Citizen AI · A Build Haifa initiative

Teach a city to build with AI.

Israel adopts AI faster than any country on earth. Yet almost no one has been taught to build with it. Citizen AI is a plan to close that gap in one city, and turn capable people the old economy left behind into builders.

23%Of Israeli workers, roughly one million people, hold jobs AI is expected to threaten (Mosaic and the Taub Center, April 2025)
95%Of Israeli high-tech workers use AI tools regularly, and 78% use them daily (Israel Innovation Authority with Myers-JDC-Brookdale, Nov 2025)
62%Wage premium for AI-skilled workers (PwC Global AI Jobs Barometer, 2026)
~10,000Jobs projected for the Haifa region with Nvidia's Kiryat Tivon campus (2027-31)
Using AI and building with it are different skills. Israeli adoption is close to universal inside high-tech. Far fewer can build agents, automations and products with it, and that is the skill the market pays for. Closing that gap is the whole point of this program. The full economic case is on the Reinvention Index →
Why this, why now

The old engine stalled. Israeli tech is cutting hard, AI is taking the routine and entry-level work first, and no country can out-spend the US and China on AI compute. The one edge Israel keeps is people who can build. So the move is not to wait for the boom, it is to retrain the displaced into AI-enabled builders, in the one city with the density to prove it can be done.

The plan

A program to turn AI users into AI builders

Not another learn-to-code bootcamp. Using AI is already common in Israel. Building with it, shipping agents, automations and real tools, is rare, and it is what employers pay a premium for. Citizen AI is designed to teach mid-career and displaced adults, olim and non-engineers to build. The intent is a funded apprenticeship, so people between jobs can afford to take part, paid for by the backers we are recruiting rather than by tuition. To be precise, because the two numbers are often confused: delivery costs about NIS 9,000 per learner, and the stipend that makes participation possible costs materially more on top. Every frontier program that achieves the results we cite pays its participants. The Index cost model prices delivery only; the pilot budget prices both. Every module ends in something shipped, not a certificate.

Who it is for
Archetype A

Displaced tech

Laid-off engineers, PMs, QA, support and ops from the 2026 cuts, with deep domain knowledge and no need for a junior-dev reset.

Archetype B

Unplaced grads

New graduates who cannot break in, who need to arrive AI-native instead of sending 700 resumes for 20 roles.

Archetype C

Olim & returnees

Mid-career professionals from Russia, Ukraine, France and North America, with skills and language barriers but no tech pathway today.

Archetype D

SMB owners

Shop, clinic and service-business owners who need AI leverage, not a computer-science degree.

Archetype E

Admin & back office

Insurance, banking, government and back-office staff whose routine desk work AI is already automating. Israeli-born or not, degree or not. The Index puts this group at the highest automation exposure of any occupation in Haifa, and it is the group this program was built for.

Archetype F

Returning reservists

Miluimnikim back from extended duty whose roles moved on without them. The workforce lost 202 average days per reservist since October 2023; this is the track for rebuilding a career that stalled while you served.

The proposed 16-week build
Weeks 1-2

Set up and mindset

A working AI workspace from day one, and the operator mindset: build with AI, do not compete against it.

Ships: live AI setup + ICP one-pager
Weeks 3-10

Build core

Real tools, week after week: a personal AI system, then agents, automations and no-code AI apps for real use.

Ships: a personal AI system + 2-3 tools
Weeks 11-16

Apply and place

An AI system that solves a real employer or city problem, built inside a host organization, then a revenue motion and Demo Day.

Ships: a deployed system + a job pipeline

Built from proven material  The curriculum draws on Yuri Kruman's own work: the book Closing the AI Wage Gap, the 12-week Career Beast Mode program and a battle-tested Claude setup protocol.

The destination

The jobs this leads to, and how you get in

The point is not a course. It is a job. Building with AI pays a large and rising premium, and increasingly it is hired on proof of what you have built, not on a computer-science degree. That is what makes a real path possible for people starting over.

RoleWhat you buildIsrael (NIS/mo)Global (USD/yr)Entry bar
No-code / automation builderAgents and automations that do the work of many inside a business function~25-40K~$86-123KLowest: portfolio, no degree
Applied AI / app developerCustomer-facing AI apps, RAG systems and assistants~30-43K~$100-150KPortfolio + some coding
AI solutions / implementation engineerScopes and ships AI into real operations, the top enterprise hire~40K+~$120-140KDomain + AI fluency
RAG / AI agent developerRetrieval-augmented and agentic systems~43K~$100-160KPortfolio, no PhD
AI-augmented operatorYour existing role, done with AI leveragein-rolerising fastYour experience + AI skills
Independent AI builderContract and freelance builds for clients anywhere, from Haifamarket$50-200+/hrPortfolio + track record

Read this honestly. No cohort has run yet, and no non-engineer has been placed by this program. These are the market rates the program targets and the ladder it aims at, not an exit salary and not a promise. A first role out of a 16-week program sits below these bands; the figures here are where the track leads over 18 to 36 months.

