# From Systems of Record to Systems of Context

Omri Bruchim · Tomer Ast

AI Engineer World's Fair 2026

[Talk page](https://ai.engineer/talks/Btk8wDUVs74-from-systems-record-systems-context) · [Video](https://www.youtube.com/watch?v=Btk8wDUVs74)

## From Systems of Record to Systems of Context

Sidekick’s architecture combines a durable model of how someone works with fresh signals about their day, preparing context before an agent is asked to prioritize.

### What should I focus on right now?

[0:53](https://www.youtube.com/watch?v=Btk8wDUVs74&t=53s)

Ask an assistant what to focus on right now, and it may return a confident list with little connection to your work. **Gemini**, **GPT** and **Claude** can all be put to this test. In Omri Bruchim’s anecdote, Claude suggested going to the gym—a memorable example of advice that missed the workplace question.

Connecting the assistant to boards, tasks, email and Slack does not automatically solve that problem. It can have access to the records without knowing how they fit together or what they imply for your priorities.

![Slide connects chat, inbox, calendar, team projects, org chart and notes beneath the statement that the AI assistant still cannot really answer.](https://ai.engineer/api/talk-essay-media/413f55adabeb7d25bf8aefd3fc4448ec8145fd52a88aedb96d4f3d62d8b6fbcf)

The assistant has access to workplace data but still cannot answer.

**Access to information is different from understanding its significance.** That distinction is the starting point for monday.com’s architecture: the missing step is interpreting the available evidence, not merely retrieving more of it. Here, understanding means being able to connect work to the person, relationships and priorities that give it meaning.

### An assistant where the work happens

[2:30](https://www.youtube.com/watch?v=Btk8wDUVs74&t=150s)

Bruchim and Tomer Ast introduce themselves as engineering managers at monday.com, where projects, tasks, decisions, meeting notes and action items accumulate as teams work. The product ambition extends beyond storing those records: a sales team might use an SDR agent to contact prospects, while finance or marketing teams might need help conducting research. The assistant has to contribute to the business outcome.

The platform divides that ambition among four offerings:

| Offering | Intended role |
| --- | --- |
| [monday sidekick](https://support.monday.com/hc/en-us/articles/26701503726610-Get-started-with-monday-sidekick) | Assist with the user’s work |
| monday vibe | Build software |
| monday agents | Create agents within the platform |
| monday workflows | Run more deterministic flows |

Sidekick is the focus here: a personal assistant intended to understand the user and their business, think and execute alongside them, and adapt to their working style while keeping them in control. Those promises make context an architectural requirement.

### Why meaning must be built before the request

[4:30](https://www.youtube.com/watch?v=Btk8wDUVs74&t=270s)

Putting one assistant above Slack, meeting notes and the rest of the workplace creates a wall of records. The first difficulty is the **agent gap**: an agent can execute a clearly specified task without knowing which task deserves attention. Ask it to draft a reply to a particular customer escalation and it has a concrete problem to solve. Ask what to focus on first and it must also understand who you are and what matters to you. Remembering something about you does not necessarily establish those priorities.

The second difficulty is that a record rarely contains its full meaning. Consider recovering the reason for a line of code:

1. Read the line and any nearby comment.
2. Use `git blame` to locate the commit and inspect its message.
3. Follow the commit to the pull request and read its description.
4. Follow the pull request’s connection to a monday.com item.
5. Discover that the work originated in a customer complaint.

The explanation emerges from the connections between records. The line tells you what the software does; the chain can explain why someone changed it.

![Slide titled “A log says what happened. Never what it means.” traces an example through git blame, a commit message, a PR description, a monday task and a customer complaint.](https://ai.engineer/api/talk-essay-media/50f108a68b57b7071e19b175e9951d211f52e96c53f575504efc50ba10f884a9)

A log records what happened; understanding why requires tracing the connections.

The third difficulty is timing. Reconstructing that meaning only when someone asks a question leaves too much work for request time. The proposed **Monday world model** builds understanding in advance: who the user is, why a fact matters, when to help and what not to do. It is a model of the work and its relationships, rather than simply a larger prompt or a longer context window.

In this diagnosis, provider connections and MCPs already make the data accessible. The remaining challenge is interpreting how entities connect. That scopes the claim about retrieval: obtaining records alone is insufficient; it does not mean the architecture eliminates retrieval.

### Three things the agent can reason over

[8:11](https://www.youtube.com/watch?v=Btk8wDUVs74&t=491s)

Ast describes collecting thousands of data points about a user across items, status changes, activity logs, messages and meetings. The architecture turns those observations into three kinds of context:

- **Work structure:** Key entities and their relationships—what depends on what, how a Slack message connects to a task, and who is blocking whom.
- **Current snapshot:** Live signals over those entities—overdue work, critical urgency, and the coworkers the user is actively working with, including why.
- **Durable profile:** What decisions, outcomes, work patterns and recurring cadences reveal about the user over time.

These outputs serve different reasoning needs. Relationships explain the structure of the work, signals describe its current condition, and the profile gives the agent a persistent account of the person doing it.

### One engine learns the person; another follows the day

[9:05](https://www.youtube.com/watch?v=Btk8wDUVs74&t=545s)

Two engines construct this context using different time windows and schedules. The **slow engine** examines weeks of activity to identify the user’s role, routines, work rhythm, collaborators, goals and current projects. It distills those patterns into a durable profile. When a pattern continues to hold, the engine reinforces it, gradually refining its account of how the user works.

The **fast engine** takes a short recent window and frequently recomputes live signals: what is overdue, what has suddenly become urgent, and which coworkers have become involved.

| Engine | Evidence window | Output |
| --- | --- | --- |
| Slow | Weeks of activity | Durable profile and recurring patterns |
| Fast | Short recent window | Current state and live signals |

The split separates relatively stable knowledge about the person from changes that may alter today’s priorities. A durable routine and a newly urgent obligation both matter, but they should not be treated as the same kind of evidence.

### Rapid updates and gradual consolidation

[10:16](https://www.youtube.com/watch?v=Btk8wDUVs74&t=616s)

Ast connects this temporal split to two precedents. The first is [complementary learning systems](https://web.stanford.edu/~jlmcc/papers/McCMcNaughtonOReilly95.pdf): rapid learning associated with the hippocampus complements slower consolidation in the neocortex. For Sidekick, the useful inspiration is the coexistence of recent experience and durable learning, not a claim that its engines reproduce the brain.

The second is [Lambda architecture](https://nathanmarz.com/blog/how-to-beat-the-cap-theorem.html). A batch layer recomputes views over historical data, while a speed layer handles recent data not yet incorporated into those results. Queries combine the two into a served view. Sidekick applies a related idea to its context model: combine slower processing with a faster account of recent changes. The profile-versus-signals division is an architectural analogy, not an assertion that the implementations are identical.

