I’ve led and observed AI initiatives across fast-moving product organizations, and one pattern is unmistakable: “The AI revolution needs a departmental leader.” When that leader is unclear, pilots stall, risk mounts, and value gets trapped in proof-of-concept purgatory. When it’s clear, AI moves from demos to durable outcomes.
In my experience, IT is uniquely positioned to play that leadership role. IT sits at the nexus of data, identity, security, and infrastructure—exactly where scalable AI capabilities live. IT also has the vantage point to connect use cases across teams, manage risk, and operationalize change without derailing core systems.
Put simply, this is the promise: “Learn the key reasons why IT teams are uniquely positioned to be the strategic leaders of your company’s AI projects.” The reasons are pragmatic—access to systems of record, stewardship of data governance, ownership of integration patterns, and accountability for reliability and compliance—yet the impact is strategic.
Here’s how I frame the operating model. IT provides strategic leadership and platform stewardship; Product owns the outcomes; Engineering delivers services and integrations; Security and Legal codify guardrails; and Finance supports cost modeling. We establish tight collaboration through product trios (Product, Design, Engineering) that plug into an IT-led AI platform, enabling empowered product teams to ship safely and quickly.
Governance turns intent into repeatable action. I use outcomes vs output OKRs to force clarity on value, pair them with lightweight QBR cadences for course correction, and require architecture reviews that cover model/data governance, observability, privacy, and vendor risk. This ensures we can scale gen ai without surprise failures or compliance gaps.
On the delivery side, forward deployed engineers embedded with business units accelerate discovery and reduce translation loss. We leverage gen ai for product prototyping to validate desirability and feasibility early, then harden solutions on our shared AI platform. This keeps experimentation fast while maintaining an enterprise-grade backbone.
Roadmapping balances ambition with throughput. I tie product roadmapping and sprint planning to value streams, not just features, and I make stakeholder management explicit—especially with customer support, finance, and operations—so we design for adoption. For example, a customer support ai strategy isn’t a chatbot alone; it’s an outcome-driven service redesign, with training, playbooks, and measurable deflection and CSAT targets.
Success demands the right metrics. Beyond typical velocity measures, I track time-to-first-value, model quality and drift, cost-to-serve, and risk posture. These roll into OKRs that link frontline improvements (e.g., resolution time) to enterprise outcomes (e.g., gross margin, retention), giving executives confidence and teams a clear definition of done.
If you lead IT, this is your moment to step into strategic ownership and elevate AI from scattered experiments to a coherent platform. If you lead Product, partner with IT to align discovery, outcomes, and guardrails so empowered teams can move fast and responsibly. Together, we can turn AI from a buzzword into a durable advantage.
AI Agents are reshaping how businesses deliver service, earn loyalty, and create measurable value. From my vantage point leading product management, I see this shift accelerating across support and CX as organizations move from experiments to production-grade systems.
Very soon, I believe AI Agents will handle the majority of customer service – and eventually, every customer interaction. Human teams won’t disappear, but their roles will evolve from answering questions to analyzing performance, improving systems, and designing better customer experiences.
The pressure to adopt AI is real. So is the opportunity. The leaders who win won’t just add technology; they’ll redesign operations to capture durable value while safeguarding customer trust.
But for many support leaders, the path forward is still unclear. Where do you start? What should success look like? How do you actually test and deploy these solutions? I hear these questions every week, and I’ve seen promising initiatives stall without a clear roadmap, evaluation framework, or governance model.
That’s why we created The AI Agent Blueprint: a strategic map for support, CX, and AI transformation leaders. It’s designed to help you launch fast, scale with confidence, and achieve meaningful business transformation with AI.
1. Launch it. Go from zero to a successful deployment. Get immediate value from an AI Agent. 2. Scale it. Rewire your organization to sustain and expand impact.
Part 1 – “Launch it”
Learn how to unlock immediate efficiency and value from an AI Agent. We cover how to build a business case, evaluate and deploy an AI Agent, and prove its impact, fast.
You’ll learn how to:
Launch it. Scale it. The AI Agent Blueprint lays out a clear framework to deploy and grow automation in customer service. Explore the step-by-step guide at fin.ai/blueprint and turn pilots into production results.
Get clear on what an AI Agent is: Discover why they’re different from chatbots and how they work. Build a business case: Prove the basic economics of AI, decide whether to buy or build, and get the buy-in and budget you need to move forward. Evaluate an AI Agent: Learn how to define success, choose the right evaluation criteria, and run a focused, high-impact assessment with our four-step framework. Deploy with confidence: Build a deployment plan that balances speed with safety. Learn what to expect at each stage. Continuously improve performance: After launch, your AI Agent becomes a system to manage. We’ll show you how to implement a repeatable process to train, test, deploy, and optimize.
Part 2 – “Scale it”
Launching AI is only the beginning. To unlock its full potential, you need to rewire your systems across three core pillars: → Customer experience → Organizational and system design → Economics
If you stop at launch, results will plateau. Your team won’t transform how they work. The system won’t evolve – and neither will the value.
The second part of the Blueprint shows you how to scale AI intentionally and sustainably. That means:
Designing AI-first customer journeys and building trust with AI. Embedding new roles, AI-first systems, governance structures, and ownership models. Rethinking how support is measured and funded in an AI-first world by exploring new metrics, ROI models, and reinvestment strategies to elevate your support function from a cost center to a strategic growth lever.
This is where AI becomes infrastructure and support becomes a lever for growth.
In practice, this is how I recommend teams approach their customer support AI strategy: start with a narrow, high-value use case, define your success metrics and guardrails, and iterate quickly with human-in-the-loop quality reviews. Once you establish confidence, expand coverage, evolve your organizational design and governance, and update your ROI model to reinvest efficiency gains into customer experience. This blueprint distills the lessons I’ve learned guiding gen AI programs from pilot to platform—so you can accelerate time-to-value and de-risk deployment.
Generative media is no longer a curiosity on the edges of product roadmaps—it’s fast becoming a core capability. Watching one company sprint from uncertainty to undeniable traction reminded me how much a decisive pivot, a developer-first brand, and ruthless focus can bend a growth curve. This is a story about finding product-market fit in real time, scaling with intention, and staying lean while the category accelerates beneath your feet.
