Every product team I talk to is using AI to move faster. Fewer are asking whether faster is actually the advantage they think it is. A study published in Nature found that scientists using AI produce 26% more citations — but their work converges toward a common position. Less exploration. More median. I’m seeing the same thing in product teams: the strategy decks sound alike, the copy sounds alike, the research synthesis sounds alike. AI is compressing the gap between your best thinking and everyone else’s. That’s not a productivity gain. That’s a moat collapsing.
The teams I’m watching win right now aren’t the fastest. They’re the ones using AI to explore more possibilities — not fewer. They’re designing AI features that expand the solution space instead of collapsing it to a single “best” answer. They’re rewarding originality of thought, not volume of output. And they’re building verification and trust into everything, because in a world where AI makes it trivially easy to look legitimate, the only durable advantage is actually being legitimate.
This edition gives you the research, the frameworks, and the stories to build that way. We spoke with the Artificiality Institute about cognitive sovereignty — the science of keeping your team’s thinking original when the tools push toward convergence. We break down why the “one-person billion-dollar company” narrative collapsed and what that means for every product competing against AI-enabled imposters. And we cover the peer influence research, the agentic AI architecture gap, and the product design failures that explain why 85% of your workforce still can’t get value from these tools.
What You’ll Learn in This Edition
Why the median pull is the most underrated risk in AI product strategy — and what to do about it
Helen and Dave Edwards on cognitive sovereignty: the framework for keeping your team’s thinking original when the tools push toward convergence
The one-person billion-dollar company is fraud — and what the Medvi collapse reveals about AI-enabled deception you’re already competing against
Why peer influence predicts AI adoption better than any training program — and why most rollout strategies ignore the one lever that works
Running your life from a terminal is an indictment of enterprise AI product design, not a flex
🎙 Episode 5: Cognitive Sovereignty — with Helen and Dave Edwards
We sat down with Helen and Dave Edwards of the Artificiality Institute — a decade of research into what AI does to human cognition, identity, and professional judgment. Not what it can do for us. What it does to us.
Their core concept — cognitive sovereignty, the ability to remain the author of your own thinking — reframes the entire AI adoption conversation. Most product teams are asking “how do we get people to use AI more?” The better question: “how do we make sure using AI makes them better, not just faster?”
Dave asked the question that should sit with every founder: “If AI can replace the humans in your business, does your business have any value at all? The value of that business is about the people and the decisions those people make.” The corollary for product builders: if your AI feature makes every user’s output identical, you’ve built a commodity machine, not a competitive advantage.
Helen’s research on the median pull has immediate design implications. When AI users converge — when everyone’s emails sound the same, when every strategy deck follows the same structure, when every analysis reaches the same conclusion — the organization loses its ability to see around corners. The teams that will outperform are the ones that use AI to explore more possibilities, not fewer. That means designing AI features that expand the solution space rather than collapsing it to a single “best” answer.
“We need to push for people to think about using these tools differently — to create new capacities that we couldn’t have before. The broader narrative is about replacing humans rather than adding to us.” — Dave Edwards
Listen now: Spotify | Apple Podcasts | YouTube
📅 The Artificiality Summit — October 22-24, 2026 in Bend, Oregon. Theme: “Unknowing.” This is where neuroscientists, cognitive scientists, philosophers, and product leaders figure out human-AI coevolution — not on a vendor stage, but in the room together. If your AI strategy depends on understanding how people actually think with these tools, this is the event. Register here.
The One-Person Billion-Dollar Company Is a Fraud — and That’s the Point
The New York Times profiled Medvi — a solo-founder telehealth GLP-1 provider projecting $1.8 billion in revenue, built with AI tools and $20,000 — as proof that Sam Altman’s prediction had arrived.
Then Gary Marcus investigated. FDA warning letter for misbranding drugs. Class-action lawsuit over spam with falsified headers. Unverified revenue claims. The company didn’t succeed because of AI — it found a pricing arbitrage and used AI to automate the parts a legitimate company would staff with humans who exercise judgment.
But the strategic lesson isn’t “Medvi is bad.” It’s that AI has collapsed the barrier to looking legitimate. Vibe-code a polished site in an afternoon. Generate thousands of personalized outreach emails. Automate customer service without human review. The products you’re building now compete against companies that look real, feel real, and may be entirely smoke. If you’re not building verification and trust signals into your product from day one, your users can’t tell the difference between you and the next Medvi. Trust isn’t a feature. It’s the product.
Brittany framed it on the podcast: “Celebrating the one-person billion-dollar company is the penultimate version of dehumanizing work — cannibalizing your own long-term value.” Dave Edwards made the math explicit: “If thousands of companies are run by single people with agent fleets, what are the rest of us doing? The math doesn’t work at the economy level.”
Featured Resource: Running Your Life From Terminal Exposes an Industry Failure
People narrating their entire day to Claude Code terminals. Building personal operating systems with Obsidian vaults and markdown task files. The examples are genuinely impressive — and they reveal that enterprise AI product design has failed the majority of users. If the most productive way to use AI requires you to architect a complex personal system, write precise natural language workflows, and become your own systems integrator, the on-ramps are built for enthusiasts, not for the 85% of the workforce who actually need to adopt these tools. Product builders should read this as a design brief, not a trend piece. The gap between power users and everyone else is your product opportunity.
Read the full article: Running your life from terminal is peak 2026 — and that’s not the flex you think it is
$490 billion in enterprise AI spending is delivering nothing. That’s not a technology failure. It’s a value creation failure. AI Value Acceleration exists to close that gap — diagnosing where AI value stalls and building playbooks that actually work. Value Assessment in 3 weeks. Value Amplification to go deep. Value Acceleration to prove what works. aivalueacceleration.com
Product Impact Resources
Every resource this period points to the same conclusion: the organizations pulling ahead aren’t adopting AI faster — they’re investing in the layers that make adoption stick. Architecture. Culture. Verification. The teams skipping these layers are building on sand.