### Precompute context, then refresh selectively

[11:14](https://www.youtube.com/watch?v=Btk8wDUVs74&t=674s)

The serving path connects those engines to the assistant:

1. Collect workplace data from monday.com, Slack, email and calendar.
2. Turn the data into structures, signals and patterns.
3. Run both engines offline, before the user engages with Sidekick.
4. At engagement time, recompute a thin slice of logic for recent activity and serve the resulting context to the agent.
5. Let Sidekick decide when and how to traverse the data model and retrieve further context.

**Precomputation prepares the agent to reason; selective retrieval still helps it investigate.** The request does not have to initiate the entire process of interpreting the user’s work.

This design also separates freshness from availability. Ast describes isolating sources so that a bad feed does not break the others. At serving time, the thin logic layer verifies part of the context against live data; elsewhere, the system retains the last verified context. The intended behavior is graceful degradation, with some information remaining older rather than the whole context becoming unavailable. This is a resilience design claim, not a measured availability guarantee.

Urgency is another intended consequence of the model. If the assistant can interpret the significance of a fact, it has a basis for deciding whether to notify the user proactively or remain silent. The goal is not simply to detect more activity, but to recognize when that activity merits an interruption.

### A growing model still trails reality

[12:30](https://www.youtube.com/watch?v=Btk8wDUVs74&t=750s)

Continued observation is intended to make the context more useful: daily capture fills in the layers, sharpens the profile and produces a model specific to the individual’s work. Ast describes adding a source as deliberately inexpensive and additive, extending the evidence available to the engines.

The architecture nevertheless has three explicit limitations:

- **Lag:** The represented world always trails what is actually happening.
- **Cold start:** New users do not yet have enough reliable history to support the same reasoning.
- **Bias and noise:** Signal definitions encode their designers’ assumptions, and distinguishing important evidence from noise remains difficult.

The proposed advantage is an architecture that can be enriched, tested and improved over time—not a claim that these problems have been solved.

### Turning Omri’s breadcrumbs into today’s context

[13:28](https://www.youtube.com/watch?v=Btk8wDUVs74&t=808s)

Return to the opening question: what should Omri focus on right now? The worked example begins with what Bruchim calls breadcrumbs: his current board, the past month’s emails, meeting transcripts and the action items from those meetings. Offline calendar processing adds recurring patterns. A regular meeting with VPs, for example, helps establish the significance of another meeting tomorrow. Recent Slack messages can also contain relevant statements that never appeared in the team’s daily stand-up.

The slow engine turns that history into a profile: Omri Bruchim is an engineering manager in Tel Aviv, working on Sidekick and Notetaker, with particular working hours and daily routines. The fast engine supplies obligations from the past day or few days. In the example, those include three commitments to other people and an email from a VP that he has not answered.

![Slide pairs “What should I focus on right now?” with stacked Serve, Build and Collect panels, including suggested priorities, fast and slow processing lanes, and workplace data sources.](https://ai.engineer/api/talk-essay-media/d57ab10c0582bdfb0e8ca3a746a341a30b34e6bd722b217eb60e7de84fd77701)

Sidekick’s answer sits above the Build and Collect layers.

The two outputs contribute different parts of a prioritization answer. The profile establishes whose work is being prioritized; the recent signals identify pending obligations that matter today. The example illustrates the evidence available to Sidekick, rather than a measured prioritization result or completed action. The commitments remain commitments, and the unanswered email remains something to address.

### Connect the work before asking the agent to act

[15:01](https://www.youtube.com/watch?v=Btk8wDUVs74&t=901s)

For Sidekick, the proposed architectural shift is to process relationships and user context before the agent needs them. A capable model with access to records still needs a basis for understanding how those records connect to this person’s work. The Monday world model supplies that basis in advance, so a question about priorities can begin with an account of the user’s situation instead of a fresh pile of records.

## Resources

- [Get started with monday sidekick](https://support.monday.com/hc/en-us/articles/26701503726610-Get-started-with-monday-sidekick) — Instructions for opening Sidekick, connecting work apps, using contextual assistance and managing personalization and memories.
- [Navigating monday's AI work platform](https://support.monday.com/hc/en-us/articles/35276662798098-Navigating-monday-s-AI-work-platform) — A guide to finding Sidekick, Vibe, agents, workflows and Notetaker within monday.com.
- [Complementary learning systems: the original 1995 paper](https://web.stanford.edu/~jlmcc/papers/McCMcNaughtonOReilly95.pdf) — The foundational account of rapid hippocampal learning and gradual neocortical consolidation through interleaved experience.
- [Nathan Marz on combining batch and realtime computation](https://nathanmarz.com/blog/how-to-beat-the-cap-theorem.html) — Explains how precomputed historical views and a realtime layer covering recent data combine to answer queries.
- [Building monday.com Sidekick: why capable agents need more than just tools](https://www.langchain.com/blog/building-monday-com-sidekick-why-capable-agents-need-more-than-just-tools) — Omri Bruchim describes Sidekick's architecture for permission-aware retrieval, specialized subagents, bounded tools, sandboxed execution and evaluation.

## Slides

### From Systems of Records to Systems of Context

Tomer Ast and Omri Bruchim introduce a shift toward software that understands connections between recorded tasks, documents, messages and status updates.

[0:20](https://www.youtube.com/watch?v=Btk8wDUVs74&t=20s) [Title slide naming Tomer Ast and Omri Bruchim, with a dark astronaut illustration.](https://assets.ai.engineer/sites/aiecode2025/sha256/0cfb4fa13ff4c53f9d95dc7ea4ee630be2f2f526dacb2c7d9028cfd9a516a085.webp)

monday AI engineering

From Systems of Records to Systems of Context

Tomer Ast
Omri Bruchim

Interface text:
Notes
Edit
Full
2 / 30

### What Should I Focus on Right Now?

The speaker uses a daily prioritization question to illustrate how assistants can give confident advice disconnected from the work at hand. In his test, Claude suggested going to the gym.

[0:53](https://www.youtube.com/watch?v=Btk8wDUVs74&t=53s) [The question appears beside a colorful astronaut illustration.](https://assets.ai.engineer/sites/aiecode2025/sha256/3773b91746a8f3ba56e2fd97134c1094edb8c6abff0f7efc08624493fad41551.webp)

What Should I Focus on Right Now?

Interface text:
Notes
Edit
Full
2 / 30

### The Assistant Has the Data but Cannot Answer

The speaker argues that connecting work records still leaves an assistant unable to understand what the user is working on. The diagram links examples across messages, email, calendars, projects, notes, and an org chart.