Gorkem Yurtseven is the co-founder and CEO of fal, the generative media platform powering the next wave of image, video, and audio applications. In less than two years, fal has scaled from $2M to over $100M in ARR, serving over 2 million developers and more than 300 enterprises, including Adobe, Canva, and Shopify. In this conversation, Gorkem shares the inside story of fal’s pivot into explosive growth, the technical and cultural philosophies driving its success, and his predictions for the future of AI-generated media.
What stood out to me first was the clarity of the pivot: “How fal pivoted from data infrastructure to generative inference.” The hardest decisions often feel like abandonment—of code, roadmap, and even identity—but the right pivot reframes everything around a higher-signal customer need. That decision, described as “The hardest decision that saved the company,” unlocked a new trajectory and set a crisp north star for the team.
Equally important was the market intuition. As they put it, “Why ‘generative media’ is a greenfield new market.” Greenfield means pattern-breaking strategy: prioritize outcomes over parity, embrace new workflows rather than retrofit old ones, and measure value in quality, latency, and unit economics—not just features. In my experience, this is where product teams win or lose: you either build the new default or get trapped perfecting the old one.
fal’s “explosive year” wasn’t luck; it was systems thinking applied to a developer platform. The team stayed small—”lean <50-person team” and “Staying nimble as a 45-person company”—and built a brand that feels genuinely for builders: “Building a brand that resonates with developers.” That shows up in everything from docs and SDKs to the cultural quirks that scale signal, like “Why fal has 500 Slack channels.” Velocity and clarity compound when communication is designed for ownership.
Early traction came from sharp use cases and fast feedback loops. I loved the transition arc from “The early adopters of the first fal product” to “The transition from toy to tool.” In a new category, the fastest path to durable usage is making something delightful and then relentlessly hardening it for production: uptime targets, deterministic APIs, transparent pricing, and repeatable performance. That’s how you move from demos to dependable workflows.
The timing call is bold and specific: “Why 2025 is the year of AI-generated video” and “Predicting AI-generated film in 2027.” If you build in gen AI, this matters. Video will force teams to optimize for cost per second, temporal coherence, and developer ergonomics across long-running jobs. The winners will combine model choice (OpenAI, Anthropic, Google DeepMind, Stability AI; “Stable Diffusion XL (SDXL)”, “Sora”, “DALL-E”, “LLaMA”) with world-class inference, smart caching, and autoscaling that feels invisible to the developer.
On the go-to-market side, I see a masterclass in founder-led GTM and developer evangelism. “Competing in a fast-moving, fragmented market” requires sharp messaging and distinctive ideas. The story behind “GPU Rich / GPU Poor” is a perfect example: a memorable narrative that encodes a real infrastructure advantage. Pair that with “fal’s greatest optimization wins” and you get a brand promise rooted in measurable performance, not just clever copy.
Culture and team design are the force multipliers. “How to build a world-class team” and “fal’s unique hiring philosophy” emphasize high-slope talent, ownership, and speed over headcount. The result is a product org that ships, learns, and iterates without bureaucratic drag. For technical founders, “Learning sales as a technical founder” is a reminder that the best sales motion often emerges from the same instincts as great product discovery: ask better questions, observe real workflows, and sell through outcomes.
Here’s how I translate these lessons into a practical playbook for product leaders working in gen ai and developer platforms: double down on developer experience (time-to-first-output, clear pricing, robust SDKs), make latency and reliability your product features, sequence the roadmap from delightful demos to dependable production tools, and stay lean enough to pivot as models and use cases evolve. Above all, treat “Why generative media is a greenfield market” as a call to invent the defaults others will copy.
Looking ahead, the path is clear: as AI-generated video normalizes in 2025 and professional-grade content follows by 2027, the products that win will combine inference excellence with a brand developers trust. If you’re building in this space, now is the moment to ship fast, optimize relentlessly, and meet creators and developers where they already work.
I obsess over building high-velocity engineering organizations that ship meaningful outcomes. When I evaluate what reliably moves the needle—across startups and scaled enterprises—it always comes back to alignment, disciplined management, and a modern view of engineering productivity. Recently, I revisited a set of insights that crystallize these themes and translate them into practical rituals any leader can adopt.
Kellan Elliott-McCrea is a Head of Engineering at Adobe, overseeing Frame.io, a newly acquired video review and collaboration platform. He is known for his experience and expertise as an engineering leader. He was previously a VPE at Dropbox, and CTO at Etsy where he built and led a team of 300 people, from tech and platform reboot through to IPO. Kellan also built and scaled teams at Flickr, and has a coaching and advising practice for companies looking to supercharge their engineering teams.
Here’s what we dig into when we talk about world-class engineering orgs: how software engineering has changed in the last 10-15 years; the future of software engineering, and the impact of AI; the importance of alignment and tactics for achieving it; how to think about and enable engineering productivity; lessons on culture from Adobe, Dropbox, and Flickr; concrete tips for being a better manager; and rituals for building business literacy throughout an org.
Let’s start with a reality I see in my own work: engineering teams are bigger than they were a decade ago, despite dramatically better tools and platforms. The reason isn’t inefficiency—it’s scope. Today’s products carry higher bars for reliability, privacy, security, compliance, and multi-surface experience. The coordination surface area has exploded. That’s why operating models must evolve: clear interfaces between teams, standardized decision-making, and reliable cross-functional rhythms are no longer nice-to-haves—they’re throughput constraints.
Alignment, then, is the ultimate speed multiplier. I’ve learned the hard way that slow teams are rarely under-skilled; they’re misaligned. “Slow teams are misaligned teams.” To counter this, I anchor on a few tactics: articulate a clear strategic narrative (why now, why us, why this), commit to outcomes vs output OKRs, and institutionalize decision logs so debates don’t reset every sprint. When teams know the customer problem, the business bet, and how their work ladders up, the flywheel starts turning.
