// fractional AI CTO · 12 years in business · powered by agent swarms

the business half

I understand the business.

Twelve years on the operating side. I know which features move revenue and which are vanity — so I kill the wrong AI bet before you fund it.

revenue sensepricinggtmhiring

the proof half

I architect the AI.

I make the senior AI calls — orchestration, MCP, evals, guardrails — then the swarm turns the idea into a working proof: a real POC in hours, not a sprint.

mcpagent orchestrationevalsguardrails

↳ POC in hours · working proof in a week · production in weeks

Not a dev army — a fractional CTO whose swarm ships the proof.

book a call see the work

typescript · bun · next.js · supabase · fly.io · mcp · turborepo · playwright · infisical · cloudflare · tailwind v4 · postgres · docker · zod ·

plainspoken

I’ll tell you what’s worth doing — and what isn’t.

Bluntly. If an idea won’t fly, you’ll hear it from me before you fund it — not at the demo.

[ 01 / the operating model ]

One operating model. Both sides of the table.

The same discipline on every engagement — whether it’s a two-week proof or a year as your fractional CTO. Nothing improvised, nothing handed to a junior.

strategy

Anchor every engagement to a business outcome

proof

Cut the work across a swarm of agents, in parallel

strategy

Kill vanity features before they cost you

proof

Ship it typed, tested, and observable in production

strategy

Own the roadmap and the hard trade-offs

proof

Own the merge gates — evals, checks, review where it counts

strategy

Set the security and compliance bar

proof

Scoped tools, least-privilege, every action audited

strategy

Stay on as fractional CTO / technical lead

proof

Hand over clean — runbooks, docs, no lock-in

strategy — what I decide

proof — what the swarm ships

Anchor every engagement to a business outcome

proof ·Cut the work across a swarm of agents, in parallel

Kill vanity features before they cost you

proof ·Ship it typed, tested, and observable in production

Own the roadmap and the hard trade-offs

proof ·Own the merge gates — evals, checks, review where it counts

Set the security and compliance bar

proof ·Scoped tools, least-privilege, every action audited

Stay on as fractional CTO / technical lead

proof ·Hand over clean — runbooks, docs, no lock-in

// fixed scope · fixed price · the governance I run in MCP Beast — policy, audited tools, no lock-in

why this exists

The demo is the easy part. Orchestrating agents to ship the rest is the job.

I watched no-code and AI demos die at the same wall — eighty percent done, never shipped. So I built the orchestration layer that gets them across: AppHandoff conducts a swarm of agents — contract-aware, gated, shipped as production code.

That’s the craft — not prompts, not prototypes. Agent orchestration, the CI fleet that keeps it honest, and the business judgment to aim it.

→ apphandoff.com

[ 02 / selected work ]

Things in production. Proof of how I work — yours ships the same way.

all projects →

lovable · diamond-tier expert

[ 03 / the studio ]

More than one person. Fewer than an agency.

One principal owns your build end to end. A swarm does the volume. When a build needs a specific human, I bring in operators I trust — by name, never a faceless bench.

Ralph

principal · accountable

Owns the spine. Makes the business calls. Sets the gates, owns the risky calls. One throat to choke.

The swarm

agents · on tap

Up to thirty coding agents cut tickets in parallel — coordinated by AppHandoff, tooling I built. Evals and CI gate every merge; I review what’s risky.

+ops

Trusted operators

a short bench

Named specialists I bring in when a build needs a specific human. Vetted, not outsourced.

// amsterdam → atlanta · one operator answers, the studio delivers

business receipts

12y

on the business side

wks

to prod, not quarters

proof receipts

90s

CI per push · was 6–9 min

products I run in production

Bring the idea. Leave with a working proof.

book a call

I understand the business.

I architect the AI.

One operating model. Both sides of the table.

The demo is the easy part. Orchestrating agents to ship the rest is the job.

Things in production. Proof of how I work — yours ships the same way.

Lovable prototypes, finished in production

One endpoint for every MCP

Bug reports in one script tag

CI from nine minutes to ninety seconds

A membership platform, end to end

More than one person. Fewer than an agency.

Bring the idea. Leave with a working proof.