[1:41](https://www.youtube.com/watch?v=Btk8wDUVs74&t=101s) [A claim about an AI assistant's limited understanding appears above a network of example work records.](https://assets.ai.engineer/sites/aiecode2025/sha256/dd5f4ac2ba4a517e707f577840e0cab1f09bbd22e7c4889335a78ce81d8a3468.webp)

Your AI assistant has every board, task, email, and message you've ever touched.

And it still can't really answer.

Dana
online
Hi, when will my order be shipped?
Your order has shipped! Arriving May 5 🚚
1:32 PM
Type a message

Inbox
1 new
SpaceX
9:41 AM
Re: launch services agreement
Confirming the payload integration window before the call...

Calendar
VP sync · 9:00–9:30

Team Projects
Buy domain — Done
New website — Working on it
#4821 Launch — Stuck

Org chart
Acme Corp · reports to VP

# launch
Today
Tomer Ast [illegible]
Hey, do you remember to send me the slide for the meeting tmw morning?

Notes — AI Engineer conf
- Open on the “still can’t answer” hook
- Records ≠ meaning
- Slow + fast engines → world model

monday AI engineering

Interface text:
Notes
Edit
Full
3 / 30

### The Challenge

The speaker identifies missing understanding as the central challenge: an AI assistant can have the data yet still struggle to answer.

[2:02](https://www.youtube.com/watch?v=Btk8wDUVs74&t=122s) [The slide says the problem is missing understanding, alongside an outlined astronaut.](https://assets.ai.engineer/sites/aiecode2025/sha256/a72e659d21c4f5165c963327adabb9b6d6166818ac7ab29d77732fac0daac814.webp)

The Challenge

Not missing data.
Missing understanding!

Interface text:
Notes
Edit
Full
‹ 4 / 30 ›

### Omri Bruchim and Tomer Ast

Omri Bruchim and Tomer Ast are engineering managers at monday.com working on the problem discussed in this talk.

[2:28](https://www.youtube.com/watch?v=Btk8wDUVs74&t=148s) [Portraits and names of the two speakers appear beside a waving hand.](https://assets.ai.engineer/sites/aiecode2025/sha256/7e163d0c7a803b835b9e76d568fb64f055823d7b8cffcc3fc98aa1dfe8b16738.webp)

Omri Bruchim
Tomer Ast

monday.com

Interface text:
Notes
Edit
Full
5 / 30

### The Best AI Work Platform

monday.com presents an AI work platform spanning teams and products. The speaker says projects, tasks, decisions, meetings, notes and action items are logged in the system.

[2:41](https://www.youtube.com/watch?v=Btk8wDUVs74&t=161s) [A monday.com overview describes its AI work platform alongside tiles for teams and products.](https://assets.ai.engineer/sites/aiecode2025/sha256/c3f97014b61432dd0780bcfcd1e6b3fbf49ff9a1f81fd6ed25ad0d14dba05e8f.webp)

The best AI work platform

monday.com

monday.com is a global software company that helps customers scale business impact and expand what teams can deliver.

Marketing
Projects
Sidekick
Vibe
Operations
Product
Design
Agents
Agent powers
IT
CRM
HR

Interface text:
Notes
Edit
Full
6 / 30

### Helping teams achieve business outcomes

monday.com maps its mission to outcomes across eight teams, from sales growth to the employee journey. The speaker proposes an SDR agent that calls prospects as one example.

[3:11](https://www.youtube.com/watch?v=Btk8wDUVs74&t=191s) [Business outcomes for sales, marketing, operations, finance, support, legal, product and HR teams.](https://assets.ai.engineer/sites/aiecode2025/sha256/3dbb5d1c9752355076693b5880f97c55a321ca0b7f773c03ebe2be6998d5a2f5.webp)

Our mission has always been to help teams achieve their business outcomes

Sales
Accelerate revenue and growth

Marketing
Accelerate demand and pipeline growth

Operations
Maximize efficiency and workflow agility

Finance
Streamline financial operations efficiently

Support
Provide service at speed and scale

Legal
Minimize risk and accelerate cycles

Product & R&D
Speed up innovation and build anything

HR
Optimize the employee journey

Interface text:
Notes
Edit
Full
7 / 30

### monday sidekick

Sidekick is presented as a personal assistant that knows your work and keeps you in control. The speaker says delivering these promises is difficult.

[3:55](https://www.youtube.com/watch?v=Btk8wDUVs74&t=235s) [Sidekick feature claims beside an example assistant interface.](https://assets.ai.engineer/sites/aiecode2025/sha256/7a176e26603569b17572532923fe6a594949d7f1105441e1b619e29522f8e459.webp)

monday sidekick

Your intelligent AI personal assistant that understands your work, thinks and executes with you

Knows you and your business
Accelerates work at every level
Works the way you do
Keeps you in total control
Delivers answers you can trust

Hey Sandra, ready to get started?
Describe what you need or ask a question...
Mention assets

Write a doc
Create a board
Research online
Analyze data
Generate an image
Build a Vibe app

Explore ready-made prompts for you

Review my recent emails and surface what needs my attention
Gmail

Create a searchable map of your company's structures...
Vibe

Build an RSVP help event organizers confirm attendance for their even...
Agents

Lunch a CSAT customer feedback form and compare it to industry [illegible]
Create social media posts for an upcoming campaign
Review all my tasks this week and recommend the next priority

Interface text:
Notes
Edit
Full
9 / 30

### One assistant for all of your work

One assistant sits amid records from boards, documents, Slack, email, calendars, transcripts, and notetaker actions. The speaker describes the challenge of connecting this overwhelming wall of records.

[4:27](https://www.youtube.com/watch?v=Btk8wDUVs74&t=267s) [An assistant icon surrounded by examples of work records from multiple sources.](https://assets.ai.engineer/sites/aiecode2025/sha256/017c5455c065157469a915c6ed5b2ffb7b3a5b4db7a6f3e626024a7509364db5.webp)

THE CHALLENGE WE'RE FACING
One assistant —
all of your work.

Q3 Launch
Sales · Acme
Escalation
Design notes
Ship hotfix
Board mtg
#general
Hiring

PRD v2
Unblock #4821
1:1 · [illegible]
[illegible]
Renewal

#launch
Sprint 24

VP sync
Auth spec
[illegible]
Kickoff call

Standup 9:00
#eng-oncall
Bug triage
Retro notes
Re: contract
Launch brief
Send notes

Boards
Documents
Slack
Email
Calendar
Transcripts
Notetaker actions

monday AI engineering

Interface text:
Notes
Edit
Full
10 / 30

### The agent gap

The speaker calls this the agent gap: an agent can draft an escalation reply when given a task, but guesses what to tackle first because it lacks the user's priorities.