Peer influence beats training for AI adoption — and most organizations are ignoring it. When AI learning stays private, adoption stalls. When people see their peers using AI visibly and successfully, adoption compounds. The implication for product rollouts: stop investing in training decks and start investing in visibility mechanisms — shared AI workflows, public experiments, team-level adoption stories. Peer Influence Can Make or Break Your AI Rollout
40% of agentic AI projects will be canceled by 2027 — and the problem isn’t agents. It’s the architecture underneath: data navigation, governance layers, orchestration, and human interface design. Deploy agents without these four layers and the first exception destroys trust in the entire program. Product teams building agent features need to design the guardrails before the capability. Gartner: Agentic AI Deployments Failing
AI is exposing which designers were designing and which were decorating. The designers who applied surface patterns without understanding constraints are being replaced. The designers who navigate ambiguity, make judgment calls, and hold complexity — the ones who were actually designing — are more valuable than ever. If you’re a design leader, this is your hiring filter. Why AI Is Exposing Design’s Craft Crisis
Domain expertise is the moat, not model performance. Wolters Kluwer grounds agents in proprietary knowledge graphs and charges for third-party queries via MCP. They’re not competing on intelligence. They’re competing on trust and domain depth. If you’re sitting on domain-specific data and haven’t built an AI access layer for it, someone else will. Wolters Kluwer’s “System of Action” Strategy
Product Impact News
The pattern across every headline this period: the distance between AI narrative and AI reality is becoming measurable — in lawsuits, in reorgs, in wasted capital.
HSBC appointed its first Chief AI Officer. David Rice, starting April 1 with a mandate to expand generative AI across the group. When a $3 trillion bank creates a dedicated CAIO, the signal is clear: AI value accountability has become a C-suite function, not an IT initiative. If your organization hasn’t named someone accountable for AI outcomes, you’re behind the curve.
Crypto.com spent $70M on AI.com, then fired 12% of its workforce. Framed the cuts as eliminating roles that “do not adapt in our new world.” When the narrative outruns execution by this much, it poisons AI adoption internally — every remaining employee now associates AI with job loss, not capability gain. AI-Washing Has Consequences
Cove AI built an AI collaboration platform. Microsoft swallowed it whole. Team acquired, product shut down, absorbed into Copilot. If you’re building adjacent to a platform company’s roadmap, your exit isn’t an IPO — it’s an acqui-hire that kills your product. Build where platform gravity can’t reach, or build something the platform can’t absorb. Platform Gravity
GPT-5.2 is tiered. Your CIO wants GPUs back on-premises. Three tiers — Instant, Thinking, Pro — plus MCP enterprise connectors. But the real story: CIOs are pulling compute back in-house for data sovereignty, favoring open-weights models over cloud APIs. Product teams building on frontier cloud APIs should be planning for the scenario where your enterprise customers won’t send their data to them. The Sovereignty Shift
Key Takeaways
The throughline across every signal this period: AI is a divergence machine. Used well, it amplifies what makes your team and your product distinctive. Used carelessly, it erases those distinctions and leaves you competing on speed alone — which is a race everyone loses eventually.
Design for exploration, not convergence. If your AI features collapse every user’s output to a single “best” answer, you’re training your team to stop thinking. The organizations that will win are the ones designing AI that expands the solution space — more options, more perspectives, more adjacent possibilities. Helen Edwards’s median pull isn’t theoretical. It’s measurable. And it’s happening in your Slack channels right now.
Trust is the product, not the feature. When the barrier to looking legitimate collapses — and it has — the only companies that survive long-term are the ones that build verification, provenance, and human judgment into every layer. The Medvi story isn’t an anomaly. It’s the new competitive environment.
Architecture before agents. 40% cancellation rate by 2027. The pattern is clear: teams that deploy agent capability before building data navigation, governance, orchestration, and human interface layers don’t just fail — they destroy organizational trust in AI for years afterward. Ship the guardrails first.
Check Out Recent Episodes
Episode 4: The Era of Agents — Your Cognition Is the Product Now — Three years of AI evolution mapped to three eras of risk. Era one gave us wrong answers. Era two gave us wrong context. Era three — agents — is giving us wrong actions. Plus four startups winning the point-solution war.
Episode 3: Context Is the New Moat — Juan Sequeda, Principal Researcher at ServiceNow, on why RAG was always a workaround for a deeper problem: your AI doesn’t understand your business.
Episode 2: Vibe Coding Changed Everything — Yoni Jozwiak, founder of Base44, on the defensibility crisis facing every AI startup when anyone can build software by describing it.
AI Strategy Jobs
Lead UX Designer, Agentic AI Platform — Deloitte (Multiple US locations — Hybrid)
Product Designer, AI Models — Figma (SF or NYC)
Generative AI Design Expert — Nike (Beaverton, OR )
Product Designer — Luma (Palo Alto, CA —Hybrid)
Product Manager, AI Models — Descript (Remote USA)
Your AI product demos well but can’t stick, scale, or justify cost? PH1 has spent 14 years helping product teams prove impact — from measuring what AI products actually deliver to improving the performance of LLM-powered experiences to defining AI vision that survives contact with real users. If the median pull research made you rethink what your AI feature is actually doing to your users’ thinking, that’s the right reaction. Let’s talk about it.
Thank You for Supporting the Product Impact Podcast
This newsletter exists because you keep showing up, sharing what resonates, and pushing back when we get it wrong. If this edition changed how you think about what AI is doing to your team’s originality — forward it to someone building with AI who needs to hear it. Browse all episodes at productimpactpod.com.