On engineering productivity, I avoid vanity metrics and favor a portfolio: flow and focus (interruptions, WIP), system signals (lead time, deployment frequency, change fail rate), and outcome alignment (how progress maps to customer value and revenue impact). Tools matter—DX investment in CI/CD, observability, and paved roads—yet the largest gains usually come from simplifying priorities and reducing cross-team coupling. Fewer, better bets will beat “more tickets shipped” every time.
The future of software engineering is inseparable from AI. In my practice, I treat gen ai and gen ai for product prototyping as core accelerators: copilots for code and tests, scaffolding services that convert specs to boilerplate, and retrieval-augmented knowledge that collapses the gap between tribal lore and action. The key is to measure impact at the team level—cycle time, defect escape, and learning velocity—so AI augments engineering judgment rather than creating hidden complexity.
Culture is the compounding edge. Lessons on culture from Adobe, Dropbox, and Flickr converge on a few essentials: invest in psychological safety and clarity of purpose, operationalize blameless learning, and make information radically accessible. “How Complex Systems Fail, by Richard I. Cook, MD” is a touchstone here—complexity punishes organizations that rely on heroics and rewards those that build resilient systems and shared mental models.
For managers, I return to a short, durable list. Schedule real one-on-ones that prioritize coaching over status. Write more than you speak; clarity scales through documents. Run crisp, time-boxed decision forums with pre-reads and owners. Close the loop on feedback—especially in moments of disagreement—by documenting trade-offs and naming the decider. These concrete tips for being a better manager build trust, accelerate decisions, and enable autonomy.
Every high-performing engineering org I’ve led invests in business literacy as a first-class ritual. I recommend monthly “Finance 101” briefings, customer support ride-alongs, and deal reviews to connect engineers to revenue realities. Pair that with tactics and rituals for enabling effective teams—weekly written updates, demo-driven reviews, and pre-mortems—and you get sharper prioritization and far better cross-functional coordination.
Why so few companies successfully go multi-product? Most underinvest in platforms, shared services, and explicit funding models for internal APIs. The remedy: treat platforms as products with clear roadmaps, SLAs, and customer empathy; align incentives so teams don’t fork capabilities in the rush to ship; and adopt technical governance that favors standardization where it compounds and freedom where it differentiates.
For compensation and career architecture, I pressure-test common models by asking: does this design reward the behaviors we say we want? If we value outcomes, impact, and enabling others, the ladders should reflect it. When the incentives match the mission, the org learns faster and scales cleaner.
My bottom line: if you want to supercharge your engineering org, anchor on alignment, measure what matters, and leverage AI to elevate—not replace—engineering judgment. Do that, and you’ll turn coordination costs into compounding advantages that show up in customer value, velocity, and morale.
The ground rules for product development have changed in the post-LLM world. I’m sharing a practical, first-person playbook—lessons I’ve pressure-tested in my own product org—to help you build AI-native products with confidence, cut through hype, and deliver outcomes that compound.
Sprig is an AI-powered user insights platform that has raised over $88m. Today’s discussion features two key individuals in Sprig’s journey so far: Ryan Glasgow, Sprig’s CEO and founder; and Kevin Mandich, Sprig’s Head of Machine Learning. Before Sprig, Ryan was an early PM at GraphScience, Vurb, and Weeby (all of which were acquired), and Kevin was an ML Engineer at Incubit, and a Post-Doctoral Researcher at UC San Diego.
In today’s episode, we discuss: Key lessons from the Sprig founding story; Product development in the pre vs. post-LLM world; How to overcome AI skepticism; How to evaluate new models and how to know when to switch; Why you need an ML engineer; Sprig’s “AI Squad” team structure; How Sprig upskills all team members on AI.
Founding story takeaways I keep returning to: conviction compounds when paired with continuous discovery. Early on, prioritize direct customer signal over elegant architectures. I’ve seen the fastest learning loops come from a tight PM–ML partnership that prototypes quickly, validates with real users, and refactors only after signal stabilizes. The Jobs to Be Done Framework: https://hbr.org/2016/09/know-your-customers-jobs-to-be-done remains my favorite lens to separate what the model can do from what the customer actually needs done.
Pre vs. post-LLM product development requires a mindset shift. Pre-LLM, we wrote deterministic systems and pushed the edge with models like Google’s BERT model: https://en.wikipedia.org/wiki/BERT_(language_model). Post-LLM, we design probabilistic systems, treat prompts like code, and invest in evaluation harnesses from day one. I routinely prototype with Chat GPT: https://chat.openai.com and scaffold experiments with Langchain: https://www.langchain.com/ to compress discovery cycles. The key is shipping guardrails and UX affordances that make non-determinism feel trustworthy.
On AI skepticism, I don’t argue—I demonstrate. I target one painful workflow, build a narrow, high-precision solution, and expose transparent failure modes with a human-in-the-loop escape hatch. This reframes AI from magic to leverage. In customer-facing settings (think customer support ai strategy), we measure deflection and satisfaction together so automation never outpaces user psychology.
Evaluating new models—and knowing when to switch—demands a clear rubric: task quality (ground-truthed), latency at p95, unit economics, privacy/compliance, and operational reliability. I run shadow evaluations before swapping production dependencies, then phase changes behind flags with canaries and backstops. Tools like Auto-GPT: https://github.com/Significant-Gravitas/Auto-GPT are useful for ideation, but I never skip rigorous offline and online evaluation before a cutover.
Why you need an ML engineer: the fastest teams pair a product manager who owns the problem framing with an ML engineer who owns the feasibility frontier. This duo translates ambiguous jobs into measurable tasks, instrumented datasets, and iterative model/UX improvements. In my experience, this partnership reduces time-to-learning more than any single tooling decision.
Sprig’s “AI Squad” team structure mirrors what I’ve seen work: a cross-functional pod with a PM, ML engineer, data engineer/analyst, design, and platform partner. The squad ships thin slices end-to-end, owns their eval suite, and meets weekly to review errors, edge cases, and customer feedback. We track outcomes vs output OKRs to ensure velocity serves impact—not the other way around.
Upskilling the entire team on AI is non-negotiable. I’ve had success with lightweight rituals: weekly demo hours, prompt libraries maintained in Jira: https://www.atlassian.com/software/jira, red-team exercises to uncover failure patterns, and internal brown bags where engineers and PMs teach each other. Small, frequent exposure beats heavyweight training.
