? Below the daily cost trends, the leader view starts with the CEO’s direct reports. Duvvuri’s usage contributes to the chief product officer’s branch because she sits somewhere in that reporting chain. Expanding the branch reveals the people who report to that leader. Continuing down reaches Duvvuri and her own reports. At that level, a reader can distinguish the team’s usage from the amount attributable to Duvvuri individually.

Changing the grouping to job family answers a question that would take several trips down the tree. A leader responsible for product, sales and engineering can compare consumption across those functions. At one customer, sales was using more than engineering, a surprise that the leader’s combined total would have hidden. For a closer look, the product breaks cost down by model and by low, medium or high thinking level, and lets the reader filter by person, job function and total cost.

### Finance follows the cost down to a department

[14:39](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=879s)

Where does the same charge land for a financial controller or analyst? The finance view opens a cost map organized around the company’s financial reporting structure. Follow a branch to its end and you reach a department. Everyone from an engineer to the CEO belongs to a department, so each person’s cost rolls up through a node in that tree. This gives finance a way to inspect the spending behind a total using the structure it already uses to report OPEX and COGS.

Role-based access control limits what any one reader can inspect. A manager responsible for team A can see that team’s usage. Looking across to team B requires administrator access in Duvvuri’s example. The product can therefore aggregate costs across the company while restricting which underlying records a manager can open.

### After launch, both the bill and its owner can change

[15:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=947s)

The team went from its first customer conversations to general availability in 10 weeks and had paying customers. Those conversations also surfaced a request that went beyond allocating cost. Could a team tie its token usage to a customer it landed, a migration it completed or interest in a new product? Duvvuri described this as ROI inference. The dashboard could identify who incurred the cost. Establishing how much that spend contributed to a business outcome remained an open question.

New requests could change the data model after people had already come to rely on it. Duvvuri illustrated the scale with a hypothetical organization of 90,000 people and 10,000 users accessing the platform. A change to accommodate a new use case still has to serve the people using the existing views. And it has to leave an engineer checking a model choice, a manager comparing teams and a controller tracking a budget able to find their own answers.

Reconciliation makes even a backward-looking question tricky. When the provider revises last month’s charge, the next fetch can change a figure finance has already used. Meanwhile, an employee can leave or move to a new manager. Applying today’s reporting tree to last month’s usage can change which manager appears to have incurred the cost, even if the provider leaves the charge alone. That leaves a concrete question for anyone reading a historical total: does a change reflect a revised bill, a change in who owns the spend, or both?

![Overlapping reports show 100 credits for day 1, 180 for days 1–2, and 240 for days 1–3. The three-day total is 240 credits. Adding the reports would double-count usage. Reconciliation, custom deals, and changing ownership can affect the reported figure.](https://ai.engineer/api/talk-essay-media/9553323ad42b7981f68c5ba04e0b07676f59f06437f82462520ab205e0ff2814)

The three reporting windows overlap. The 240-credit total already contains the earlier windows.

## Key takeaways

- Model choice depends on the value of an improvement for the task. A rise from 96% to 99% on an internal evaluation was real and still too small to justify the more expensive model for that use.
- Interpreting a charge requires the customer’s pricing terms and the relevant ownership hierarchy. A per-credit threshold changes what usage costs. Reporting lines and departments determine whose total includes it.
- A daily refresh can bring in a correction to an earlier bill. Changes in reporting relationships raise a separate question about how historical team costs should appear.

## Slides

### Who cares about Tokenomics?

An engineer weighs which model fits the task and the usage limit. An engineering manager looks at how teams use AI to decide where to put budget and support. Finance asks who paid for the work and whether the return justifies the cost.

[5:30](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=330s) [An engineer asks about tool choice and usage limits, an engineering manager asks about adoption and budget, and a finance representative asks who owns the spending and what return it delivers.](https://assets.ai.engineer/sites/aiecode2025/sha256/519dfc51c734c85b84bd8c49c8e1b1c829e398295e121948362118d7033c3555.webp)

Who cares about Tokenomics?

Engineer
Which tools, skills and models fit my work?
How can I get more from my usage limit?

Engineering manager
Who is using AI, and how?
How should I allocate budget and support?

Finance
Who owns the spend?
What work did it fund?
What return justifies a budget or exception?

03

### Doesn’t this already exist?

Five laptops show five separate reports. To see a team’s AI usage, someone has to piece together reports across providers, accounts and workspaces. Finance also needs to know whose budget covers that usage.

[6:44](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=404s) [Five laptops show separate reports. The bullets explain that data is split across providers, accounts and workspaces, making it hard to see usage across teams or assign spending to an owner.](https://assets.ai.engineer/sites/aiecode2025/sha256/88eb108e19778d6a55145c44fb2ecc43fb9f9b661d37f3cb72db89e86482d70f.webp)

Doesn’t this already exist?