[5:00](https://www.youtube.com/watch?v=Btk8wDUVs74&t=300s) [Agents can execute specified tasks but struggle to identify what needs attention.](https://assets.ai.engineer/sites/aiecode2025/sha256/5b342dfb5f4961027b7e09d5bb7fa83587c2eb3eddde819e0919e5751ae91415.webp)

THE AGENT GAP

01

Sharp at doing tasks.
Lost at finding them.

Point an agent at a problem and it's brilliant. But “what should I focus on?” isn't a problem to solve — it's asking the agent to find what the problems even are.

“Draft a reply to this escalation” → nails it.
“What should I work on first?” → guesses.

monday AI engineering

Interface text:
Notes
Edit
Full
‹ 11 / 30 ›

### Records and meaning

A log shows a deadline change; its reason emerges by tracing from git blame through the commit, PR and monday task to the customer complaint.

[5:40](https://www.youtube.com/watch?v=Btk8wDUVs74&t=340s) [A log records a change; understanding its reason requires tracing related work and context.](https://assets.ai.engineer/sites/aiecode2025/sha256/f92cb485b30faa70e315170577a0e3cf2dff0e5abee24873650998017f989912.webp)

RECORDS ≠ MEANING

02

A log says what happened.
Never what it means.

A record is like a line of code someone wrote months ago. It tells you what changed — never why. To recover the meaning you trace it back, one hop at a time.

The log: launch.due += 5d — what changed.
The why lives a hop away, every time:
git blame → the commit message → the PR description
→ the monday task that asked for it → the customer who complained in the first place.

monday AI engineering

Interface text:
Notes
Edit
Full
12 / 30

### Built ahead of time

Turning work records into meaning takes layers of processing. The speaker argues that this context must be built before a question arrives, because an LLM cannot reconstruct the whole work history at runtime.

[6:40](https://www.youtube.com/watch?v=Btk8wDUVs74&t=400s) [Slide explains that understanding work history must be built before an agent needs it.](https://assets.ai.engineer/sites/aiecode2025/sha256/838a514a6a9499e0af4031ffaca04c783fe3b58a5b17e30c543a704c1a06e528.webp)

03

BUILT AHEAD OF TIME

At runtime,
it's too late.

Turning records into meaning takes layers of processing — signals, patterns, connecting the dots. An LLM can't reconstruct your whole work history the instant you ask.

Understanding has to be built ahead of time — waiting, so the moment the agent needs it, it's already there.

monday AI engineering

Interface text:
Notes
Edit
Full
13 / 30

### The monday.com world model

The monday.com world model builds context ahead of time about who you are, your work and what needs action, giving an agent information to reason over when deciding how and when to help.

[7:09](https://www.youtube.com/watch?v=Btk8wDUVs74&t=429s) [Slide introduces a world model that follows work and gives an agent context for reasoning.](https://assets.ai.engineer/sites/aiecode2025/sha256/c79380e873dcdc099593d5507d0486452a4f3c7b4ee38875759d85a89b15cd5b.webp)

WHAT WE BUILT

Not a bigger prompt.
The monday.com world model.

why this matters
how to help
when — & when not — to act

Context that follows your work: who you are, what you're working on, what needs action — a model the agent can reason over to answer what it can't derive on its own.

monday AI engineering

Interface text:
Notes
Edit
Full
‹ 14 / 30 ›

### It isn't retrieval

Giving an agent every board, doc and message can still leave it lost. The model needs to understand how work and its entities connect.

[7:44](https://www.youtube.com/watch?v=Btk8wDUVs74&t=464s) [Slide describes the model as a working understanding of work rather than a search system.](https://assets.ai.engineer/sites/aiecode2025/sha256/374f71ac2a70b5d0be0025f2fcbdddf2b7eee6a328f79d0014c61317f44c46a6.webp)

FIRST, WHAT IT'S NOT

It isn't retrieval.

The problem was never getting the data — hand an agent every board, every doc, every message, and it's still lost. So the model isn't a smarter search over your work. It's a working understanding of it.

monday AI engineering

Interface text:
Notes
Edit
Full
15 / 30

### Three faculties of understanding

The model connects work entities and relationships with live signals and patterns learned from decisions and outcomes. Each fact is stamped in time, preserving what was true and when it changed.

[8:13](https://www.youtube.com/watch?v=Btk8wDUVs74&t=493s) [Three cards describe structure and connections, the present, and patterns learned over time.](https://assets.ai.engineer/sites/aiecode2025/sha256/9f2ac0c517f68812c8b8d2a7b1de36a658b5e0eb788b55e95213d391c34e6132.webp)

WHAT THE MODEL UNDERSTANDS

Three faculties of understanding.

HOW IT'S STRUCTURED
Structure & connections
Items, people, meetings, boards — and the relationships that carry the meaning: what blocks what, who's waiting on whom, where a thing got decided.

WHAT'S TRUE RIGHT NOW
The present
Live signals read off your current state: what's overdue, what's piling into a crunch, who you're suddenly in the loop with.

WHAT IT'S LEARNED
Learned over time
Episodes, decisions and their outcomes — distilled into durable patterns: how you actually work, not how a job description says you do.

every fact stamped in time — what's true, what was true, and when it changed

monday AI engineering

Interface text:
Notes
Edit
Full
16 / 30

### Two engines on different clocks

The slow engine learns the user and their work over a long window. The fast engine reads what is happening now and how it affects that work.

[9:03](https://www.youtube.com/watch?v=Btk8wDUVs74&t=543s) [Two illustrated characters represent a slow engine that learns who you are and a fast engine that reads what is happening now.](https://assets.ai.engineer/sites/aiecode2025/sha256/6f2210f527c9c4ecc75b19549e7a4f30d28471237e8b74e520e51de1fc69efcc.webp)

Two engines, on different clocks.

HOW THE MODEL IS BUILT

SLOW ENGINE
Learns who you are

FAST ENGINE
Reads what's happening now

monday AI engineering

Interface text:
Notes
Edit
Full
17 / 30

### Two engines, on different clocks

The slow engine mines weeks of activity for a profile that strengthens as patterns hold. The fast engine refreshes live indicators over a recent window to track the user's current state.

[9:23](https://www.youtube.com/watch?v=Btk8wDUVs74&t=563s) [The slow engine mines patterns across weeks, distills a durable profile, and reinforces patterns that hold.](https://assets.ai.engineer/sites/aiecode2025/sha256/b30640efdec43adeec9b6664fd0cc49193059260fb7647d3749f9ce9254b0bdd.webp)

HOW THE MODEL IS BUILT

Two engines, on different clocks.

SLOW ENGINE
Learns who you are

Mines patterns across weeks
Distills a durable profile
Strengthens as patterns hold

It knows you.

monday AI engineering

Interface text:
Notes
Edit
Full
18 / 30

[9:54](https://www.youtube.com/watch?v=Btk8wDUVs74&t=594s) [Two illustrated characters accompany a description of the fast engine, which refreshes indicators from a recent window.](https://assets.ai.engineer/sites/aiecode2025/sha256/846b2b761f1c8441f04dfd52baf74aad549e7160be90d3f252694a9ce17107d5.webp)

HOW THE MODEL IS BUILT

Two engines, on different clocks.