For deeper exploration and hands-on experimentation, I reference: Auto-GPT: https://github.com/Significant-Gravitas/Auto-GPT; Chat GPT: https://chat.openai.com; Google’s BERT model: https://en.wikipedia.org/wiki/BERT_(language_model); Jira: https://www.atlassian.com/software/jira; Jobs to Be Done Framework: https://hbr.org/2016/09/know-your-customers-jobs-to-be-done; Langchain: https://www.langchain.com/; Sprig: https://sprig.com/.
Timestamps: (02:50) Intro (04:57) What attracted Kevin to Sprig (05:53) Kevin’s background before Sprig (07:56) How Ryan gained conviction about Kevin (09:55) Key technical challenges and how they solved them (18:46) How to overcome AI skepticism (21:47) The early difficulties of building an ML-enabled product (25:06) Evaluating new models and knowing when to switch (35:09) Using Chat GPT (37:23) Product development in the pre vs. post-LLM world (39:53) The impact of AI hype on Sprig’s product development (45:36) Balancing AI automation with user-psychology (48:47) Do recent LLMs reduce Sprig’s competitive advantage? (51:00) The importance of “selling the vision” to customers (54:40) How Sprig structures teams (57:25) How Sprig upskills all team members on AI (60:25) 3 key tips for companies trying to navigate AI (66:05) Major limitations with LLMs right now (70:27) The future of AI and the future of Sprig
Three guiding principles I use daily: first, reduce surface area—start with one high-value job and earn trust with reliability. Second, treat evaluation as a product—version prompts, log failures, and continuously retrain on your own data distributions. Third, design for collaboration—pair AI with human judgment and transparent controls so users feel empowered, not replaced. Post-LLM success isn’t about chasing models; it’s about building resilient systems, teams, and learning loops.
I sat down with Dan Siroker to explore the product, fundraising, and AI strategy lessons behind Rewind AI’s rapid rise — and to reflect on what I would adopt in my own product management practice today. Dan Siroker is the co-founder and CEO at Rewind AI, a personalized AI powered by everything you’ve seen, said, or heard. Dan launched Rewind to an emphatic response on Twitter, and used a public pitch video to fundraise at a $350m valuation. Prior to starting Rewind, Dan co-founded Optimizely, which reached $120m ARR before being acquired by Episerver, a content management company. Dan was also the Director of Analytics for Obama’s first presidential campaign.
What stood out immediately was Rewind’s journey to Product Market Fit and how deliberately the team instrumented learning loops. As a product leader, I pay close attention to how founders reduce ambiguity: narrow the target segment, ship thin slices, measure engagement cohorts, and iterate fast. Rewind’s early focus on utility and trust — not novelty — created the conditions for PMF while the team resisted the temptation to over-scope.
I was especially interested in how Rewind works and how the team managed scope while building a category-creating product. By focusing on personalized recall powered by on-device intelligence and a clear privacy narrative, they avoided the common trap of trying to solve everything for everyone. My own rule of thumb is to enforce brutal prioritization around the highest-intent jobs-to-be-done, then earn the right to expand. That same discipline shows up in Rewind’s cultural mantra for shipping and validating fast.
Lessons from Optimizely echo throughout. Being a second-time founder sharpens pattern recognition — from building high-clarity cultural values to operationalizing product-market fit. I’ve found that codifying operating principles early helps a team move faster with fewer collisions, and Dan’s approach to open feedback and public learning raises the bar for transparency.
On product positioning as a category creator, the team leaned into outcomes over features, which is critical when the mental model is new. Rather than compete in a features arms race, they framed a compelling before-and-after: instant, searchable memory that augments cognition. In my experience, that level of narrative clarity drives founder-led GTM and accelerates word-of-mouth.
We also dug into where to build in AI, and what makes a “wrapper” thin versus thick. My take: thin wrappers add shallow convenience on top of foundation models; thick wrappers integrate proprietary data, workflow depth, distribution advantages, and durable UX moats. Founders should aim for thick wrappers with unique data flywheels, not commodity interfaces easily displaced by platform shifts.
Operationalizing Product Market Fit remains a craft. I routinely use leading indicators like activation rate, day-7/day-30 retention for key actions, and sentiment via structured PMF surveys. Rahul Vohra’s framework for measuring and optimizing Product Market Fit: https://review.firstround.com/how-superhuman-built-an-engine-to-find-product-market-fit is a proven playbook. Pair that with cohort-based instrumentation and tight audience segmentation to reveal the “sharpest edge” of value.
On AI hype, we aligned on a pragmatic view: real value accrues where latency, accuracy, and privacy meet workflow depth. Apple’s Silicon: https://www.macrumors.com/guide/apple-silicon/ and on-device acceleration will keep unlocking new consumer experiences, while ChatGPT: https://chat.openai.com/ has reset expectations for natural interfaces. The cautionary tales of Google Glass: https://en.wikipedia.org/wiki/Google_Glass and Google Wave: https://en.wikipedia.org/wiki/Google_Wave remind me that timing, social acceptability, and use-case clarity matter as much as technical novelty.
Data privacy is now a core buying criterion, not a checkbox. I see a clear trend toward local-first approaches, explicit consent, and user agency — especially for products that touch memory, identity, and personal archives. Framing value through Maslow’s Hierarchy of Needs: https://www.simplypsychology.org/maslow.html helps prioritize trustworthy utility over gimmicks.
Dan’s one-of-a-kind Twitter fundraising strategy was a masterclass in founder-led GTM. By sharing a public pitch and engaging directly with early users and supporters, he compressed feedback cycles and aligned community, product, and capital. For reference, see Dan’s public Twitter fundraise: https://twitter.com/dsiroker/status/1646895452317700097 and Dan’s Rewind demo tweet: https://twitter.com/dsiroker/status/1638799931891920897. The transparency extended to leadership practice as well, with Dan publicly sharing his own 360 performance reviews: https://twitter.com/dsiroker/status/1689763756459675650 — a bold move that builds trust.