Not really...

• Separate reports for each provider
• Multiple accounts and workspaces within a tool
• No single view across tools, people and teams
• Finance needs spend mapped to ownership

04

### What a company brings in

Provider credentials let the product fetch AI usage daily. The company’s org chart ties that usage to people and departments. An optional finance hierarchy identifies who owns the cost.

[9:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=562s) [The slide lists workforce hierarchy data, provider account details for usage reporting, and an optional finance hierarchy. Three illustrated colleagues review charts beside the list.](https://assets.ai.engineer/sites/aiecode2025/sha256/60a566390fcae5f7902b4303cb23b84c202203b98f41fa427a536b0f40a1ad27.webp)

What a company brings in

01  Workforce hierarchy
People, managers and departments
CSV, or a directory integration route

02  Provider accounts
Organization / workspace ID and reporting API credentials

03  Finance hierarchy
Optional financial ownership structure

05

### What we’re adding next

The team plans to help customers forecast AI spending and set budget limits. To estimate the return on that spending, it wants to connect token usage to outcomes such as a completed migration or a new customer.

[15:48](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=948s) [Beside a meeting photo, four bullets cover forecasting, alerts and budget limits, inferring ROI from evidence about the work, provider telemetry and possible device collection, and finance auditing and reporting across quarters.](https://assets.ai.engineer/sites/aiecode2025/sha256/e1afd96645b24a88df7aefa0f0f0ac6d2b4833c37eaa8a466e9e69003ccd2a62.webp)

What we’re adding next

• Forecasting, alerts and budget limits
• ROI inference from evidence about the work
• Provider telemetry and possible device collection
• Finance auditing and reporting across quarters

07

### Challenges

Duvvuri shifts from incoming feature requests to the challenges her team faced while building the product.

[16:39](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=999s) [A dark section divider displays the word “Challenges” in large white type.](https://assets.ai.engineer/sites/aiecode2025/sha256/f167356462603fb3896c9b8e619310e33951990cf307d5f42062f104658ce686.webp)

Challenges

08

### Challenge 1: Startup speed, Cisco scale

Four people gather around a laptop, with a much larger workforce behind them. The team had to design for Cisco’s workforce while still learning what Cisco needed. A new request could force them to rethink the data model.

[16:43](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1003s) [Beside an illustration of a team around a laptop, three bullets describe learning from Cisco as a customer, adapting to changing requirements, and designing for a large workforce. The slide asks how to scale a product still being defined.](https://assets.ai.engineer/sites/aiecode2025/sha256/7073c7a4c0b71b3ae088846a92216407d6017624d53e0fa30b79dda1558660e1.webp)

Challenge 1: Startup speed, Cisco scale

• Learn from Cisco as a customer
• Iterate while requirements change
• Design for a huge workforce

How do we scale a product we’re still defining?

09

### Challenge 2: Three audiences, one product

The provider’s usage data maps to an individual engineer. From there, engineering managers can trace usage to a team, while finance can follow the same usage through the engineer’s department to the responsible finance group and budget.

[17:30](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1050s) [Provider usage points to a person. One branch connects the person to their team and manager. The other connects the person to a department, then to its finance group and budget.](https://assets.ai.engineer/sites/aiecode2025/sha256/5de5bcdcaee041b06e40adfa943ecef6449c0739b75019ecb5eee1ac66e853c9.webp)

Challenge 2: Three audiences, one product

The management tree and the finance tree answer different ownership questions

Provider usage
Person
Team and manager
Department
Finance group and budget

10

### Challenge 3: Reports change. Hierarchies change. Data changes…

In the example, the three-day total of 240 credits already includes the 180 reported for days 1–2. Adding the reports counts earlier usage again. Reconciliation, custom deals and changes in ownership can also affect the reported figure.

[17:59](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1079s) [Overlapping reports show 100 credits for day 1, 180 for days 1–2, and 240 for days 1–3. The three-day total is 240 credits. Adding the reports would double-count usage. Reconciliation, custom deals, and changing ownership can affect the reported figure.](https://assets.ai.engineer/sites/aiecode2025/sha256/9553323ad42b7981f68c5ba04e0b07676f59f06437f82462520ab205e0ff2814.webp)

Challenge 3: Reports change. Hierarchies change. Data changes during reconciliation.

Overlapping reporting windows

Report window	Credits
Day 1	100
Days 1–2	180
Days 1–3	240

Illustrative example

240 credits
across the three days

Adding the overlapping reports would double-count usage

Reconciliation updates the data. Custom deals and changing ownership also affect the number we show.

11

### What does Tokenomics mean to you now?

Duvvuri leaves the definition of tokenomics open. She asks readers to connect how many tokens a team uses and who owns that usage with what the tokens cost and what the work delivers in return.

