AI Engineer World's Fair 2025
From Hype to Habit: How We’re Building an AI-First SaaS Company—While Still Shipping the Roadmap
Loading slides…
Read the talk
From Hype to Habit: How We’re Building an AI-First SaaS Company—While Still Shipping the Roadmap
Sprout Social’s transformation connects product strategy, discovery, and people development while preserving the customer commitments that fund the next wave of innovation.
From a talk by Rossella Blatt Vital and Deepsha Menghani
Making AI-first real inside an existing company
How do you make AI-first real inside a company without breaking the business, the team, or its values? For Rossella Blatt Vital and Deepsha Menghani, Sprout Social’s transformation is an ongoing attempt to answer that question. There is no finished playbook: the work involves trade-offs, and the right choices depend on the company making them.
Rossella opens the definition with a teenage-sex analogy: everyone talks about it, assumes everyone else is doing it, and claims to be doing it themselves. The joke captures the gap between adopting the label and knowing how to put it into practice.
AI-first changes how a company plans, builds, and delivers value. Adding intelligence to individual features is only a starting point; the larger shift puts AI into strategic decisions about what the company could become. Rossella contrasts this with cloud-first, mobile-first, and DevOps: in her framing, AI presses on product, architecture, people, process, and ethics simultaneously. That breadth makes transformation difficult to isolate within a single team.
This is a spectrum, not a switch or an instruction to rebuild everything overnight. A company needs to understand its current position and move deliberately across three dimensions:
| Dimension | Central question |
|---|---|
| Strategy | What do we prioritize, and why? |
| Ways of working | How do we build, ship, and adapt? |
| People | How must teams and skills evolve? |
The dimensions provide a way to organize the work without pretending that every company will follow the same path.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Fund the future without abandoning the present
The strategy shift starts by changing the question used to select work. An AI-enhanced strategy asks where intelligence could improve an existing experience. An AI-first strategy asks what experiences have become possible that could not have existed before. That opens the search to previously unsolvable problems and possibilities customers may not yet imagine.
But customers still need features today. Investing too heavily in the present risks missing the technological shift; investing only in the future risks disappointing customers, slowing revenue, and starving innovation of the resources it needs. Rossella invokes the innovator’s dilemma to describe this tension, using “10 times harder” as rhetorical emphasis on AI’s added pressure. Enterprise discipline and startup curiosity have to operate in parallel: steering a ship while launching a rocket. The organizational challenge is to build the startup inside the existing company.
Deepsha illustrates the planning pressure with a three-month roadmap that can feel stale in three weeks. As capabilities change, planning moves away from a deterministic account of what teams will build months in advance. Learning and discovery shape the path—and can change the destination itself, not just the implementation used to reach it.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Woofwell: an experience that crosses team boundaries
The change in planning leads to a change in design. Customers expect an intelligent interaction to work across workflows and roles, even when the underlying features belong to different teams. Deepsha makes this concrete with Woofwell, an illustrative puppy health app, and its customer, Pineapples. The app includes meal logging, activity tracking, digestive insights—the poop log—and supplement recommendations. Each feature has a separate team behind it.
When Pineapples says, “I'm feeling kind of blah,” he is not selecting a department or a feature. A meal-focused answer might point to skipped breakfast. A digestive answer might point to abnormal poop. Each uses relevant information, but each sees only a narrow part of the situation.
A unified response could connect new kibble, declining activity, abnormal poop, and a newly introduced supplement. It could then suggest pausing the supplement and checking how Pineapples feels tomorrow. Within this illustrative scenario, that is a proposed response, not an action the app has executed.
| Experience | Context used | Response scope |
|---|---|---|
| Meal feature | Skipped breakfast | One possible explanation |
| Digestive feature | Abnormal poop | One symptom |
| Unified interaction | Food, activity, digestion, supplements | Connected explanation and proposed next step |
The value comes from combining context across feature boundaries. Separate ownership need not dictate separate customer experiences. This broader value can still be unlocked incrementally through the roadmap.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Make discovery a repeatable part of delivery
Building across workflows also changes how teams collaborate and learn. Deepsha describes an earlier pattern of reactive innovation: someone proposes an idea, a team tests it in isolation or runs a spike, and attention moves to the next idea. That can produce inspiration without becoming a sustained driver of strategy.