I’m watching what’s next for Rewind with interest, particularly around thicker integrations, extensibility, and collaboration patterns. In the next decade, I expect assistive AI to become ambient, multimodal, and context-aware — an ever-present copilot that feels less like a tool and more like an extension of cognition.
Referenced: Paul Graham: https://twitter.com/paulg
Referenced: Rahul Vohra’s framework for measuring and optimizing Product Market Fit: https://review.firstround.com/how-superhuman-built-an-engine-to-find-product-market-fit
Referenced: Rewind AI: https://www.rewind.ai/
Referenced: Scribe (which morphed into Rewind): https://www.scribe.ai/about
Where to find Dan Siroker: Twitter: https://twitter.com/dsiroker
Where to find Dan Siroker: LinkedIn: https://www.linkedin.com/in/dsiroker
Where to find Dan Siroker: Personal website: https://siroker.com/
Where to find Dan Siroker: Blog: https://medium.com/@dsiroker
My takeaway for founders and product leaders: obsess over segmentation, instrument for learning, and tell a crisp narrative that earns trust. Thick wrappers, privacy-first design, and founder-led GTM are how you win the next wave of AI.
When I study category-defining products, I look for the decisive moment where simplification unlocks scale. Few stories illustrate this better. Guillermo Rauch is the CEO of Vercel, a frontend-as-a-service product that was valued at $2.5b in 2021. That headline number matters, but the underlying playbook matters more: simplify the developer path to value, create a default that feels inevitable, and let adoption compound.
Vercel serves customers like Uber, Notion and Zapier, and their React framework – Next.js – is used by over 500,000 developers and designers worldwide. For a product creator, those are the signatures of extreme product-market fit: an obvious customer set, a loveable default framework, and a platform that scales with developer ambition. From a product management leadership lens, this is a masterclass in developer evangelism and founder-led GTM, not just technology.
Guillermo started his first company at age 11 in Buenos Aires and moved to San Francisco at age 18. In 2013, he sold his company Cloudup to Automattic (the company behind WordPress), and in 2015 he founded Vercel. I read this arc as a sequence of product discovery moments: start with a sharp user problem, ship the smallest credible solution, and iterate where the usage is loudest. The throughline is obsession with experience—reducing friction until the product’s default path feels like magic.
Reflecting on the Cloudup era, I see a blueprint for outcomes vs output OKRs. Shipping features is easy; aligning them to a few hard outcomes is what prepares a company for scale or acquisition. That discipline shows up later in Vercel’s sequencing: tight technical scope, clear constraints, and relentless measurement of time-to-first-value.
On origin and early validation, the insight was deceptively simple: give frontend teams a zero-config way to build, preview, and ship. The V1 product wasn’t a kitchen sink—it was a clean, repeatable flow from commit to deploy. The early skeptics (and there are always skeptics) helped refine the edges; real usage pressure-tested the defaults. The paradox of developers is alive here: we demand power without complexity. The genius is delivering depth without exposing every knob on day one—Next.js did exactly that.
My advice on finding product-market fit mirrors this path: collapse the distance between intent and impact. Design the onboarding so one successful path feels pre-ordained. Put forward deployed engineers beside the customer, and treat their feedback as your fastest route to truth. Keep founder-led GTM longer than you think; it’s the most direct signal path you’ll ever have.
An open source business becomes successful when adoption is the front door and the cloud is the living room. Open source monetization works when you resist taxing the developer and instead charge for the operational guarantees that companies need at scale: performance, security, governance, and global reliability. Next.js as a community engine and Vercel as a managed “frontend-as-a-service” is a textbook pairing.
The trend toward a “Front-end Cloud” is structural. As teams modularize on services like AWS and adopt modern stacks with Next.js, React Native, and headless partners such as Contentful or Shopify, the frontend becomes the primary assembly layer. That’s why people now pay so much attention to the front-end: it’s where the brand lives, the iteration cycles are fastest, and the performance budget is now a business KPI.
Positioning and category creation here relied on clarity over cleverness. Name the job-to-be-done, anchor on speed and reliability, and make the default workflow visibly better than the DIY alternative. When the default wins, you earn the right to go multi-product. The key is sequencing: expand from core strengths and ship adjacent capabilities that shorten time-to-value across the same journey.
On AI, I’m seeing gen ai shift from novelty to necessity. The immediate wins are in gen ai for product prototyping (faster ideation, copy, and component scaffolding) and in developer experience (test generation, refactors, and safe migrations). The long arc over 10–20 years points to engineering where we curate constraints and verify outcomes, while machines propose implementations. That raises the bar for PM rigor: better problem statements, tighter acceptance criteria, and sharper product discovery.
My enduring heuristics for building better product experiences are simple. Eliminate decisions the user shouldn’t have to make. Make the fastest path the default path. Optimize for the preview moment because that’s where confidence is built. And measure success by how little the user has to think to achieve a powerful result. If you apply that mindset—plus disciplined developer evangelism and thoughtful open source monetization—you give your product a real shot at extreme product-market fit.
I build and ship AI products in an environment where the frontier changes weekly, so my planning system has to be adaptive, evidence-driven, and unapologetically outcome-focused. In this piece, I share the frameworks I use to set goals for generative AI, balance research with product execution, and scale responsibly — drawing sharp lessons from one of the most influential applied AI companies operating today.
Consider Runway, an applied AI research company shaping the next era of art, entertainment, and human creativity. Runway has raised $237m and was one of Time Magazine’s “100 most influential companies” in 2023. Runway has been a persistent viral sensation in recent years, and is behind many of the most famous AI demos online.
I’ve learned to be cautious about “The limitations of being “customer-driven” when building in AI”. Traditional product discovery assumes needs are legible and solutions are relatively deterministic. In generative AI, user desire often follows model capability, not the other way around. The job is to triangulate: run tight user loops to validate perceived value, instrument objective model quality, and explore novel interaction patterns that customers can’t yet articulate. I treat this as a portfolio of discovery bets — some customer-led, some capability-led, all evaluated against clear outcome thresholds.
Balancing research development with product development requires organizational design that prevents context-switching tax while preserving velocity. I pair research pods with product pods, supported by forward deployed engineers and domain PMs who translate evaluation metrics into user-visible milestones. Safety and content moderation sit on the critical path, not as afterthoughts — think policy definition, classifier tooling, abuse red teaming, and clear escalation playbooks. This balance is how you move from a great demo to a dependable product without losing momentum.
