What humans should focus on when AI does the building

Claude can write the code. It can't decide what's worth building, who to trust, or what the brand means. Here's the split I actually work by — for myself, and for teams I advise.

Talk it through → See my resume ↗

02 Where humans stay essential

The hard parts were never about “building”

Knowing what to build, getting people to care, navigating organizations, earning trust, deciding when data is ambiguous. Claude replaces none of that.

01
The hard parts were never about "building"
Knowing what to build, getting people to care, navigating organizations, earning trust, making decisions when data is ambiguous. Claude replaces none of that.
02
Distribution and relationships
You can vibe-code a product in a weekend; convincing 1,000 paying customers to change their workflow is still a 2-year job. The bottleneck has shifted from "can we build it" to "can we get anyone to use it."
03
Taste and judgment
When everyone can generate, curation becomes the scarce resource. Someone still decides what's worth building, what UX feels right, what the brand means. Bad taste at that level kills good execution.
04
Coordination and trust
Companies don't buy software, they buy confidence it won't blow up. Compliance, security, onboarding, support, legal — all still very human.
05
If work = only receiving requests, you get replaced
Pure order-taking (do X, wait, do Y) is exactly what AI automates. Value now = shaping the request itself: spotting the problem before it's asked, pushing back, deciding what's worth doing at all.
06
Creativity
Coming up with the idea worth building in the first place, not just executing on a brief. AI recombines; a human still has to point it at something new.
07
Leadership
Setting direction, owning the call when it's political or personal, and being the person others trust to decide when the answer isn't in the data.

03 Where AI should take over

Own the boring parts, escalate the sharp ones

Execution speed is AI's whole advantage. Don't gate it behind human hands once direction is set.

01
Build
The actual code, drafts, designs, scripts. Execution speed is AI's whole advantage; don't gate it behind human hands once direction is set.
02
Hold and surface memory
Remember decisions, context, past attempts, so humans don't re-litigate settled questions or repeat failed approaches.
03
First-pass everything
First draft, first analysis, first option set. Humans edit and decide; AI shouldn't wait to be told to start.
04
Grunt-level research and triage
Scan job boards, read logs, check prices, monitor feeds. Anything repetitive and well-specified belongs to AI by default.
05
Flag, don't decide, on ambiguity
When data is unclear or a call is political/relational, surface the tradeoff and stop. Don't fake confidence to avoid asking.
06
Own the boring 80%, escalate the sharp 20%
Routine execution stays with AI; anything touching trust, money, or relationships routes to the human, explicitly, not silently absorbed.
07
Follow-ups
Chasing the loose thread, the unanswered message, the reminder nobody sent — so nothing quietly drops between people.
08
Reporting
Turning what happened into a clean summary, on schedule, without being chased for it.

Humans decide what's worth doing and why.
AI does the doing, and remembers what happened.

"A small team that operates much larger."

Building AI systems for complex product companies.

Work with me → More about how I build ↗

Malmö, Sweden — Product Owner turned AI operator. See the full resume.