Ritualized discovery reserves time inside planning cycles for experiments, hackathons, and learning forums. The findings need to be visible and actionable across the company’s strategy, so discovery can influence what happens next rather than remain inside an isolated experiment.
The purpose of an MVP changes accordingly. It is a way to validate direction, not merely a way to launch faster. Some bets will not land, and some features will not behave as expected. Those results can still reduce ambiguity by clarifying which direction deserves further investment.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Treat process as a product
Processes built for linear roadmaps and predictable risks can become a poor fit when timelines shift and capabilities evolve monthly. Yet adding more process is not automatically the answer. Deepsha reports that Sprout learned the cost of introducing too many new processes at once: overhead and drag replaced the intended help.
Treating process as a product means evaluating it against outcomes:
- Direction: Does it make the next decision clearer?
- Flow: Does it unblock teams?
- Decision quality: Does it help people make better decisions faster?
If a process fails those tests, iterate on it or remove it. Purposeful process earns its place by helping teams move with clarity.
Deepsha recalls the previous day’s keynotes describing execution as the moat, then adds the condition that makes speed useful. Speed without direction produces chaos; direction without momentum produces stagnation. Sprout calls the desired combination smart velocity: clarity, momentum, and adaptability together. Rapid prototypes and MVPs serve that combination, rather than making shipping an end in itself.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Keep the depth and expand the range
Strategy establishes direction and ways of working shape execution, but people make both real. Rossella describes AI-first primarily as a cultural transformation: it changes how people think, work, lead, and feel. Sprout’s people priorities therefore span both the evolution of specialist talent and AI fluency across the organization.
The new expectations build on existing expertise. Deep AI specialization remains essential, but versatility and the ability to imagine and build new experiences become more central. T-shaped talent combines depth with the breadth to prototype quickly, collaborate across silos, and bring complete systems to life.
Rossella’s comparison is Indiana Jones: the professor and adventurer combined. The specialist brings knowledge into unfamiliar terrain rather than leaving it behind. As roadmaps and technology shift, teams need people who can navigate ambiguity while retaining their expertise. Sometimes there is no path to follow yet; imagining one becomes part of the builder’s job.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Make AI participation possible across the company
Specialist development is only part of the work. As Rossella puts it, “You can't go AI-first if the rest of the company is still AI-last.” Marketers, designers, product managers, and care agents also need enough understanding and confidence to work with AI. The aim is to make it usable, safe, and relevant to everyday tasks.
Sprout supports that fluency through AI newsletters, podcasts, cross-functional show-and-tell, and empowering teams to use tools that help them work smarter. These channels make exploration a shared activity across functions.
Fluency also needs a path to delivery. Rossella describes an aim to build a self-service platform through which product and engineering teams could prototype and ship AI-powered features without deep involvement from the AI team. The platform is a planned enabler in this account. Its purpose is to make AI thinking and exploration routine without requiring everyone to become an AI expert.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Keep customer needs and human judgment central
The scale of change does not remove the product fundamentals. Useful AI features still solve customer problems rather than merely showcase novelty. User experience, performance, reliability, and trust remain non-negotiable.
Human creativity, judgment, and care remain central to leadership and decisions. That requires honesty about trade-offs and investment in people alongside models and agents. Deepsha’s call to “Kill the roadmap if needed” follows those commitments: plans must remain revisable as teams learn. Be bold, learn openly, and give an unusual idea a chance to ship.
The framework is a flashlight in a maze, not a guarantee of a linear journey. Teams do not need every answer before starting; they need questions that help them learn and the conviction to evolve as the answers emerge.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
When transformation becomes everyday work
Rossella closes by connecting technological change to daily life: transformative inventions become revolutionary when people incorporate them into ordinary routines. The opportunity carries a responsibility to guide teams, companies, and communities through changes in how they live, work, and connect.
That makes AI transformation a human undertaking as much as a technical one. Turning possibility into habit takes time, grit, and patience through setbacks, doubt, and growing pains. Those are part of the work of making the technology useful in everyday life.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Resources
Further reading
Clayton Christensen's foundational account of why successful firms can miss new waves of innovation.
- Sprout Social’s April 2025 product releasesDocumentation
Historical release notes covering AI-assisted post generation and other product improvements shortly before the talk.
Full transcript
Loading transcript…