FAST ENGINE
Reads what's happening now

Recomputes live indicators
Over a recent window
Your current state, refreshed

It knows your day.

monday AI engineering

Interface text:
Notes
Edit
Full
19 / 30

### This split isn't ours

The monday.com world model applies a split seen in complementary learning systems and Lambda architecture to the agent’s data model.

[10:15](https://www.youtube.com/watch?v=Btk8wDUVs74&t=615s) [The slide links complementary learning systems and lambda architecture to the monday.com world model.](https://assets.ai.engineer/sites/aiecode2025/sha256/52b76c1ed259211f80a13eba6356f6878bc48942d7b3ff63cea4e5bdc6ab4281.webp)

THE LINEAGE

This split isn't ours.

COGNITIVE NEUROSCIENCE
Complementary Learning Systems

DATA INFRASTRUCTURE
Lambda Architecture

Two different fields, the same shape. We applied it to the one place no one had:
the monday.com world model.

monday AI engineering

Interface text:
Notes
Edit
Full
20 / 30

### The brain already runs slow + fast

A child’s burn illustrates complementary learning systems: the hippocampus captures the episode quickly, while the neocortex distills a lasting lesson—don’t touch.

[10:35](https://www.youtube.com/watch?v=Btk8wDUVs74&t=635s) [A child and fire illustrate a fast episode captured by the hippocampus and a lasting lesson distilled by the neocortex.](https://assets.ai.engineer/sites/aiecode2025/sha256/6574f94ba7e441980577c63f64c930fe562d052eb8e9f1a80f12a5f5de81513b.webp)

The brain already runs slow + fast.

THIS SPLIT ISN'T OURS · THE BRAIN

COGNITIVE NEUROSCIENCE
Complementary Learning Systems

FAST The hippocampus grabs the episode — the burn, in an instant.
SLOW The neocortex distills the durable lesson — don't touch.

THE EPISODE
THE LASTING LESSON

monday AI engineering

Interface text:
Notes
Edit
Full
‹ 21 / 30 ›

### So does the data stack

Lambda architecture merges a fast layer for a recent realtime window with a slow batch layer that recomputes the full history, producing one served view.

[10:49](https://www.youtube.com/watch?v=Btk8wDUVs74&t=649s) [A diagram combines a fast realtime stream and a slow batch stream into one served view.](https://assets.ai.engineer/sites/aiecode2025/sha256/3ba525d127ee570812af901f464deeb108539d817b8cf77ae118c1a905c0b8c0.webp)

THIS SPLIT ISN'T OURS · THE DATA STACK
So does the data stack.

REALTIME STREAM · FAST
BATCH STREAM · SLOW
SERVED VIEW
one current answer

DATA INFRASTRUCTURE
Lambda Architecture
FAST  Speed layer — a recent realtime window, refreshed.
SLOW  Batch layer — the full history, recomputed.

monday AI engineering

Interface text:
Notes
Edit
Full
22 / 30

### Heavy below, live on top

Two engines precompute context from work data offline. When a user engages, a thin live slice updates recent activity, and Sidekick can retrieve context from the model.

[11:12](https://www.youtube.com/watch?v=Btk8wDUVs74&t=672s) [A layered diagram shows data collection, offline slow and fast engines, online serving, and monday Sidekick.](https://assets.ai.engineer/sites/aiecode2025/sha256/53b0802971142a41eee6ee020e00c0a7e296e1f025c983eed009fe2a68e802de.webp)

HOW THE MODEL IS BUILT — AND SERVED
Heavy below, live on top.

monday Sidekick
↑ PULL · FOLLOW CONNECTIONS · COME BACK ↻

Serve ⚡ ONLINE · WHEN YOU ENGAGE
A thin live slice — the agent reads a model that's already built.

Build ⌛ OFFLINE · AHEAD OF TIME
SLOW ENGINE
Learns who you are
FAST ENGINE
Reads what's happening now

Collect FROM EVERYWHERE YOU WORK
structure · signals · patterns

monday AI engineering

Interface text:
Notes
Edit
Full
23 / 30

### Behavior we never had to bolt on

Isolated sources and fallback to the last verified context help the model withstand a bad feed. The model also weighs urgency to decide when to notify the user or stay quiet.

[11:52](https://www.youtube.com/watch?v=Btk8wDUVs74&t=712s) [Two panels describe graceful degradation through isolated sources and fallbacks, and restraint in deciding when to interrupt.](https://assets.ai.engineer/sites/aiecode2025/sha256/4b0387baefae5d4781760c353a78c01afb9d811bcf01ca95cce8fed1b0ccc813.webp)

WHAT FALLS OUT OF BUILDING IT THIS WAY

Behavior we never had to bolt on.

RESILIENCE
It degrades — it doesn't fail.

"But precomputed means stale." So it's built to bend, not break.

isolated sources — one bad feed can't sink the rest
a freshness budget — serve the last good answer, not nothing
model-steps fall back down a chain

RESTRAINT
It knows when to stay quiet.

The agents that don't understand can't tell — so they spam you. Restraint isn't a feature we add; it falls out of understanding.

In the model it's just another signal: worth interrupting for, or leave them alone.

monday AI engineering

Interface text:
Notes
Edit
Full
‹ 24 / 30 ›

### And it compounds

The personal world model connects monday, meetings, mail, calendar and Slack. As daily work fills its layers, the profile sharpens; adding sources cheaply lets it keep growing.

[12:29](https://www.youtube.com/watch?v=Btk8wDUVs74&t=749s) [A personal world model connects to work sources beside text explaining how it grows richer over time.](https://assets.ai.engineer/sites/aiecode2025/sha256/774eb383b565de55ecba529f82aba78be0804cd7b9725e82ea62048242b23467.webp)

THE PAYLOAD
And it compounds.

YOURS ALONE
monday.com
world model

monday
meetings
mail
calendar
Slack

Every day you work, the layers get richer and the profile gets sharper. Adding a source is deliberately cheap — so the surface only grows.

The more it sees, the more it understands. The more it understands, the more you lean on it.

monday AI engineering

Interface text:
Notes
Edit
Full
‹ 25 / 30 ›

### The hard parts

The world model trails live work, has little to learn from for new users, and must separate useful signals from noise. The speaker notes that signals also carry biases.

[12:55](https://www.youtube.com/watch?v=Btk8wDUVs74&t=775s) [A slide lists eventual consistency, cold starts, and separating signal from noise alongside an illustrated character.](https://assets.ai.engineer/sites/aiecode2025/sha256/ed2609c5982a07205457886f5cdaeb88bd52ff3c9378042a398e048b70f7f0ef.webp)

WE'RE NOT PRETENDING THIS IS SOLVED

The hard parts.

