The build is the easy part.
Getting your people to use it takes twenty years of experience.
Most AI projects end at launch.
Ours start there.
The bottleneck moved.
Building software is no longer the hard part. We can put a working system in front of your users in three weeks. The hard part is everything around it: whether your team adopts it, whether it survives contact with your IT department, and whether anyone still uses it in six months.
Most AI consultancies solve the first problem and hand you the other three. I spent twenty years on the adoption side before I ever shipped a line of AI‑assisted code.
Services
All servicesStart here
Talks and workshops
Not every organization needs a system built. Some need a room in which to understand what changed. Each session is written for the room: your words, your material, your tools. Not a generic tour.
AI Readiness Audit
In two weeks, we map where AI actually fits in your operation, what it would cost, what would break, and what your people would need to learn. You get a prioritized plan and a build estimate. No obligation to build it with us.
Then build
Build Sprint
Three to six weeks from brief to a working system your users can log into. Built with Claude Code, handed over with CI/CD, access control and documentation, so your IT team signs off instead of pushing back.
AI Enablement Program
Your team gets the tools, the workflows and the guardrails, written up as an AI playbook for your business. Designed the way I designed training for 150 instructors across 15 countries: role by role, with real tasks, measured adoption, and a governance layer so nobody ships something dangerous on day three.
Work
See the workStartupKid Mentors
A multi-track recruitment platform for an education company
- 4 tracks
- 500+ candidates
- 83% storage reduction
An end-to-end hiring system that takes a candidate from a public form through video screening, dual review, interviews, scheduling and contract, across four tracks. In production and in daily use, with about ten operators behind it.
Rebuild
A live workshop app that stress-tests assignments against AI
- Built in 1 day
- Live audience
- 7 prompt iterations
Teachers redesign a real assignment from their phones and watch it land on a wall, where a server-side stress test tries to do it the way a student with AI would. The room generates the evidence about its own material, live.
A two-sided platform for a martial arts federation
Management for the federation and an app for every student, across countries, with hard data boundaries
- Two sides, one database
- 3 role tiers
- RLS-enforced
Two sides on one database: the federation runs enrolment, ranks, instructors, regional administration, finances and live analytics, and every student gets an app with their own progression. A director sees their own country, a student sees only themselves, and leadership sees an accurate roll-up of all of it.
Maximo
Real-time multiplayer where the secrets stay on the server
- 5 rounds
- Real-time
- Live classrooms
Pairs of players make five simultaneous choices; each choice is held in tables the client cannot query and revealed only after both have committed. Built for mentor-led classroom sessions, in Hebrew, right to left.
Who actually does the work
-
Ittai Perez
Strategy, learning design, product and build
I design how organizations learn to work with AI, then build the systems they end up using. I lead every engagement myself: I map where AI fits, design what your team practises and in what order, and run the build. This is what turns an AI tool into something your people still use six months later.
LinkedIn -
Bar Abudi
Production and governance
A DevOps and AI infrastructure engineer who runs Kubernetes-based AI platforms in production for hundreds of users. He owns CI/CD, secrets and access management, code review standards and risk classification. This is what turns a fast prototype into something your IT department will sign off on.
LinkedIn -
Shira Segev
Automation and integration
An information systems engineer who builds the connective tissue: workflow automation, API integrations, chatbots, and the technical specs that keep a build honest.
LinkedIn
Twenty years of training architecture, working software in weeks.
Book a 30-minute call