[18:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1116s) [The slide lists usage, ownership, cost, and the ROI and value of the work. A note says the definition is still evolving. A QR code on the right invites viewers to connect on LinkedIn.](https://assets.ai.engineer/sites/aiecode2025/sha256/7964104a9971add4c01c5d885cd724fe61e8e2872bf198c283ff870be36611b3.webp)

What does Tokenomics mean to you now?

Usage
Ownership
Cost
ROI and value of the work

The definition is still evolving...

Connect on LinkedIn

12

## Full transcript

[0:01](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1s) [music]

[0:12](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=12s) for 7 seconds cuz longer than that

[0:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=14s) you'll sleep. Um, close your eyes and I

[0:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=17s) want you to think about what tokconomics

[0:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=20s) means to you. Humor me for a second.

[0:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=22s) Close your eyes and think about that.

[0:30](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=30s) Okay, that's about 7 seconds. Now,

[0:33](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=33s) here's the thing. I'm willing to bet a

[0:35](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=35s) million dollars that no two of you in

[0:37](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=37s) this room have the same answer. Whether

[0:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=40s) it's Hi there or Christy there, like

[0:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=42s) nobody has the same answer here. And I

[0:43](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=43s) can tell you this with confidence

[0:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=45s) because we surveiled at least a hundred

[0:48](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=48s) different customers/companies.

[0:51](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=51s) We talked to engineers, we talked to

[0:53](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=53s) PMs, we talked to engineering managers.

[0:56](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=56s) CFOs,

[0:57](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=57s) like the whole spectrum, right? And the

[1:00](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=60s) variety of answers we got from, yeah,

[1:02](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=62s) it's just about observability, right?

[1:04](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=64s) Like how much what is my token costing

[1:06](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=66s) to no, we want a full financial

[1:08](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=68s) breakdown of the ROI on every single

[1:10](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=70s) team. It was a pretty crazy wide

[1:13](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=73s) spectrum. This is probably a good time

[1:15](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=75s) to tell you that I don't have an answer

[1:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=77s) to what tokconomics really is, but I can

[1:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=80s) humbly try to give you my version of

[1:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=82s) tokconomics.

[1:24](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=84s) It's probably also a good time to say

[1:25](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=85s) hi. I'm Prushia. Um, we well I was at

[1:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=89s) Galileo which was acquired by Splunk

[1:31](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=91s) which itself was acquired by Cisco. So

[1:33](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=93s) we have a Russian doll situation. Um,

[1:35](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=95s) and I was a tech lead for agent

[1:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=96s) observability distributed tracing most

[1:39](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=99s) recently this crazy topic tokconomics.

[1:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=102s) And while I'd love to just jump in and

[1:44](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=104s) show you a product, I want all of us to

[1:46](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=106s) sort of sail together on this crazy

[1:48](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=108s) journey. So let's come to the same page

[1:51](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=111s) on three trends in AI and if you dispute

[1:54](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=114s) me like please do like I'd love to hear

[1:56](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=116s) from you after the session. The first

[1:58](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=118s) one is we all know that tokconomics was

[2:01](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=121s) definitely like a status symbol of some

[2:03](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=123s) sort right or token maxing sorry. Um,

[2:06](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=126s) and what I mean by that is our friend

[2:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=127s) here, Jensen, literally said, this was

[2:09](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=129s) in May, if I remember correctly, if a

[2:12](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=132s) 500k employee is not spending more than

[2:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=134s) or at least around 250K in tokens or

[2:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=137s) token spend, I am terrified. Now though,

[2:21](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=141s) like just yesterday or well, I woke up

[2:23](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=143s) this morning and saw Twitter, there was

[2:25](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=145s) a post from Eric, I forgot his last

[2:27](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=147s) name, the CEO of RAMP, uh, one of them,

[2:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=149s) and he said, uh, our our CFO doesn't

[2:32](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=152s) want me to post about this, but I'm

[2:34](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=154s) going to do it anyway. we spent $1.5

[2:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=156s) million in one week and he freaked out.

[2:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=160s) So there there's this trend towards

[2:43](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=163s) let's talk about efficiency, right? And

[2:46](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=166s) that's also probably a good time to

[2:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=167s) segue into the second point here, which

[2:49](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=169s) is before let's just use the soda model

[2:52](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=172s) for whatever task, right? I don't care

[2:54](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=174s) what I'm going to use Astra Opus 5.5.