• Eventual consistency — the model is always a little behind the live world. A feature, but you design for it.
• Cold start — new users, before there's anything to learn from.
• Signal from noise — the hardest one, and the same one underneath the whole talk. That's the work.

monday AI engineering

Interface text:
Notes
Edit
Full
26 / 30

### The bottleneck was understanding

A network connects people and work tools. The speaker argues that an agent needs to connect those dots to understand the work instead of guessing.

[15:01](https://www.youtube.com/watch?v=Btk8wDUVs74&t=901s) [A network of connected work tools illustrates the claim that understanding is the bottleneck.](https://assets.ai.engineer/sites/aiecode2025/sha256/70d99e12bf7bbd445c820f526b8a71e18f5cf986e85ad9a5f4ec71d6ce824251.webp)

THE LESSON UNDERNEATH ALL OF IT

The bottleneck was never raw capability.
It was understanding.

The most capable agent in the world, pointed at your work without it, is still just guessing.

Calendar icon: 31

monday AI engineering

Interface text:
Notes
Edit
Full
28 / 30

### The monday.com world model

The speakers close with their vision for a monday.com world model that helps Sidekick understand a user's work, beyond answering questions.

[15:22](https://www.youtube.com/watch?v=Btk8wDUVs74&t=922s) [A closing slide says Sidekick aims to understand the user's work, beside illustrated characters.](https://assets.ai.engineer/sites/aiecode2025/sha256/d9b46367e0187b713714b0d733092d49e6feda6555edc989234b934596ef08f9.webp)

The monday.com world model.

That's what we're building — the understanding. Sidekick won't just answer your questions — it'll understand your work.

It doesn't just answer you... it knows you.

TOMER & OMRI · MONDAY.COM   THANK YOU.

monday AI engineering

Interface text:
Notes
Edit
Full
29 / 30

### AI Engineer World's Fair

[15:45](https://www.youtube.com/watch?v=Btk8wDUVs74&t=945s) [White AI Engineer World's Fair logo on a black background.](https://assets.ai.engineer/sites/aiecode2025/sha256/d9bf31daa0f2a065564382cc7ebcfbf3229c61ee1336e4f8906a13239e993117.webp)

AI Engineer
World’s Fair

HTTPS://AI.ENGINEER

## Full transcript

[0:00](https://www.youtube.com/watch?v=Btk8wDUVs74&t=0s) [upbeat music] [audience applauding] Um, hey everyone.

[0:16](https://www.youtube.com/watch?v=Btk8wDUVs74&t=16s) Uh, we are super excited to be here. Thanks for having us. Uh, today we're gonna talk about how we shift monday.com from, uh, a system of record, uh, into a system of context.

[0:27](https://www.youtube.com/watch?v=Btk8wDUVs74&t=27s) And honestly, uh, the title tell the whole story, uh, in just a single line. Uh, for decades, we build, uh, software that record what happened, um, every task, every document, every message, um, every status update, um, just, uh, put into the record.

[0:45](https://www.youtube.com/watch?v=Btk8wDUVs74&t=45s) What we want to talk today is like take it a st-step further. Uh, we want, uh, software that actually understand the connection between them.

[0:53](https://www.youtube.com/watch?v=Btk8wDUVs74&t=53s) So I wanna start from a simple question that each one of us ask himself every morning, um, what should I focus on right now? Um, it sounds almost trivial, but, uh, to be honest with yourself, if you ask your agent whether Gemini, GPT, or even Claude, um, if you ever typed this question, you probably got list of

[1:16](https://www.youtube.com/watch?v=Btk8wDUVs74&t=76s) bullets, um, not related to each other. Um, list of like item dressed up like a, a confident paragraph. Um, but, but it's not really, um, um, connected to what you're working on.

[1:30](https://www.youtube.com/watch?v=Btk8wDUVs74&t=90s) Actually, I tested la-last week and, uh, Claude asked me to go to the gym. I don't know if it's a compliment or not, but, uh, this is what he suggested.

[1:38](https://www.youtube.com/watch?v=Btk8wDUVs74&t=98s) Um, and, and what really make it so frustrating is like your AI assistant, um, have all this data. It has all the boards, the tasks, the emails, the Slack messages, everything that you ever touched if you connect it.

[1:54](https://www.youtube.com/watch?v=Btk8wDUVs74&t=114s) Um, but it still can't really answer it. Um, it has all the data, um, but it has zero understanding.

[2:04](https://www.youtube.com/watch?v=Btk8wDUVs74&t=124s) Um, so this is the real challenge we are facing. Um, e-- this is the heart of the entire talk. Uh, the problem was never the missing of data, the retrieval.

[2:12](https://www.youtube.com/watch?v=Btk8wDUVs74&t=132s) Uh, the problem is like the missing understanding. Uh, those are two totally different things, and almost everyone, uh, mix between them. Understanding is the word that we're gonna focus, uh, the entire wo-- uh, the entire talk.

[2:25](https://www.youtube.com/watch?v=Btk8wDUVs74&t=145s) Um, not context, not memory, not retrieval, understanding.

[2:30](https://www.youtube.com/watch?v=Btk8wDUVs74&t=150s) So, um, quick about ourselves. My name is Tomer. My name is Omri. This is Tomer. Um, uh, we're both engineering manager at monday.com, working on the exactly, uh, the problem that we're gonna talk about.

[2:43](https://www.youtube.com/watch?v=Btk8wDUVs74&t=163s) A little, a little bit context about where we're coming from. Uh, monday.com is a global, uh, software company. Uh, the-- we build a AI work platform, uh, used by hundreds of thousands of teams.

[2:54](https://www.youtube.com/watch?v=Btk8wDUVs74&t=174s) Um, and the part that really matter for the talk is that Monday is where the work, um, lives. Uh, we help companies to doing the work, not just like saving records.

[3:04](https://www.youtube.com/watch?v=Btk8wDUVs74&t=184s) Every project, every task, every decision, every meeting, the notes, the action item, everything logs, logged into the system. And our mission has always been to help the teams to achieve their business outcome, whether you are a salesperson, so we want to help you, like, create an SDR agent that call to your prospect and help you sell the

[3:25](https://www.youtube.com/watch?v=Btk8wDUVs74&t=205s) product. And if you are like, uh, a, a finance team or marketing, we want to help you, uh, with the, uh, research. So each one of the discipline, we want you help you to doing the job.

[3:37](https://www.youtube.com/watch?v=Btk8wDUVs74&t=217s) And together we have like, uh, four b-big bets on the AI Work Platform. We have like monday sidekick, we're gonna talk about it mostly today. Uh, monday vibe, if you want to build your own software.