Goal-setting amidst constant change in AI starts with outcomes vs output OKRs. I write OKRs in terms of user impact and model performance thresholds — for example, target ranges for latency, quality scores against a golden dataset, or creator retention — then let teams choose the highest-leverage outputs (data pipelines, fine-tuning, UX improvements) to get there. Why I don’t plan very far ahead: I treat the annual view as a vision and bet map, the quarterly view as a constrained slate of outcomes, and the 6–8 week cycle as the execution heartbeat. AI roadmaps are hypotheses; evaluation harnesses and launch gates are the truth.
Community is a force multiplier. Forming a vocal community and fostering community requires real access and real listening: early release cohorts, office hours, and transparent changelogs. How they picked users for early release matters — diversity of use cases, sophistication of workflows, and willingness to give crisp feedback. Expanding past the first 100 users of Gen-2 demands readiness: evaluation parity across modalities, scalable infra, and safety coverage. Done well, this motion compounds learning while building authentic advocacy.
For founders, my advice echoes the core lessons above. Start with a narrow, high-intent wedge and prove durable value fast; let founder-led GTM compress the feedback loop; instrument everything from day one; and resist the urge to over-plan features before you’ve nailed outcomes. Product-market fit lessons in AI often arrive via small, fast experiments — not grand, long-range plans. Ship thin slices that demonstrate unmistakable value, then iterate toward a system, not a single feature. When in doubt, shorten the loop and improve the evaluation harness.
People often ask: Will AI replace video editors? My view is that AI will replace zero editors who master these tools — and many who don’t. The winners blend taste, storytelling, and generative leverage. The products we build should honor this reality: design for control, iteration, and co-creation, not just automation.
For builders focused on gen ai for product prototyping through production: keep your demo muscle strong, your evaluation stronger, and your outcomes strongest. Invest in community, treat safety as a feature, and let your OKRs steer what ships — not the other way around.
I spend a lot of time helping teams reconcile two pressures that define modern product management: ship fast enough to learn and compete, but slow enough to be safe, ethical, and useful. Studying Bard offers a crisp blueprint for navigating that tension and leveling up how we build with Generative AI.
Jack Krawczyk is a Senior Director of Product at Google, building Bard. Bard is Google’s collaborative, conversational, and experimental AI tool that’s bridging the gap between humans and bots, while addressing ethical considerations around AI. After joining the project in 2020, Jack helped ship Bard in less than four years. Bard sources information directly from the web, and now enables users to inquire about and summarize YouTube videos.
From a product management lens, the most valuable takeaway is the sequencing: problem definition → principled constraints → rapid public learning with clear guardrails. I’ve seen this order de-risk speed. When we anchor teams on a tight product thesis and ethical framework, we unlock faster iteration without drifting into feature theater.
Shipping early—especially with a Large Language Model (LLM)—can feel risky. Yet the decision to open Bard to the public quickly reflects a disciplined bias toward learning velocity. In my experience, the longer we delay real-world feedback with LLMs, the more our internal assumptions calcify. Early exposure surfaces edge cases, calibrates safety systems, and drives better prioritization than any lab-only evaluation can.
Ethics in AI is not a separate workstream; it’s a product requirement. I anchor cross-functional reviews on harm modeling, transparency, and user agency. Bard’s framing makes this explicit: collaborative, conversational, experimental—language that signals co-creation and responsible exploration rather than unfettered automation. That positioning matters for trust and sets expectations for both quality and limitations.
Differentiation in AI assistants increasingly hinges on live context and modality. Bard sources information directly from the web, and now enables users to inquire about and summarize YouTube videos. In practice, this moves Bard beyond static Q&A toward dynamic sensemaking. I advise teams to ask: what fresh, authoritative context can our system responsibly ingest to reduce hallucinations and increase actionability?
On development speed, I look for a culture that marries ambition with measurable risk reduction. That means small, end-to-end vertical slices; evaluation harnesses aligned to user outcomes, not model vanity metrics; and weekly red-teaming that actually changes the roadmap. Outcomes vs output OKRs are critical here—optimize for quality-adjusted learning per unit time, not just feature count.
Early user research should be embedded, not episodic. I’m a proponent of forward deployed engineers paired with product and research to observe failure modes in the wild and close the loop quickly. With LLM-based experiences, qualitative signals (confusion, trust breaks, cognitive load) often precede quantitative ones; instrument both and let them inform each other.
Deciding when to ship comes down to clear thresholds. I pressure-test launch criteria with two prompts: what would change my mind tomorrow, and what could break if we’re right but too early? For AI features, I also require recovery paths—explanations, undo, source attribution—so that small misses don’t become trust-ending moments.
As for the competitive landscape—Bard versus ChatGPT, and others—users ultimately reward utility, reliability, and workflow fit. I encourage teams to pick a sharp use case, lean into their unique distribution or data advantage, and prove value in minutes, not weeks. “Generative AI” is table stakes; reliable outcomes in a real job-to-be-done is differentiation.
Zooming out, I see three fronts shaping the future of LLM, Generative AI, and AGI: model capability, grounding and retrieval quality, and product ergonomics. Most teams overinvest in capability and underinvest in grounding and UX. The fastest wins often come from better retrieval, tighter prompts, and clearer affordances—not just a larger model.
For aspiring AI developers, start narrow and instrument deeply. Pick a workflow with painful status quo, ship a thin slice, measure correctness and confidence, and iterate with real users. For non-LLM companies, the mandate is different: augment your core product where AI reduces friction or unlocks frequency—don’t bolt on a chatbot because everyone else did.
For product leaders, AI changes the craft in two ways. First, prototyping is faster—use this to expand the option space early. Second, evaluation requires new muscles—build an experimentation and safety stack that blends qualitative red-teaming with quantitative reliability and cost controls. The leaders who thrive will combine taste with statistical rigor.
