1
Outcomes over output
Shipped features don't count. Moved metrics do. Every piece of work needs a number attached before it starts.
How I work
Tested on clients. Refined over 9 years.
Early stage. Enterprise. Every engagement.
9 years of client work across early stage and enterprise built this system. It's not a methodology. It's just what works — clarity, ruthless prioritisation, calm delivery.
How I think
Before I do anything.
2
Decisions that stick
Most chaos comes from re-litigating the same decisions. Trade-offs are made explicit, logged, and don't get reopened.
3
Systems over heroics
I'm not the hero. The system is. Delivery becomes predictable without adding heavy process or more meetings.
The operating system
Six steps. Every engagement.
Applied to every client. Without exception.
1
Align on goal + success metrics
One shared target. Everything else flows from that. No metric, no work.
2
Set decision owners + a decision log
Make ownership explicit so decisions are fast and traceable. No more relitigating last week's call.
3
Ruthless scope — what we do / what we don't
Cut what doesn't need to ship yet. Protect what does. Scope is never about what's cool.
4
Backlog quality — AC, edge cases, QA steps
No ambiguous tickets. Engineers ship with confidence. Done means done.
5
Regular shipping cadence + release readiness
Predictable rhythm. Checks in place. No release surprises.
6
Measure → learn → iterate
Every release is a learning loop not a finish line. What moved? What didn't? Ship again.
"Shipped it. Metrics said no. Shipped something else. That's the job."
How AI fits in
AI across the full product lifecycle.
Not hype. Just proof.
I use AI across the full product lifecycle. Research, discovery synthesis, spec writing, build loops, iteration planning. Not as a gimmick — as a way to compress timelines from months to weeks at every stage.
Discovery
User research synthesis, pattern finding, hypothesis generation. Faster signal, less noise.
Spec writing
PRD-lites, acceptance criteria, edge cases. What used to take days takes hours.
Build loops
Used across every engagement. I also build my own products with AI — no agency, no bloated team.
Iteration
Faster feedback loops. Faster decisions. Faster shipping. That's the whole point.
9 years of client work. Now applied to my own products too. All of it shipped with AI.
Week to week
Lock priorities. Ship weekly. No surprises.
1
Plan
Lock priorities + success signals. Confirm what good looks like before anything starts.
2
Build
Async check-in on risks and blocks. Scope stays stable. No mid-sprint priority changes.
3
Release readiness
QA confidence. Edge cases covered. Final scope check. No last-minute surprises.
4
Ship + learn
What shipped / what's next / what's blocked. Track outcomes. Iterate. Repeat.
Day to day
Async-first. Direct. No theatre.
Async-first by default
Written over verbal. Decisions don't wait for a meeting to happen.
1–2 calls per week max
Planning and decisions only. No check-in calls. No status updates that should be a doc.
Written updates every week
Shipped / next / blocked. Always. No exceptions. No surprises.
Your tools, not mine
Jira, Linear, ClickUp, Notion. I work in whatever you use. No onboarding overhead.
Builder mindset
I don't just advise on building products.
I build them.
9 years of client work built the system. Building my own products — Heat, Miguelito, whatever ships next — pressure-tests it. Nothing sharpens your product instincts faster than having skin in the game. 3,000 followers before a single line of code shipped. Built with AI. Documented publicly.
3,000+ followersShipping April 2026Built with AI
See Heat →Tell me what you're building.
15 minutes. No slides.
Book a 15-min call →Abdi Bedel · AI-Native Product Lead · London · Remote · 2026