[2:57](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=177s) But now it's more about okay how can I

[2:59](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=179s) get the most efficiency out of my tool

[3:02](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=182s) calls out of my model usage and there's

[3:06](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=186s) a nuance here that I want all of us to

[3:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=187s) think about for example at Galileo well

[3:10](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=190s) now Cisco um we had this internal use

[3:13](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=193s) case where we're using a cheaper model

[3:15](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=195s) like a Luna model which gave us 96% in

[3:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=197s) eval

[3:19](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=199s) right um just for fun we ran evals on

[3:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=202s) like Astra instead like GPT6 Astra and

[3:25](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=205s) our eval jump from 96% to 99%. Did we

[3:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=209s) change the model to Astra? No, because I

[3:32](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=212s) don't care. I want to save money, right,

[3:34](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=214s) when I'm inside the company. But if I'm

[3:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=216s) trying to do anything external facing,

[3:38](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=218s) if I'm trying to talk to my customers,

[3:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=220s) use an agentic like harness to get the

[3:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=222s) best utility for my customers, I'm

[3:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=225s) willing to take that 3% hit for that

[3:48](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=228s) extra, you know, like boost of customer

[3:50](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=230s) satisfaction that I will get. The third

[3:52](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=232s) trend here and this is probably the most

[3:55](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=235s) that mo the biggest one that people

[3:57](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=237s) don't think about which is that before

[3:59](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=239s) open anthropic cursor you name it

[4:02](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=242s) whatever model provider they would tell

[4:04](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=244s) you okay here's our cost per credit here

[4:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=247s) are the models this is our license plan

[4:10](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=250s) thank you very much we don't do any

[4:11](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=251s) customization today as we talk to all of

[4:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=254s) the spectrum of like telos fintech

[4:16](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=256s) companies I mean you name it everybody

[4:19](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=259s) has a different like agreement with

[4:21](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=261s) these providers. For some people it's 4

[4:23](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=263s) cents per credit. For some people it's

[4:25](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=265s) three cents per credit. For some people

[4:27](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=267s) it's oh up until $30 we won't charge you

[4:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=269s) anything. But the moment you cross that

[4:30](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=270s) it's 10 cents per credit. Everyone has

[4:33](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=273s) their custom way of defining like costs

[4:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=276s) for tokens which is especially why

[4:39](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=279s) tokconomics starts to become important.

[4:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=282s) Um before I jump into my version of that

[4:44](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=284s) definition once again, we as I mentioned

[4:46](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=286s) before surveiled a hundred different

[4:48](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=288s) companies and different customers in

[4:50](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=290s) each of these different companies,

[4:51](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=291s) right? And I will tell you it is

[4:53](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=293s) extremely hard to bucket all of your

[4:56](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=296s) ICPS into just three categories, but we

[4:58](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=298s) had to try. I mean there's only one way

[5:00](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=300s) you can build a product, right? You got

[5:02](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=302s) to have some version of ICPS. And what

[5:04](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=304s) we came up with are these three. Number

[5:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=307s) one, probably the most that some of us

[5:09](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=309s) could relate to, the engineer, right?

[5:11](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=311s) Number two, the engineering manager. And

[5:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=314s) then the third one, completely something

[5:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=317s) that most tech folks don't think about.

[5:18](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=318s) I will admit it. The finance folks. So,

[5:21](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=321s) the engineer is worried about, okay,

[5:23](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=323s) well, uh, what models should I use? How

[5:25](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=325s) much did I like? I have a $10,000 budget

[5:27](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=327s) for this month. Am I am I blowing past

[5:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=329s) it? Like, where am I on that spectrum?

[5:31](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=331s) What skills am I using? Should I use

[5:33](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=333s) like sonnet instead of opus? You know,

[5:35](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=335s) whatever it is. The engineering manager

[5:37](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=337s) has a lot more concerns. That person is

[5:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=340s) like, "Okay, I have, let's say, an

[5:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=342s) observability team and I have the

[5:43](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=343s) product team." The observability team is

[5:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=345s) getting me results, but they're also

[5:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=347s) blowing past their costs. The product

[5:49](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=349s) team doesn't seem like they're getting

[5:51](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=351s) me, you know, the KPIs that we promised.

[5:54](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=354s) However, the costs are lower. What

[5:56](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=356s) exactly is happening? What's the trend?

[5:58](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=358s) That person is worried about like

[6:00](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=360s) broader usage and adoption. Um, and then

[6:04](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=364s) the one I promised you was completely

[6:05](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=365s) orthogonal. Finance folks, these guys

[6:08](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=368s) are worried about budgets, about

[6:10](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=370s) alerting. They need to justify to their

[6:12](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=372s) bosses or the CEO or the board or

[6:15](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=375s) whoever is there, what exactly is the

[6:18](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=378s) ROI on this spend, right? And the

[6:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=380s) problem with tokconomics, I think I read

[6:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=382s) a really interesting article like again

[6:24](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=384s) this morning um on X. It was uh some guy

[6:27](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=387s) was saying that token usage is sort of

[6:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=389s) like you're driving a taxi with a meter

[6:31](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=391s) and you have no clue like what the taxi

[6:33](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=393s) is going to show you at the end. It's

[6:35](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=395s) sort of like eh we landed on 2 million.