If you want to go deeper, these references are useful: Bard: https://bard.google.com/; ChatGPT: https://chat.openai.com/; Duet AI: https://cloud.google.com/duet-ai; Free courses on machine learning by Andrew Ng: https://www.andrewng.org/courses/; Google Assistant: https://assistant.google.com/; Introducing Google Assistant to Bard: https://blog.google/products/assistant/google-assistant-bard-generative-ai/; Large Language Model (LLM): https://en.wikipedia.org/wiki/Large_language_model; Meena: https://blog.research.google/2020/01/towards-conversational-agent-that-can.html.
In sum, the Bard blueprint reinforces a simple truth: ship with a thesis, learn in public with care, and let principled constraints accelerate—not slow—your path to product-market fit. That’s how we create value fast, build ethically, and stay ahead in the next era of AI.
I’ve learned that the fastest route to product-market fit blends ruthless focus, a tight-knit community of power users, and a clear-eyed understanding of founder psychology. As a VP of Product, I’ve seen how aligning strategy and self-awareness creates compounding advantages—especially when you commit to a vertical, build with your most advanced users, and make decisions faster than your market shifts.
Start narrow. Building for a specific customer forces clarity: one ICP, one core job to be done, one measurable outcome. That single-threaded focus removes ambiguity from product discovery, sharpens prioritization, and accelerates iteration. Only after unmistakable pull—retention, compounding usage, and customer-led expansion—do I widen the aperture to a broader customer base.
Clay is a lead-generation software that uses AI to scrape 50+ databases and help companies scale their outbound campaigns. When I evaluate products like this, I look for a crisp vertical wedge (for example, outbound sales teams or growth marketers), a clear “time-to-first-value” path, and strong affordances for advanced workflows. Winning a vertical creates a reliable beachhead for expansion without diluting the core value proposition.
Power users are the engine of product evolution. I actively identify and convene them—by analyzing power-law usage patterns, high-complexity workflows, and frequent integration touches—then invite them into hands-on feedback loops. I’ve found small, recurring sessions where we co-design in Figma and document patterns in Notion to be especially effective. These users don’t just validate; they reveal emergent use cases, inspire templates, and shape the roadmap. The result is a community that evangelizes organically and sets a high bar for everyone else.
Speed is a strategy. I front-load decisions with clear success criteria, kill-switch thresholds, and a cadence of two-to-four-week sprints. The discipline isn’t just “moving fast”—it’s committing to bounded bets, reducing work-in-progress, and measuring outcomes over output. Focus is often misunderstood as doing less; in practice, it’s doing the essential few things completely and letting the data make the hard calls. This mindset unlocks faster iteration cycles and cleaner OKRs that reflect real customer value.
My principles for product-market fit are simple but demanding: undeniable engagement (habitual use without prompts), concentrated love from a specific customer archetype, willingness to endure friction for core value, expansion motions that begin with usage not discounts, and a backlog shaped by customer pull rather than internal aspiration. When these signals converge, you don’t ask, “Do we have PMF?”—you ask, “How do we scale responsibly?”
Founder psychology shapes the product more than most admit. A company is the reflection of its founder’s personality, from appetite for risk to tolerance for ambiguity. I align my own psychology with the business through honest self-inventory, explicit constraints, and a cadence of reflection. I’ve found the life spiral framework helpful to contextualize growth phases, and techniques from Internal Family Systems to reduce reactive decision-making. The outcome is a calmer operating system that scales with the company rather than against it.
Translating this into a customer journey, I design for a sharp “aha” moment within minutes, a guided path to first successful workflow, and a clear unlock that turns a single task into a repeatable motion. From there, I build leverage: templates, automations, and integrations that accelerate outcomes for advanced users while remaining accessible to newcomers. This is how individual success becomes team adoption—and team adoption becomes the basis for expansion.
If you’re moving from first-time to second-time founder (or leading product through that evolution), the mindset shift is profound: less attachment to ideas, more attachment to evidence; fewer bets, bigger conviction; and a deeper respect for focus as an accelerant. Lean into vertical excellence, invest in your power users, and do the inner work. That’s how you achieve product-market fit—and keep it.
I study breakout platforms to refine how we build and scale product at HighLevel, and one story I keep returning to is how a modern dev tool can outpace entrenched competitors by reducing friction and amplifying distribution. Replit’s trajectory is a masterclass in both. As a product leader, I wanted to capture the strategic levers I see at work—and how any product creator can adapt them.
Replit is an online platform designed for collaborative coding in multiple programming languages. Replit boasts over 30m users, has secured $200M in venture funding, and was recently valued at $1.2B. These facts aren’t just impressive milestones; they’re signals of product-market fit compounding through sharp positioning, relentless iteration, and a distribution engine that turned usage into growth.
Here’s the core insight I take from Replit’s rise: the most durable advantage came from collapsing the distance between idea and software. By making it trivial to start, share, and iterate, the product converted curiosity into creation, and creation into distribution. In practice, that looks like zero-setup environments, multiplayer by default, and a UX that rewards shipping. When the platform itself becomes the marketing, you’ve found the secret lever.
AI is accelerating this shift. Integrating gen AI into the flow of work doesn’t just speed coding; it broadens who can build. I see this daily with product teams using AI for scaffolding prototypes, refactoring tricky edge cases, and translating intent (“what should this do?”) into working software. This is where the new “software creator” role emerges—part product thinker, part prompt engineer, part builder—unlocked by copilots and smart defaults rather than heavyweight toolchains.
For me, this reframes the strategy question from “How do we add features?” to “How do we lower activation energy?” The drivers of growth are then predictable: faster time-to-first-value, social proof embedded in artifacts users already share, and a distribution engine that compounds. Think of every project as a portable, runnable demo—content, onboarding, and virality in one.
There’s also a leadership lesson in the origin story: resilience and contrarian conviction often precede acceptance. The path included fundraising difficulties and multiple near-misses with Y Combinator—“Why YC almost rejected Replit four times” is a reminder that consensus is a lagging indicator. Credit where due, timely belief from people like Paul Graham can change the arc, but the throughline is persistence paired with user obsession.
On monetization, the strategy I favor—and see reflected here—is to let the free tier fuel creation and community, then monetize depth: private workspaces, performance, collaboration, compute, enterprise governance. In other words, price the power, not the curiosity. This aligns the business model to the distribution engine and avoids taxing the very behaviors that drive growth.
