When the codebase is the bottleneck - not the team.
Good engineers, doing their honest best, inside an architecture that quietly fights them every week. That’s usually the actual problem.
Releases got slow.
Six months ago you shipped daily. Now it’s a fortnight, three Slack channels, two reverts and a quiet promise to refactor next quarter.
One person knows everything.
The same name appears in every PR, every incident, every onboarding doc. They’re a heroic single point of failure, and they’re tired.
The team ships, the AI shelf gathers dust.
Cursor licences paid, Claude subscriptions billed, nobody actually changed how they work. Seniors still write boilerplate by hand. Juniors paste prompts into a chat window.
The roadmap is a wish list.
No estimates anyone trusts. No way to say no. Engineering becomes a feature factory and morale quietly leaks.
You inherited the build.
The original team left. The docs are old. You’re a non-technical founder making technical bets on hope.
A CTO function, in cycles.
Eight tracks of work. Pick the one or two that matter this quarter. We run them with you; your team grows into them.
Architecture review & roadmap
Two weeks deep into your code, infra and decisions. You walk out with a 12-month architectural plan and the three changes worth making first.
AI & automation implementation
Where in the company AI actually pays off — ops, sales, support, finance. We pick the workflows, build the agents, plug them into your stack, and train the humans next to them.
Delivery practices & SOPs
Sprint cadence, definition of done, code review, incident process, on-call rota. Boring rituals, ruthless about keeping them.
AI adoption for engineering
Pick the tools, write the prompts, build the eval harness, run the rollout. Your team ships measurably faster — and you can prove it on the board deck.
Engineering metrics
DORA + the four metrics that matter for your product. Honest measurement, not vanity dashboards.
Tech due diligence (sell-side)
You’re raising or being acquired. We write the architecture deck, answer the DD questionnaire in your voice, and sit on the calls so you don’t have to bluff a stack diagram on a whiteboard.
Platform & cost engineering
Cloud bill audit, infra simplification, build vs. buy calls. Often the second-largest line item on your P&L — treat it like one.
Engineering brand & docs
The README that makes seniors apply. Engineering blog. Public RFC process. The signals that compound your hiring.
Twelve weeks. A running team.
Four phases, four artefacts. No theatre — each one is something your team is using the day we hand it over.
An honest map — before anyone touches the code.
1:1s with every engineer. A codebase walk-through. A risk inventory. Friday of week 2: the baseline doc lands in your shared drive.
Most CTOs prescribe before they’ve listened.
Pick the layer you need now.
Sequence them. Pause them. We hand back a stronger team each cycle.
Compass
Six-week strategic engagement. The architecture & team plan you should have written six months ago.
- Architecture & risk review
- 12-month tech roadmap
- Three written ADRs
- Investor-ready tech narrative
Embedded CTO
Ongoing. Two days per week, embedded with your founders & engineering leads.
- Everything in Compass
- Weekly leadership cadence
- Delivery + on-call practices
- Engineering metrics dashboard
- Board / investor tech updates
AI Adoption Sprint
Six weeks. Your engineering team stops paying for AI tools they don’t use — and starts shipping measurably faster.
- Tooling audit & consolidation
- Internal prompt library & review agents
- Eval harness — quality, not just speed
- Rollout plan: senior + junior tracks
- Weekly office hours with the team
- Hand-off to an internal AI lead
What changes in ninety days.
Honest, measurable, durable. Numbers our last four engagements actually moved.
You asked. We answered.
When it’s the highest-leverage thing in the room, yes — usually proofs-of-concept, infra spikes, the first ADR-worthy refactor. We’re not your hands-on senior engineer; we coach the ones you have and hire.
Walk us through your README.
Half an hour, founder-to-founder. We’ll point at the three risks the team has stopped seeing — whether or not we end up working together.