These are the applied tier, where enterprise hiring volume actually is and where a retrained adult can realistically land. The elite ML-research tier pays far more but needs a computer-science or quant degree, so it is not this program's destination. Israel figures are directional, from 2025 recruiter reports (Globes, GotFriends); global figures are from salary aggregators and are self-reported. For scale: the overall Israeli tech average was about NIS 37,100/month in 2025, rising to about NIS 39,800 in 2026, and AI-specialized companies paid the most. Watch the trend as well as the level: GotFriends' H1 2026 report has AI-engineer pay up only 5% year on year, behind security research at +25%. The AI premium is real, and it is no longer the fastest-moving line in the market.

Demand, in Israel, not the US  AI hiring in Israel doubled in the first half of 2025 versus all of 2024 (GotFriends). The country has 2,200+ AI companies it cannot fully staff, the state is importing 200 AI experts from abroad because it cannot grow them fast enough (Israel Innovation Authority), and Nvidia is building a campus for up to 10,000 people twenty minutes from Haifa.

The honest part  The old "junior developer" door is narrowing as AI absorbs entry-level work, and overall Israeli tech headcount is flat. So the position that wins is not junior coder, it is an AI builder who ships. That is exactly what this program is built to make you.

The real path
Step 1Build in publicLeave with a portfolio of shipped, deployed AI tools that prove capability.
Step 2Audition on the jobThe final month is a real project inside a host employer who has watched you work for weeks.
Step 3Convert or contractTurn the host project into a hire, or start earning on contract and freelance builds.
Step 4Grow the ladderMove from first role into higher-leverage AI-builder and product work as demand rises.
Getting hired

We build the bridge to employers ourselves

This is the hardest and most important part, so we are direct about it. The one lesson from every Israeli program that actually places people is that the employer is inside the training, not contacted after graduation. We do not rely on any university's career office. We build the placement engine ourselves.

The strongest lever

The capstone is the interview

The final project is a real brief from a real northern employer, who watches the work for weeks before deciding. Employer-embedded capstones are the common feature of Israel's strongest placement records (Israel Tech Challenge has reported 94-96% historically, and about 82% at six months in more recent third-party reporting; Excellenteam about 75%). Those results come from heavily screened, largely degree-holding cohorts, so they are a ceiling to aim at, not a rate to assume.

The demand pool

A northern employer consortium

We recruit employers as founding partners before the first cohort: Matam park in Haifa (~15,000 professionals across ~80 companies), the Yokneam and Nvidia ecosystem, defense and health-tech, and local SMBs adopting AI.

Who can work where

We route by clearance, honestly

Much of the north's hiring is defence, and defence work needs security clearance that recent olim cannot hold for years after arrival. Pretending otherwise would waste people's time. Clearance-eligible participants are routed to defence-track capstones; olim and returnees are routed to the Matam multinationals, health-tech, SMBs and the freelance bridge, which is where their work is actually placeable now.

The scouting floor

Hackathons where employers watch

Sponsors set the briefs, scout builders live and hand out interviews. A hackathon is the cheapest way a company ever auditions AI talent, and where graduates get found.

The credential

Portfolio first, plus a freelance bridge

Your shipped tools are the credential in applied-AI hiring. Contract and freelance work is an honest supplement and a bridge to build a track record, not a promised income.

What we still have to prove  No one has yet run an applied-AI program placing non-engineers into northern employers at scale. Generating that evidence, cohort by cohort and measured independently, is the actual work. We will not claim a placement rate we have not earned.

Running a program in a live conflict  Reserve call-ups and rocket alerts are facts of life here, not hypotheticals, and a plan that ignores them is not a plan. Every session has a remote fallback and every venue has a shelter route. A participant called up mid-cohort keeps their place and rejoins the next intake with no penalty and no repeat cost. Reservists are reported as a named sub-cohort in the independent evaluation, so the effect of service on retraining outcomes becomes evidence rather than an excuse.

Who is accountable  This is currently one person carrying execution risk, which is a real dependency and we will not pretend otherwise. Before cohort 1 opens: a fiscal sponsor named with its registration number, and a three-to-five person advisory board including an Israeli workforce-development professional, a figure from the mixed-city civic sector, and a finance signatory who is not the founder. No institutional funder should commit before that exists, and we are not asking one to.

Citizen AI Hackathons

Where the city builds, out loud

Hackathons are the heartbeat of Citizen AI: weekend build sprints where residents, cohorts and newcomers ship real AI tools for real Haifa and employer problems. They are the front door, the showcase and the hiring floor all at once.

The front door

Anyone can show up and build something with AI in a weekend. No prerequisites. It is how people discover they can do this, and how the program fills its cohorts.

The showcase

Real tools, judged and demoed in public on real city and business briefs. It is the proof the idea works, and the reason press, backers and partners keep showing up.

The hiring floor

Employers sponsor briefs, scout builders and hand out prizes and interviews. It is where a company auditions AI talent cheaply, and where graduates get found.

The pilot

Small, provable steps, each one funding the next

Every stage is gated on the last one's measured proof, so the second cohort is paid for by the first cohort's results. The evidence is the whole strategy.

First

Citizen AI nights

Monthly builder evenings, open to all, on a host campus we are approaching. Low-risk, high-signal, the top of the funnel.

