3 skills compound when ai does the constrained work
3 skills compound when you build solo with agents: negotiation, problem solving, leadership. mckinsey just named them as the three that get MORE valuable, not less. solo builders are doing all three already, mostly without naming it.
negotiation as a solo dev means brokering with customers, with your future self, with whatever past-you committed to in v0.1. agents propose options. you broker between them. that's the work.
problem solving splits in two. agents are extraordinary at constrained problems. they're still poor at the unconstrained kind. for solo builders, EVERY problem starts unconstrained because nobody else framed it for you. the skill is shaping it until you can hand it to an agent and trust the output.
leadership is workflow design. who proposes, who decides, what the agent owns, what you own. solo builders write this exact document for themselves every week. most teams have never written it on paper.
3 exercises i'm doing this quarter: one cross-org negotiation rep with someone who isn't a customer (legal, partner, accountant). re-frame three currently mis-scoped projects until i can hand each framing to an agent. write the human-AI workflow doc for the most expensive project. done by end of june.
which of the three are you most underinvesting in right now?
Mykola, your insight on solo building as a "leadership and workflow design" exercise is exactly why the technical barrier is no longer the bottleneck—the bottleneck is now the clarity of the frame you provide to the agent. By moving from "builder" to "broker" between AI options and customer needs, you’re highlighting the exact shift toward high-bandwidth leadership that McKinsey is betting on.
I’m currently running Tokyo Lore, a project that highlights high-utility logic and validation-focused tools—specifically those that leverage the exact human-AI workflow you’ve outlined. Since you’re building the definitive case for how solo entrepreneurs compound their value through agentic leverage, entering your project could be the perfect way to turn this strategic framing into a winning case study while your odds are at their absolute peak.
I've reused the same eval spec across 4 different agents - writing it once was the real work. criteria don't expire just because the task did.
yeah the broker framing lands well. i've noticed the same thing - my agents don't fail on model quality, they fail when i'm vague about success criteria. 'improve the sprint plan' gets garbage output, 'flag stories with missing acceptance criteria' gets something i can actually use.
Exactly — that shift from “generate” to “evaluate against clear criteria” is where things start working.
You’re basically turning the agent into a system with boundaries, not a guesser.
The interesting part is once you define success clearly enough, you can:
→ reuse it
→ delegate it
→ and even measure it
That’s where it compounds.
Curious — are you starting to standardize these as reusable checks, or still defining them per task?
Also, this “broker + success criteria” framing would actually make a strong case study.
I’m running a small round (Tokyo Lore) where we surface ideas like this with a focused group of builders — could be a good fit while it’s still early.
Happy to share more if you’re open 👍