[6:37](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=397s) Oops. Um very different from preplay

[6:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=400s) prepaid plans, right? So we're worried

[6:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=402s) about all three of these folks. And now

[6:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=405s) is a good time to talk about okay well I

[6:48](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=408s) I hear you but we already have

[6:50](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=410s) dashboards. Like I log into my OpenAI

[6:52](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=412s) portal. I can see a dashboard. I can see

[6:54](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=414s) skills used. I can see my cost. I can

[6:55](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=415s) see maybe a projection, sure, but that's

[6:58](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=418s) just OpenAI. Then you have Enthropic,

[7:00](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=420s) then you have GitHub Copilot, then you

[7:02](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=422s) have like four other providers, and then

[7:04](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=424s) how are you having a one-stop shop to

[7:06](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=426s) see all of that information? Plus, if

[7:08](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=428s) you really think about it, I'm a power

[7:10](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=430s) user of those dashboards, right? Before

[7:12](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=432s) well, my my own tok to tokconomics

[7:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=434s) product. So, um before that, I had to a

[7:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=437s) shuffle between all of these and I'm an

[7:19](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=439s) engineer. I'm not like a finance person

[7:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=440s) or an engineering manager. So, you're

[7:23](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=443s) not really servicing all your ICPS with

[7:26](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=446s) those products even though they're good.

[7:28](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=448s) Plus, something I'm going to touch on in

[7:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=449s) like a couple of um a couple of slides

[7:32](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=452s) is that there's this crazy process

[7:33](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=453s) called reconciliation where a lot of

[7:35](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=455s) things that OpenAI might have said, oh,

[7:37](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=457s) you know, we have 100K uh that we're

[7:39](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=459s) charging you for this month might

[7:41](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=461s) actually not be true. They might come

[7:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=462s) back to you and say, oops, it was 120.

[7:44](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=464s) It's just our recon process caught it.

[7:46](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=466s) So, we need a way to catch all of that

[7:48](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=468s) and like give you the one true like

[7:50](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=470s) honest place for you to see all of your

[7:52](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=472s) usage across all of your providers. Um,

[7:55](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=475s) oh, also that last bullet point, finance

[7:57](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=477s) needs spend mapped to ownership. Yeah,

[7:59](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=479s) that's that's a little different. See,

[8:01](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=481s) in in most companies, you have your

[8:03](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=483s) management structure, right? You have

[8:04](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=484s) the workday structure. So, it's me, my

[8:06](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=486s) boss, that person's boss, all the way up

[8:08](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=488s) to, in my case, Chuck Robbins. Um,

[8:11](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=491s) that's not the way finance sees things.

[8:13](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=493s) Sure, it's the same guy. You have like

[8:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=494s) one CEO, right? But the way they map out

[8:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=497s) the finance hierarchy of a company based

[8:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=500s) on opex, COGS, whatever they're thinking

[8:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=502s) about is very different. And their their

[8:24](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=504s) notion of ownership is very different of

[8:26](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=506s) like from workday or my manager like

[8:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=509s) that notion of ownership. If that was a

[8:31](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=511s) mouthful, don't worry, we'll come back

[8:32](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=512s) to that in just a second. Okay, this is

[8:35](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=515s) a good time to talk about my definition

[8:37](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=517s) of tokconomics.

[8:38](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=518s) Tokconomics is the observability and

[8:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=522s) interpretability of token usage and

[8:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=525s) cost.

[8:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=527s) I kind of just like shied away from a

[8:49](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=529s) more complex definition there and tried

[8:51](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=531s) to hide behind these crazy words. But if

[8:53](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=533s) you really think about it, what is

[8:55](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=535s) interpretability? It's ROI. It's trying

[8:58](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=538s) to figure out like which what is a set

[9:00](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=540s) of models I'm using? It's trying to

[9:01](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=541s) figure out the proper usage and

[9:05](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=545s) categorization depending on your use

[9:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=547s) case. The point is that these words are

[9:09](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=549s) very like vague, but when you really

[9:11](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=551s) jump into like your ICPS and think about

[9:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=554s) what they want, that's when you can

[9:16](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=556s) think of what the best product to build

[9:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=557s) is. So, um, before I actually show you

[9:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=560s) the demo itself, I also want to talk to

[9:23](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=563s) you about what we need from you. So,

[9:25](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=565s) let's say I'm going to show you my

[9:26](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=566s) product like, you know, you're going to

[9:27](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=567s) use tokconomics tomorrow. It's going to

[9:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=569s) be great. Before that, the first thing

