We’re in the biggest behavior-change moment in tech since the 1990s. Back then, everyone had to learn how to type, use computers, and rewire how work got done. Today, we’re doing that again but this time at AI speed.
When markets rewire this fast, product-market fit doesn’t just evolve. It renegotiates itself weekly.
In my conversation with Spence Green (CEO/Co-founder of Lilt) on Inside the Silicon Mind, one thing became clear:
PMF in 2025 is a wartime game and agents are the new battlefield.
Here’s what stood out.
3 Insights
1) PMF now depends on leaders modeling new behavior
Spence said something I’ve now heard echoed across Silicon Valley:
“Leadership is more important than ever but it has to be player-coach leadership.”
We’re not just shipping products. We’re asking humans to change habits.
Habit change is hard. People don’t adopt new workflows because of strategy decks. They adopt them because they watch leaders use them first.
If your product requires behavior change, adoption becomes a leadership problem before it becomes a product problem.
2) Speed is a product feature in frenzy markets
Spence described this era as a “speculative frenzy” like the early oil rushes.
“If you’re on vacation when they strike oil, you lose.”
That’s the moment we’re in.
The velocity of change means consensus becomes dangerous. Wartime markets reward direction, tempo, and iteration, not deliberation theater.
In 2025, speed isn’t just execution. It’s defensibility.
3) Agents aren’t a feature - they’re a new architecture for PMF
This is the deepest product shift in the conversation.
Spence made the point that enterprise problems aren’t transactional - they’re workflow-based.
Translation isn’t “run model → done.” It’s a coordination layer across teams, approvals, and publication systems.
Agents flip the stack:
● Old world: fixed UI + hardcoded business logic
● New world: tools + agent orchestration + dynamic workflow
The product becomes a toolbox, and the agent becomes the workflow engine.
That’s why this isn’t a hype cycle. It’s a complete rewiring of how PMF gets built in enterprise software.
3 Plays (for founders/operators)
Play 1: Lead adoption from the front
If AI changes workflows, your job is to show usage before asking for it.
Ask yourself weekly:
● What new behavior am I modeling publicly?
● Where am I delegating what I should be demonstrating?
Founders who stay close to the product win adoption wars. Founders who drift up into “air-traffic controller leadership” lose them.
Play 2: Separate one-way doors from two-way doors
Most teams waste time debating reversible decisions and rush irreversible ones.
Two-way doors → ship fast, adjust fast.
One-way doors → slow down, memo it, commit.
One-way doors are the decisions that define your trajectory:
● business model shifts
● core architecture rewrites
● government/regulatory entry
● exec hires that reshape culture
Your PMF story is basically the sum of those few doors.
Play 3: Build agent-ready primitives, not bloated workflows
If you’re feeling roadmap bloat, agents are your escape hatch.
Stop building bespoke logic for every new segment.
Instead:
build durable tools
expose them cleanly
let the agent compose workflows per use case
This is how you scale PMF without turning your product into a feature factory.
Closing Reflection
This conversation reframed something for me:
The next decade of PMF isn’t about which model you use. It’s about which workflows you replace and how quickly humans trust you to replace them.
Agents make software adaptable. But only leadership makes adoption inevitable.
We’re in a race right now. Not for the best demo but for the fastest path to trust.
That’s where PMF lives in 2025.
Until next time,
Firas Sozan
Your Cloud, Data & AI Search & Venture Partner
Find me on Linkedin: https://www.linkedin.com/in/firassozan/
Personal website: https://firassozan.com/
Company website: https://www.harrisonclarke.com/
Venture capital fund: https://harrisonclarkeventures.com/
‘Inside the Silicon Mind’ podcast: https://insidethesiliconmind.com/