[3:47](https://www.youtube.com/watch?v=Btk8wDUVs74&t=227s) monday agent, if you want to create your own agent into the platform, and like if you want some more, some more deterministic flow, we have like monday workflows. But today we're gonna focus on Sidekick.

[3:57](https://www.youtube.com/watch?v=Btk8wDUVs74&t=237s) Um, Sidekick is, uh, your intelligence AI personal assistant that's understand your work, think, and execute with you, uh, like a bird on your shoulder. Uh, he know you, he know your business.

[4:09](https://www.youtube.com/watch?v=Btk8wDUVs74&t=249s) Um, it help you with your work on every aspect. Uh, he work, work the way you do, uh, with your tones. Um, um, it's keep you totally in control.

[4:19](https://www.youtube.com/watch?v=Btk8wDUVs74&t=259s) But like putting all these bullets and, uh, building all these promises is really hard. Um, so let's talk about why it's hard. Uh, there are three point for that.

[4:30](https://www.youtube.com/watch?v=Btk8wDUVs74&t=270s) Um, first, like picture one assistant sitting on top of absolutely everything. Uh, you have your Sloc-- uh, your, uh, uh, Slack messages, you have all the notes for your meetings, um, absolutely everything.

[4:45](https://www.youtube.com/watch?v=Btk8wDUVs74&t=285s) And it's beautiful and overwhelming at the same time because as it stand, like they have wall of records everywhere. Um, so what-- why it's making so hard to connect between them?

[4:59](https://www.youtube.com/watch?v=Btk8wDUVs74&t=299s) There are three reason for that. One, it's something that we call the agent gap. Your agent, every agent is really sharp in doing task, um, when you-- when he know what to do.

[5:11](https://www.youtube.com/watch?v=Btk8wDUVs74&t=311s) Um, but he sometimes lost, find them-- find the problem. Um, if I ask my agent like, "Please help me draft and, and, and reply to like the escalation from a customer," he nail it.

[5:22](https://www.youtube.com/watch?v=Btk8wDUVs74&t=322s) He know how to do it. He take the context, like doing the work. But asking him, "What should I focus on first?" He guess, because he doesn't understand what is my priority.

[5:34](https://www.youtube.com/watch?v=Btk8wDUVs74&t=334s) Who am I? It doesn't matter if he have a memory of something, he still don't know what is the problem.

[5:41](https://www.youtube.com/watch?v=Btk8wDUVs74&t=341s) The second problem is like, um, y- we have the records, but we don't have the meaning. Um, a log never said what it mean. Let's take an analogy from our world.

[5:55](https://www.youtube.com/watch?v=Btk8wDUVs74&t=355s) Let's say that you have one line of code in your GitHub. You look at a code, maybe there is a comment on top of it, nice, but you don't really understand why someone wrote this line of code.

[6:06](https://www.youtube.com/watch?v=Btk8wDUVs74&t=366s) Today, never-- n-none of us writing code, but Somewhere in the past, someone wrote this code, line of code, and if you really want to know, you can go to git blame and understand, uh, from the commit log what, why, why someone do it.

[6:20](https://www.youtube.com/watch?v=Btk8wDUVs74&t=380s) But if you want to go further, you can go to the PR and maybe read a description. And if you really want to go hard, you can go to your Monday board and see which PR connected to this item and understand that this line of code came because some customer complained about something.

[6:38](https://www.youtube.com/watch?v=Btk8wDUVs74&t=398s) So this is what we are trying to build. And the third point that is, uh, a bit challenging as well is like, um, it's really hard to build ahead of time, um, it, at runtime, um, the meaning of things.

[6:51](https://www.youtube.com/watch?v=Btk8wDUVs74&t=411s) Um, it's really too late. Um, you simply can't do, uh, something like that the moment someone asks the question. So understanding, understanding of the context has to be ahead of time.

[7:04](https://www.youtube.com/watch?v=Btk8wDUVs74&t=424s) Um, you need to build it much before someone asks a question. Um, and this is why we are build what we are building. We are building, uh, the Monday world model.

[7:15](https://www.youtube.com/watch?v=Btk8wDUVs74&t=435s) This is what we call. The Monday world model help you, um, understand why this matter, um, how to help you, who you are, uh, when and, and what's not to do.

[7:28](https://www.youtube.com/watch?v=Btk8wDUVs74&t=448s) Um, it's a context that follow your work. Um, he understand who you are, and it's simply not a bigger prompt. Uh, it's not a longer, uh, context window. It's a totally different, um, um, from, from what we know until now.

[7:45](https://www.youtube.com/watch?v=Btk8wDUVs74&t=465s) First, before, like, Tomer gonna talk about how we build it, what it's not. It's not a retrieval problem. The problem never been getting the data. We don't-- We have all the data.

[7:56](https://www.youtube.com/watch?v=Btk8wDUVs74&t=476s) We have all the connection to what all the provider that we want to get, all the MCPs. The problem is really to understand how it works, hun- understand how each one of these entity connected to each other.

[8:08](https://www.youtube.com/watch?v=Btk8wDUVs74&t=488s) So go ahead, Tomer.

[8:11](https://www.youtube.com/watch?v=Btk8wDUVs74&t=491s) Thank you, Omri. Hi, everybody. Um, so what's the data model? We collect thousands of data points on the user, every item, status change, their activity log, messages and meetings, and construct three things the agent can reason over.

[8:28](https://www.youtube.com/watch?v=Btk8wDUVs74&t=508s) The first is how the user's work is structured, their key it- entities and relationships and connections between them, what depends on what, how a message in Slack connects to a task, who's blocking whom.

[8:42](https://www.youtube.com/watch?v=Btk8wDUVs74&t=522s) The second is a current snapshot, live signals over those entities, what's overdue, what's critically urgent right now, which coworkers you've been actively working with and why. The third is what we can learn about the user over time, their decisions and outcomes, work patterns and cadences distilled into a durable profile.

[9:05](https://www.youtube.com/watch?v=Btk8wDUVs74&t=545s) So how do we build that data model? Um, we use two engines running on different time windows and schedules, a slow engine that runs on a long time window and learns the user and their work, and a fast engine that reads what's happening right now and how it affects the user's work.

[9:20](https://www.youtube.com/watch?v=Btk8wDUVs74&t=560s) One knows you, and the other one knows your day. First, the slow engine. It takes as context user's activity over weeks and mines it for patterns and the type of persona the user is, their routines, their work rhythm, who they collaborate with, their main goals and current projects.

[9:42](https://www.youtube.com/watch?v=Btk8wDUVs74&t=582s) Those patterns get distilled into a durable profile, and every time a pattern holds, it's reinforced. This engine tries over time to learn who exactly the user is and how they work.