[9:31](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=571s) we're going to tell you is, hey, we need

[9:32](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=572s) a workday hierarchy from you or ADP or

[9:35](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=575s) whatever like tool you use to store your

[9:37](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=577s) like company structure or tree, we need

[9:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=580s) that information. The second thing we

[9:43](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=583s) need from you is we need your API keys

[9:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=587s) and or ID. So for OpenAI, Anthropic,

[9:49](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=589s) whoever you're using for your provider,

[9:51](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=591s) we need that information. Why? because

[9:53](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=593s) we're going to talk to them every single

[9:55](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=595s) day to get that data and then service

[9:57](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=597s) you like the latest data on like a

[9:58](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=598s) cadence of one day freshness. And the

[10:01](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=601s) third one, the the really gnarly one is

[10:03](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=603s) the finance hierarchy, which differs

[10:05](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=605s) from company to company. And it's it's

[10:08](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=608s) honestly crazy how finance people deal

[10:10](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=610s) with that. Um, okay, let's jump to the

[10:13](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=613s) product.

[10:15](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=615s) By the way, uh, fun fact about workforce

[10:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=617s) hierarchy, people are very sensitive

[10:18](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=618s) about what data they share, which makes

[10:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=620s) sense, right? like they're not going to

[10:21](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=621s) share their entire like workday history.

[10:23](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=623s) So what we've noticed when we talked to

[10:25](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=625s) customers was we just want to share with

[10:27](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=627s) you like the names, the ids, the job

[10:30](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=630s) functions, but I don't want to tell you

[10:31](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=631s) anything else like tenure, duration,

[10:33](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=633s) etc. So based on your use case, we have

[10:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=636s) made our workflow so simple that we

[10:38](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=638s) literally accept a CSV file. We're like,

[10:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=640s) bro, just send us a CSV file with like

[10:43](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=643s) these six things. If you want to give us

[10:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=645s) more information, we'll slice and dice

[10:46](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=646s) and show you more stuff. But at the very

[10:48](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=648s) least, we need these six things from

[10:50](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=650s) you. Anyway, back to the product. Oh, I

[10:55](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=655s) I had to do this thing where I blurred

[10:59](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=659s) the information because I didn't want

[11:01](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=661s) like to get fired by Galile or Cisco and

[11:04](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=664s) they did not give me permission to show

[11:05](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=665s) you the actual numbers. So, sorry about

[11:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=667s) that. But apart from the actual numbers,

[11:09](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=669s) you have everything else in terms of the

[11:10](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=670s) product showcased here. So as you log on

[11:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=674s) and if I I'm running out of time feel

[11:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=677s) free to like pause me for questions

[11:18](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=678s) towards the end. Um you log into the

[11:21](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=681s) tokconomics dashboard and the first

[11:23](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=683s) thing you see here is different ranges

[11:26](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=686s) to filter right like you can filter

[11:28](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=688s) monthto date. You can filter by like

[11:30](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=690s) weeks. This is a fiscal calendar for

[11:32](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=692s) finance folks. Notice that on the top is

[11:35](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=695s) this right? Yeah top right you see the

[11:37](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=697s) leader view. This is not for finance

[11:39](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=699s) folks although they can access this

[11:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=700s) view. This is for the engineers and the

[11:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=702s) engineering managers. Um as you see here

[11:46](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=706s) there is the cost dashboard and the

[11:49](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=709s) adoption dashboard. Cost is well as the

[11:52](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=712s) name suggests how much money am I

[11:53](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=713s) spending the tokens etc. Adoption is

[11:56](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=716s) very different. It's hey is is my team

[11:59](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=719s) adopting these AI tools are they is the

[12:01](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=721s) codeex usage increasing decreasing like

[12:03](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=723s) you see these like red numbers here

[12:06](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=726s) trends. That's the trend over like

[12:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=727s) whatever time range you've chosen for

[12:09](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=729s) adoption and cost increase and decrease.

[12:12](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=732s) And then this one is actually one of my

[12:13](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=733s) favorites. We got this after talking to

[12:15](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=735s) customers where a lot of folks mention

[12:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=737s) the biggest stresses for both

[12:19](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=739s) engineering managers and finance folks

[12:21](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=741s) is forecasting. People don't know like

[12:23](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=743s) are they going to blow through their

[12:25](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=745s) budget in like a week or a month or like

[12:27](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=747s) what that looks like. And so this

[12:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=749s) ghostlike graph that you see here, this

[12:31](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=751s) weird like opening graph, um yeah,

[12:33](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=753s) that's the projected usage. And if you

[12:35](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=755s) think there's a lot of variance, I agree

[12:37](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=757s) with you. But that's also because that

[12:39](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=759s) just is how the AI world is today with

[12:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=762s) like new things coming out, new models

