Roadmap-Less Agility for AI-Era Building
Drop the explicit roadmap and jump on capability drops the moment they land
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 4
- Confidence
- 82%
When your business is exposed to AI, rigid roadmaps become a liability because major new capabilities drop unpredictably and can reshape priorities overnight. Masad's answer is to run with no explicit roadmap, stay resilient to rapid change, and reallocate the whole team onto a new capability within days of it appearing. This pairs with a deliberately fluid, low-silo org where people slide along the designer-engineer-PM spectrum.
Origin
Amjad Masad's advice to founders and company leaders, drawn from how Replit reacted to anthropic's computer-use release.
Core principles
- 01Roadmaps are dangerous when anything AI could affect is in scope
- 02Being able to switch priorities right away beats committing to a fixed plan
- 03The way we work is going to change rapidly — build resilience to that change
- 04Fluid roles (designer↔engineer↔PM) let you redeploy people onto new capabilities fast
- 05Flexibility is uncomfortable for many people — you must build a culture and milieu that tolerates it
How to run it
- 1
Hold priorities loosely instead of a fixed roadmap
Replace a locked roadmap with a short list of directions you can drop instantly. Accept that for anything AI touches, a hard-committed plan will be wrong when the next capability lands.
Watch out Only override the roadmap for capability drops that genuinely affect your business — not every new release warrants a full priority switch.
- 2
Monitor for capability drops that hit your business
Watch for new model or platform capabilities (e.g. computer-use) and judge which ones materially affect your product versus which are noise.
- 3
Jump on the relevant ones immediately
When a capability that matters lands, slot it in and start building within days rather than queuing it behind existing plans. Replit began building and launching around computer-use almost as soon as it dropped.
Pro tip Speed of reaction is the edge — 'be able to jump on it really really quickly.'
- 4
Build a fluid, low-silo org that can absorb the pivot
Cultivate people who span the designer-engineer-PM spectrum so you can redeploy talent onto the new priority without reorg friction; send design-engineers to design crits and some designers to engineering meetings.
Pro tip Hire and grow hybrids — Replit has people who started as designers and became PMs or engineers.
Watch out This flexibility is 'uncomfortable for a lot of people'; you must deliberately build a culture that welcomes it rather than assuming people will adapt.
In the wild
When anthropic dropped its computer-use capability, Replit had no explicit roadmap slot for it. Rather than deferring it, the team immediately jumped on it and started building and launching features around it.
→ Replit shipped capabilities on top of computer-use quickly and kept building on it, staying ahead of a capability that could otherwise have blindsided a roadmap-bound competitor.
Common mistakes
Committing to an explicit roadmap in an AI-exposed business
A locked roadmap makes you slow to react when a major capability drops, so you either miss the opportunity or waste weeks building something the new capability made irrelevant.
Enforcing rigid role silos
Hard boundaries between designers, engineers, and PMs create low-bandwidth, easily-misinterpreted handoffs and prevent fast redeployment of people onto a new priority.
Is it for you?
Best for
Founders and leaders of small-to-mid AI-exposed companies who need to reprioritize on days' notice
Not ideal for
Large organizations or regulated environments where predictability, compliance, and long-lead commitments outweigh reaction speed
From the transcript
“one thing that I think is really difficult now is having road maps especially if you're doing anything in AI”
“we don't really have it explicit road map we we like immediately jumped on it and started building things”
“you just got you got to be fluid right because you know again when when designers can code and uh and Engineers can design”
“which is uncomfortable for a lot of people”
From the episode
Behind the product: Replit
Amjad Masad (co-founder and CEO)