[9:56](https://www.youtube.com/watch?v=Btk8wDUVs74&t=596s) The fast engine is the opposite. It takes as context a short recent window and recomputes a set of live signals over the user's current state. What's overdue, what's suddenly urgent, which coworkers you've been pulled in with.

[10:11](https://www.youtube.com/watch?v=Btk8wDUVs74&t=611s) This engine tries to understand your day and updates frequently.

[10:16](https://www.youtube.com/watch?v=Btk8wDUVs74&t=616s) Uh, this split isn't something we invented. It's present in two totally different fields. In neuroscience, this split is referred to as complementary learning systems, and in data processing architecture, it's referred to as a Lambda architecture.

[10:31](https://www.youtube.com/watch?v=Btk8wDUVs74&t=631s) We apply the same concepts to how we construct the agent's data model.

[10:37](https://www.youtube.com/watch?v=Btk8wDUVs74&t=637s) Our brain uses the same split. Uh, every important experience gets captured instantly by the hippocampus, and o- over time, the neocortex distills those into durable lessons.

[10:50](https://www.youtube.com/watch?v=Btk8wDUVs74&t=650s) In data infrastructure, the same-- there's the same split, a fast speed layer over a recent real-time window and a slow batch layer over the full history that gets recomputed, and the two are merged into a single served view.

[11:07](https://www.youtube.com/watch?v=Btk8wDUVs74&t=667s) Two different fields landed on the same idea, and that's what we're trying to apply to our data model.

[11:14](https://www.youtube.com/watch?v=Btk8wDUVs74&t=674s) So how does it all come together? We collect data from everywhere our users work, uh, Monday, Slack, emails, calendar, and we turn it into data structures, signals and patterns.

[11:27](https://www.youtube.com/watch?v=Btk8wDUVs74&t=687s) Both engines precompute on top of that offline and ahead of time, and when a user engages with Sidekick, a thin slice of logic is recomputed for recent activity, and the entire context is served to the agent.

[11:43](https://www.youtube.com/watch?v=Btk8wDUVs74&t=703s) Sidekick can then decide when and how to traverse and retrieve context from the data model itself, and it's primed to reason on top of it.

[11:53](https://www.youtube.com/watch?v=Btk8wDUVs74&t=713s) And building it this way gives us two behaviors out of the box. The data model, it's resilient. Sources are isolated, so a bad feed can't break the rest. And the thin layer of logic that runs at serve time verifies part of the context against live data while the rest falls back to the last verified context.

[12:13](https://www.youtube.com/watch?v=Btk8wDUVs74&t=733s) So it degrades gracefully, but it doesn't fail.

[12:18](https://www.youtube.com/watch?v=Btk8wDUVs74&t=738s) Second, it actually understands the urgency of facts. It's able to understand when and how it should be proactive and notify the user, and when it should stay silent.

[12:30](https://www.youtube.com/watch?v=Btk8wDUVs74&t=750s) And the crucial part is that it compounds. Every day, the data is captured, the layers fill in, and the profile sharpens, and adding a new data source is deliberately cheap and only contributes, so the surface only grows.

[12:45](https://www.youtube.com/watch?v=Btk8wDUVs74&t=765s) The more it sees, the more it understands, and the more it understands, the more you can lean on it. And the data model is unique. It's unique to how you work.

[12:56](https://www.youtube.com/watch?v=Btk8wDUVs74&t=776s) We're not pretending this solves everything. Uh, the model itself is always trailing the actual li- life world. New users have no reliable data to reason from yet, and signals have our own biases built in, so the hardest part is actually telling the important parts from the noise.

[13:16](https://www.youtube.com/watch?v=Btk8wDUVs74&t=796s) But this is an architectural design that we can enrich and test and improve over time. Thank you. [audience applauding]

[13:28](https://www.youtube.com/watch?v=Btk8wDUVs74&t=808s) So back, uh, back to the original, uh, question that we had, what should I focus on right now? Remember the, the question. So now Sidekick can really answer that.

[13:38](https://www.youtube.com/watch?v=Btk8wDUVs74&t=818s) Uh, let's see how. So first, like, uh, Tomer said, we collect all the data point. We called it, uh, breadcrumbs. Uh, for example, the, the board that I working on right now, um, all the emails from the past, uh, month, um, the transcripts from the meeting, all the action item that I got from this meeting, and we

[13:54](https://www.youtube.com/watch?v=Btk8wDUVs74&t=834s) processing them offline. Um, we see a full picture of your calendar. This is what help us to understand the pattern. So if you have, like, bi-weekly with your VPs, I understand that tomorrow you have another meetings.

[14:07](https://www.youtube.com/watch?v=Btk8wDUVs74&t=847s) And even, uh, uh, Slack messages from, uh, the last few days to understand if you said something in the Slack that's relevant that you didn't say in your daily, uh, uh, stand-up with the team.

[14:17](https://www.youtube.com/watch?v=Btk8wDUVs74&t=857s) We take all, all this data, and we, uh, uh, process them either fast or slow. Um, the slow builds some kind of a profile on you, so you can see it on the top, like Omri Bruchim is an engineering manager.

[14:30](https://www.youtube.com/watch?v=Btk8wDUVs74&t=870s) Uh, uh, we-- I'm working on two projects, Sidekick and Notetaker. I'm from Tel Aviv, and this is how many hours I have, uh, per day, and, like, how I work every day.

[14:40](https://www.youtube.com/watch?v=Btk8wDUVs74&t=880s) And the fast is something that's helped me to understand what is the action item. I have, like, three commitment that I promised to other people. Uh, I have to, to reply to a VP that send me email, and I didn't.

[14:50](https://www.youtube.com/watch?v=Btk8wDUVs74&t=890s) Um, so this is only from the past, uh, uh, day or few days window. This is something that's relevant more for today, and this is how Sidekick can answer those questions.

[15:01](https://www.youtube.com/watch?v=Btk8wDUVs74&t=901s) So to summarize it, the bottleneck was never, uh, the capability of where you ca- where we take all the data from. It was the understanding, how you connect each one of the dots to each other.

[15:12](https://www.youtube.com/watch?v=Btk8wDUVs74&t=912s) Um, the most capable agent in the world, whether it's, like, uh, Claude and Gemini, it doesn't understand you. He need to process it beforehand. This is what we're trying to solve at Sidekick.

[15:22](https://www.youtube.com/watch?v=Btk8wDUVs74&t=922s) Um, this is the Monday world model. Uh, this is what we are building. Um, if you have any question, uh, we are out of time. We are here, uh, uh, near the stage.

[15:31](https://www.youtube.com/watch?v=Btk8wDUVs74&t=931s) Thank you ve- very much for the time. Uh, you can follow us on LinkedIn. Uh, we post things on, uh, on the area. Thank you very much. [audience applauding] [outro jingle]