[12:43](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=763s) coming out and cost. Um, so if we look

[12:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=767s) down here, this is the probably the most

[12:49](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=769s) colorful part of the product where you

[12:52](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=772s) actually see the cost trends over each

[12:54](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=774s) day, right? But then here we have Chuck

[12:56](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=776s) Robbins and everyone under Chuck Robbins

[12:59](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=779s) uh as like direct reports, we see all of

[13:01](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=781s) their usage roll up. So I think I'm

[13:03](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=783s) under this guy Ju. Let me see if there's

[13:05](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=785s) like a slightly clearer view. Okay. So,

[13:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=787s) Ju is our CPO, chief product officer. I

[13:10](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=790s) report to him in some way. And my usage

[13:13](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=793s) is under Juu Patel's usage, right? But

[13:15](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=795s) and Ju Patel obviously reports to the

[13:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=797s) CEO. So, you have that. And if you like

[13:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=800s) scroll a little past, let me see if I

[13:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=802s) can tease my way here. Yep. You see that

[13:25](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=805s) Juu has people that report to him and

[13:27](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=807s) and it goes on and on and on. And then

[13:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=809s) you can very fast because that's very

[13:31](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=811s) important for user experience. see each

[13:34](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=814s) person's usage in their team level and

[13:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=816s) also like an individual usage as you

[13:38](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=818s) keep going down this this path. Um,

[13:41](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=821s) another really cool thing and this is

[13:43](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=823s) just me continually going down until I

[13:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=825s) reach me and people who report to me.

[13:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=827s) But what's also cool here is you see the

[13:50](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=830s) consumption by job family. Think of my

[13:53](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=833s) manager's manager's manager, right? That

[13:55](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=835s) guy is responsible for product team, a

[13:57](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=837s) sales team and uh the engineering team.

[14:00](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=840s) That person wants to know, okay, well,

[14:02](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=842s) uh, is my sales team using more tokens

[14:05](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=845s) or have a higher cost than my

[14:06](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=846s) engineering team? That's crazy, right?

[14:08](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=848s) Funny enough, that was actually one of

[14:09](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=849s) our customer stories where like sales

[14:11](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=851s) was using more than engineering and that

[14:13](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=853s) was a little little worrisome. Um, but

[14:15](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=855s) that's exactly the kind of slicing and

[14:16](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=856s) dicing that we provide to you, right,

[14:18](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=858s) with this sort of like one-stop shop.

[14:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=860s) Um, the rest of it is more intuitive, I

[14:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=862s) would say. Like, yeah, you want to know

[14:24](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=864s) your cost per model, per thinking

[14:26](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=866s) pattern, high, low, medium. uh you want

[14:28](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=868s) to be able to filter by the person,

[14:30](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=870s) their job function, their total cost,

[14:34](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=874s) etc. I want to I want to spend more time

[14:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=876s) on this, but I know we're running out of

[14:37](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=877s) time, so I'm going to move towards the

[14:39](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=879s) finance view, which again, as someone

[14:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=882s) coming from a pure engineering

[14:43](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=883s) background was a little crazy. Um, but

[14:48](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=888s) once you really understand how finance

[14:50](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=890s) works and like the COGS and OPEX

[14:51](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=891s) reporting structure, it's it's not that

[14:53](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=893s) bad. It is also a hierarchy of some

[14:55](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=895s) sort. So we have the same thing here for

[14:57](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=897s) finance. We have um their finance

[15:00](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=900s) hierarchy cost map uh a tree with

[15:03](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=903s) different nodes and then all the way at

[15:05](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=905s) the end you hit departments. And when

[15:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=907s) you hit departments all the way from

[15:10](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=910s) someone like me to Chuck Robbins the

[15:12](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=912s) CEO, everybody is part of a department.

[15:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=914s) And so everybody's cost rolls up into

[15:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=917s) some node in this tree. And then you

[15:19](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=919s) have a financial controller and an

[15:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=920s) analyst. And I can keep going but that's

[15:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=922s) going to like make you guys get a little

[15:23](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=923s) lost in my explanation. But the point is

[15:26](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=926s) you have strict rback. You have strict

[15:28](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=928s) viewership rules. Like if I'm a manager

[15:30](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=930s) of team A, I should not be able to see

[15:32](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=932s) team B's usage unless I'm an admin. We

[15:35](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=935s) take care of all of that in our product

[15:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=936s) where you only see what you're supposed

[15:38](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=938s) to see.

[15:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=940s) Um though what was left in this demo was

[15:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=942s) honestly just me talking through the

[15:43](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=943s) settings, but for the purposes of

[15:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=945s) interest and time, we will skip that

[15:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=947s) part. Um this is a very very evolving

[15:50](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=950s) product. I know I spoke through a lot,

[15:52](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=952s) but you guys will be alarmed if I say we

[15:54](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=954s) went from zero, like talking to

[15:56](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=956s) customers, right? That that zero to

[15:59](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=959s) building a full-fledged product and

[16:01](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=961s) shipping it to GA in 10 weeks.