Then

The first hackathon

A weekend build on real briefs, with employer sponsors scouting and the first cohort applicants surfacing.

Then

The first cohort

About 24 learners, funded and placed, measured by an independent evaluator. Cohort 1 is displaced tech professionals only: the shortest distance to real placement, and the fastest way to produce evidence. The other archetypes are the roadmap, not the first class. Every frontier program we borrow from runs a single, screened intake before it widens.

Then

Scale & train the trainers

Two tracks, then graduate-facilitators, aligned to the Kiryat Tivon hiring ramp.

Why it can work

The models we borrow from

We are a pilot with no track record yet, so we do not claim to match these programs. We borrow from the best of them: Singapore's AI Apprenticeship Programme (funded, employer-embedded, placement over 90%), Year Up in the US (the one such program with randomized proof of durable wage gains), and France's Simplon (inclusion-first, where placement is treated as real work, not an afterthought). And we commit to third-party, longitudinal outcome measurement from the first cohort, because in this field verified evidence is the scarcest asset, and the fastest way to earn the next backer.

The coalition

The partners who make this real

Citizen AI is ready to run. What it needs now is its founding partners, and here is exactly who they are and what each one does. These conversations are underway, not signed, and we say so plainly. But the plan is built, the evidence is live, the founder has already made this move himself, and every ask below is small, specific and built to say yes to.

The Technion, first

Host the first builder nights

One evening session, one named room, capped at 30 people, evaluated before anything renews and with no Technion branding until it has. Public liability insurance is carried by the fiscal sponsor and the host venue is indemnified; registration and a capacity cap apply, and the events are 18+. In return: facility fees at standard third-party rates, first-look access to cohort outcome data for the Technion's own labour-market research, and a defined referral channel into Technion continuing education for graduates who want to go further.

The first move
Haifa City Hall

Champion a workforce win

A municipal champion and a letter of support, time-boxed to the first cohort's independently measured results, revocable at the city's request at any time, scoped to the Citizen AI program only and never to the Index or its research claims, and not transferable to any later ask without a fresh decision. No data to share, no budget the city underwrites. Any funding conversation with the Haifa Economic Corporation is a separate decision on a separate timeline and does not follow from this letter.

The civic play
Northern employers

Become a founding employer partner

Sign a capstone-host MOU: one real brief, one internal mentor, a per-seat contribution and a right to first interview. Cohort size is capped at the number of signed seats, so the program cannot outrun real demand. Target: five signed before cohort 1 opens. This is the instrument every program we borrow from has and we do not.

The demand side
Haifa Development Corp

Sponsor the pilot

NIS 260,000 for cohort 1: 24 learners at the published per-learner blend, plus the stipend that makes participation possible for people between jobs, over one 16-week term. In return: named founding sponsorship, the outcome data, and first refusal on cohort 2. We are seeking this as matching capital alongside an anchor funder and signed employer partners, not as the first institutional cheque, and the full cost derivation is published on the Index methodology page.

The economic bet
Who is building this

The founder made this exact move

Citizen AI is not a theory. It is its founder's own story. Yuri Kruman is an oleh hadash who left New York, landed in Israel and had to reinvent himself to survive and support his family, making the exact corporate-to-AI-builder transition this program teaches, from three chief people officer seats into building with AI, in public. He is a Top-6 global HR thought leader on the future of work, wrote the books that are the program's spine, built the course it runs on, has raised high-net-worth money in this exact city and lives in Haifa. He is asking Haifa to do, at city scale, what he already did himself.

The funding we are assembling
  • Target anchor

    IIA Human Capital Fund matched funding, built for exactly this population

  • Approaching

    Google.org, Rashi, Edmond de Rothschild and diaspora philanthropy

  • Demand side

    Employers who host projects, sponsor hackathons and hire

  • Doors we work

    Mosaic, Tsofen and northern employers, for thesis, cross-sector reach and hiring. Haifa is one of Israel's principal mixed cities, roughly 10-11% Arab residents, and a program that does not reach them is not a Haifa program. Target: at least 15% of every cohort, with Arabic alongside Hebrew in recruitment materials, and Tsofen involved in co-recruitment rather than named as a logo.

Measured

An independent evaluator, a university social-policy group or an equivalent contracted research body, named before cohort 1 opens and never the program itself. Pre-registered outcomes: placement or income gain at 6 months, wage delta against a matched comparison, retention at 12 and 24 months. Participant consent and data protection are agreed in writing before intake. We report what shipped, not what we pledged, including the results that go against us.

Accountable

We are seeking a fiscal sponsor, an established registered nonprofit to receive and account for funds, and no money moves until one is named here with its registration number. There is no vehicle of our own yet and we will not solicit against one that does not exist. No city bank accounts, city letterhead or city staff are involved in any donor process.

Replicable

Built as a model, not a one-off: a method a national office could run in any Israeli city once Haifa proves it.

Get involved

Build Haifa, with Haifa.

Whether you want to learn, host, hire, fund or just follow along, the movement starts with a conversation. Tell us who you are and we will be in touch.

Use the form and we will reply personally.