As AI reshapes engineering, I expect team topologies to evolve. I’m already deploying forward deployed engineers who sit with customers, use gen ai for product prototyping, and collapse feedback loops from weeks to hours. Combined with outcomes vs output OKRs, this makes room for velocity without sacrificing quality: ship thin slices, observe real behavior, let data and user value—not internal preferences—pull the roadmap forward.
If you lead product, here’s the playbook I’d reuse tomorrow: remove setup friction until “start” feels inevitable; turn every creation into content people naturally share; instrument for learning and iterate weekly; layer gen AI where it erases toil and unlocks new builders; and keep the monetization strategy aligned to usage intensity, not entry.
A few references that continue to shape my thinking—and that surfaced in this story—are worth bookmarking: 7 Powers: https://www.amazon.com/7-Powers-Foundations-Business-Strategy/dp/0998116319/; The Innovator’s Dilemma: https://www.amazon.com/Innovators-Dilemma-Technologies-Management-Innovation/dp/1633691780/; Mythical Man-Month: https://www.amazon.com/Mythical-Man-Month-Software-Engineering-Anniversary/dp/0201835959; On the Naturalness of Software: https://people.inf.ethz.ch/suz/publications/natural.pdf. For practitioners, I’d also keep an eye on OpenAI: https://openai.com/, Hacker News: https://news.ycombinator.com/, and ecosystems like Python: https://www.python.org/.
Summing it up: distribution is a product choice, not a marketing afterthought. When you design for creation, collaboration, and shareability from day one—and amplify with AI—you don’t just chase growth; you manufacture it. That’s the lever I’m pulling across my teams, and the mindset I’d recommend to every product creator aiming to build category-defining software.
Enterprise AI is exhilarating and unforgiving. I’ve seen gorgeous demos fall apart under real-world constraints and seemingly modest pilots unlock outsized value at scale. In this reflection, I share the playbooks I rely on to build, scale, and sell generative AI in the enterprise—what actually moves deals, secures product-market fit, and sustains trust with the C-suite and the front line.
Why is it so difficult to scale AI products for enterprise? Because the bar is higher on every dimension: data governance, security, extensibility, integration depth, reliability, and measurable ROI. An enterprise-grade, full-stack generative AI platform isn’t just a model; it’s the surrounding system—observability, evals, policy, workflow, and human-in-the-loop—that makes outcomes predictable, auditable, and safe. The fastest path to adoption is simple: deliver on-brand, on-policy content and decisions using a customer’s first-party data, and prove that quality holds up under load.
My north star is dependability over demo magic. The number one challenge is making model output dependable across messy, high-variance enterprise inputs. I build an evaluation harness early, with gold datasets, task-specific metrics, and human adjudication. Every change ships behind guardrails and is measured against cost, latency, and quality SLOs. When governance, change management, and procurement show up (they always do), I treat them like first-class product requirements, not hurdles.
Champions are the secret to winning complex accounts. I map the org, find operators who feel the pain daily, and quantify that pain in dollars and hours. Then I define success criteria upfront—time-to-value in under 30 days, measurable uplift (e.g., deflection, conversion, cycle time), and a plan for scale. I deploy forward deployed engineers alongside the business to co-design workflows, refine prompts and evaluators, and document before/after outcomes. Champions don’t just approve pilots; they co-author the business case and defend it.
To win the enterprise, trust architecture matters as much as model architecture. I lead with clear answers on data residency, encryption, SSO, RBAC, DLP, and retention policies; I address whether customer data trains models, default behaviors, and opt-in controls. I offer flexible deployment (VPC or private networking when needed), transparent pricing, and SLAs with real teeth. I also integrate where work already happens—CRM, help desk, knowledge bases—so value shows up in the flow of work.
Signs of healthy product-market fit are unmistakable: pull from lookalike buyers, multi-threaded expansions, champions who present results internally without me in the room, and usage that moves from experimentation to business-critical. I watch for weekly active usage above pilot thresholds, POCs converting to multi-year deals, and adjacent teams (Support, Marketing, Legal, RevOps) asking to onboard with minimal push. PMF feels less like persuasion and more like coordination.
Scaling large language models for specific use cases requires ruthless focus. I constrain scope to tightly defined workflows, pair retrieval with structured knowledge, and mix model strategies (base models, fine-tunes, tools, and function calling) based on cost and latency budgets. I codify policy-as-code and deploy guardrails at the orchestration layer, not just the model layer. Continuous evaluation—both automatic and human—is the heartbeat of quality.
My advice to AI founders in 2024 is pragmatic. Start with outcomes, not demos. Establish outcomes vs output OKRs that tie directly to revenue, cost, risk, or customer experience. Use gen AI for product prototyping to shorten discovery cycles, but graduate quickly to instrumented workflows in production. Align early with InfoSec and Legal; your speed will be gated by trust, not code. And when in doubt, ship smaller, safer increments faster.
Healthy co-founder relationships look the same across winning companies: clear domains, fast escalation, and a shared appetite for “disagree and commit.” I keep a decision log, time-box debates, and make moments-of-truth visible to the team and board. You’ll know it’s working when you have more energy after hard conversations than before.
The future of agentic AI is deeply enterprise: multi-agent workflows that plan, act, and verify with human oversight where it matters. The winners will combine reasoning, tool use, retrieval, and policy with audit trails that satisfy compliance while keeping velocity high. Think of it as re-engineering business processes around AI-native steps, not sprinkling AI on top of legacy workflows.
Culture turns strategy into reality. I anchor my teams on “connect, challenge, and own.” Connect means obsess over the customer problem and internal alignment. Challenge means we red-team our ideas, run experiments, and measure impact. Own means we accept outcomes, not just output, and we iterate until the business moves. This is how a customer support ai strategy becomes a durable moat, not a slide.
If you’re a product creator or product management leader, the above playbooks are meant to be lifted and adapted. Start where the pain is loudest, quantify the win, and let champions carry the story. The compound interest of disciplined product discovery, strong governance, and relentless evaluation is a generative AI business that sells itself—and scales.