[16:04](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=964s) That was an insane sprint, off the

[16:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=967s) record, [redacted], but it

[16:09](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=969s) is what it is. Uh, and we're getting

[16:11](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=971s) paid. So, um, I will say that with our

[16:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=974s) customer like outreach, we're getting a

[16:15](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=975s) lot of like requests. As I mentioned,

[16:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=977s) forecasting, that's something that's not

[16:18](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=978s) public yet, so please don't go talk

[16:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=980s) about it, but it's going to come out in

[16:21](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=981s) [redacted]. Um, we

[16:23](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=983s) have ROI inference. So, what how do you

[16:26](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=986s) tie business outcomes, whether that's

[16:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=989s) new customers landed, a migration

[16:31](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=991s) completed, a new product that's

[16:32](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=992s) attracting some eyeballs, how do you tie

[16:34](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=994s) that to token usage? And I can keep

[16:37](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=997s) going on like all the feature requests

[16:39](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=999s) that are coming, but I want to spend at

[16:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1000s) least a little bit of time on the

[16:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1002s) challenges we faced.

[16:44](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1004s) So the way I would describe tokconomics,

[16:46](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1006s) right, and at this point I think you'll

[16:48](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1008s) understand me when I say this. You're

[16:50](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1010s) riding or sailing, I don't know what the

[16:52](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1012s) right word is, a ship where you don't

[16:54](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1014s) even know the destination and it's a big

[16:56](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1016s) ship. So it's very hard to like steer

[16:58](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1018s) your product that you're building for

[17:00](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1020s) scale when you don't even know the

[17:02](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1022s) destination because it's an evolving

[17:03](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1023s) subject. And the engineers in the room

[17:06](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1026s) will be able to empathize with me when I

[17:08](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1028s) say imagine you have like 90,000 like

[17:11](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1031s) people that you're servicing, right? you

[17:12](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1032s) have this like data model built, you

[17:14](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1034s) you're like so ready and then you get a

[17:16](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1036s) crazy new request that completely

[17:18](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1038s) changes that and now you have to account

[17:20](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1040s) for like 10,000 users accessing your

[17:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1042s) platform while also accounting for the

[17:24](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1044s) new product users. It's just very hard

[17:26](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1046s) to do that when you're like a nent

[17:28](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1048s) product. But we managed to do it. So I'm

[17:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1049s) pretty proud of the team. Um the second

[17:32](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1052s) challenge I will touch on here is the

[17:34](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1054s) three different audiences. I know I made

[17:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1056s) it sound hopefully a little easy when I

[17:39](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1059s) was talking to you about engineers,

[17:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1060s) engineering managers and the finance

[17:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1062s) folks. But if you really think about it,

[17:44](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1064s) the true challenge of a product,

[17:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1065s) especially in the age of AI, is not just

[17:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1067s) building something and like shipping it,

[17:49](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1069s) right? The true challenge is how do you

[17:51](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1071s) make it so simple and dumb and easy to

[17:53](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1073s) use that everybody just gets what they

[17:56](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1076s) want from it. That's very hard to do

[17:58](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1078s) when you have crazy different ICPS.

[18:00](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1080s) And then the third of many challenges is

[18:04](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1084s) the thing I mentioned before

[18:05](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1085s) reconciliation. It's like this data is

[18:07](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1087s) not constant. It changes over time.

[18:09](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1089s) Especially the finance folks in the room

[18:11](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1091s) can empathize with me here. And how do

[18:13](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1093s) you like fetch that data, the new data

[18:15](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1095s) that was so different from the old data

[18:17](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1097s) that you promised you would not change.

[18:19](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1099s) But that is just how providers are. Um

[18:22](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1102s) not just that your people come and go,

[18:24](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1104s) right? I might quit tomorrow. You might

[18:25](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1105s) like have a new manager tomorrow. Like

[18:27](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1107s) things can change. How do you find a

[18:29](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1109s) nice way to loop that in to your

[18:31](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1111s) visualization of like attribution? It's

[18:34](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1114s) it's a pretty hard problem. With that

[18:36](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1116s) being said, I know we're towards the

[18:38](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1118s) end, so I want to leave at least I well

[18:40](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1120s) 20 seconds for questions. And I also

[18:42](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1122s) want to end with please keep thinking

[18:45](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1125s) about this definition of tokconomics

[18:47](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1127s) like what does it now mean to you? Thank

[18:49](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1129s) you.

[18:51](https://www.youtube.com/watch?v=KdG9OSOtJOA&t=1131s) [applause]
