Hai. hi + AI — Appaya Ltd
Appaya is the London studio of Luke Czak — fractional CPTO and hands-on engineer, previously Global Head of Technical Product Management at HSBC, and a product leader at Zopa, Smarkets and Shieldpay. A fixed-fee audit of what you have built, then ongoing technical judgement once you know.
Appaya is an AI-native studio with three productised offers, not a generic agency menu. I direct a fleet of AI engines that build in parallel, and every change is proven before it reaches you.
Orchestrates Claude · Codex · Gemini · + whichever engine earns the job
01 — How I work with you
Most clients start with a decision. Some stay there. Others move to ongoing leadership, then to delivery — start wherever fits today.
You built it with AI and it works. Find out what is actually underneath it before real users arrive.
Timing Delivered in 5 working days.
You leave knowing exactly where you stand
Good fit when you have shipped something with Claude Code, Lovable or Cursor and need an honest read on it.
See the AuditOngoing senior technical leadership — technical direction, architecture, hiring, and delivery discipline, without a full-time hire.
Rhythm Days that suit the work — and reviewed openly as it goes.
The seat filled by someone who has held it
Good fit when you have engineers but no senior technical head accountable for the calls.
See FractionalProduction software built by AI agent fleets I direct, shipped with real cross-engine review rigor.
Timing Set per engagement, once scope is locked.
The thing built, working, and yours
Good fit when you have one bounded outcome to ship and are out of engineering capacity.
Start a build02 — Data & confidentiality
The public enquiry below is not the place for anything sensitive — here's the boundary, plainly.
Business context, the problem you're deciding or auditing, and non-sensitive documents. Access, where an engagement needs it, is scoped read-only by default.
Credentials, production data, payment details, special-category data, or confidential source archives — not in the public enquiry, and not before a signed data schedule is in place.
Work runs through the AI providers and access controls agreed in the engagement's data schedule. Working data is deleted on a stated schedule after closeout, subject to statutory financial-record requirements.
An NDA or DPA is available before I handle anything sensitive — just ask, before anything changes hands.
03 — Process
However we work together, the shape is the same: agree what good looks like, then prove it — you see the work as it happens, not at the end.
Tell me the outcome you need. I come back within a working day with an honest read on whether I am the right person for it — and I will say so if I am not.
What you'll actually have at the end, written down plainly — so there is never a gap between what you pictured and what arrives.
No discovery theatre and no onboarding month. The first real look at your code happens in the first days, not the third week.
You see the work as it happens, not a reveal at the end. Anything serious reaches you the day I find it.
You read it, push back on anything that doesn't ring true, and I fix it. Then it is yours to keep, use, or hand to anyone.
A numbered balance invoice goes out. The final report, code, or assets hand over by bank transfer against that invoice, once funds clear — stated plainly, never on this site.
04 — Proof
No invented clients, no logos I have not earned, no metrics I cannot show you. Everything below is Appaya's own — built to make an agent fleet trustworthy, and the products that came out the other end.
Anyone can say they build with AI. The question worth asking is what they built to make it safe to build with AI — because the speed is free and the rigour is not.
Routes every task to the right engine — Claude, Codex, Gemini, Copilot — instead of hardcoding one vendor. It is what makes cross-engine review possible: the builder and the reviewer are never the same model.
The project knowledge layer: acceptance criteria, proof requirements, and the record of what was actually verified. A task closes on evidence, not on an agent reporting success.
Per-product analytics pulled from every source into one place, so a decision about what to build next is made against numbers rather than a hunch about which thing feels popular.
The deployment brain: one path to production per platform, credentials resolved from a vault, and a health check after every deploy. This site ships through it.
A workout tracker built natively for Apple and Google, with offline-safe logging that never loses a set mid-session and syncs when you are back online. Native on both platforms, approaching launch.
The operating system for a one-or-two person AI company — running a portfolio of products with agents that actually finish things. Built by exactly the operation it is built for.
05 — How I work
Every change passes an adversarial cross-engine review before it ever reaches you. That's how I guarantee the quality.
The rigour isn't a promise on this page — it's a gate in the pipeline, and the proof ships with the work.
Work lands in the open as it's made — you're never waiting until the end to find out what you got.
A different engine reviews every change before it reaches you — the builder never marks its own homework.
Anything that matters reaches you the day I find it — never saved up for a report or an invoice.
06 — Questions
Start with the audit unless you already know exactly what needs building. You have shipped something with AI assistance and want an honest read on it → AI Codebase Audit. You need ongoing senior technical judgement, not a one-off → fractional leadership. You know exactly what you want built → tell me the outcome.
With a conversation, not a quote. Tell me what you're trying to find out or get built; I'll tell you honestly whether I'm the right person, and what I think the real shape of the work is. If we both want to carry on from there, we agree what good looks like and I put it in writing. Nothing is metered by the hour, and nothing arrives as a surprise.
Agents do the volume, working in parallel lanes across Claude, Codex, Gemini, and more. I direct every lane, and nothing ships until it passes an adversarial cross-engine review. Speed comes from the fleet; trust comes from the gate.
You do. On the Sprint offer, full ownership of the code and assets transfers on cleared final payment. On both audits, the report and its findings are yours outright on delivery.
I never ask for credentials, production data, or payment details in the public enquiry. See Data & confidentiality above — an NDA or DPA is available before anything sensitive changes hands.
Use the enquiry form below, or send one email to hai (at) appaya.co with what you're deciding, auditing, or building. I reply within one working day.
Who's behind Appaya
Appaya is founded and directed by Luke Czak — ex-CPTO, with 13 years leading product and technology in UK regulated fintech, from Series A startups to global banks.
I scope the work, steer the fleets, and sign off every build before it ships.
More about Luke — lukeczak.comHai. Let's talk.
Tell me what you're deciding, auditing, or building. I reply within one working day, personally — and if I'm not the right person for it, I'll tell you who might be.
* required — everything else is optional
I'll reply within one working day to confirm fit and the likely offer. Nothing is billed, scoped, or agreed yet — that all happens after this.
Something went wrong on my end. Please email me directly instead and I'll pick it up from there:
hai (at) appaya.co