What Pakistan becomes
when AI is built in.
نیا پاکستان، مصنوعی ذہانت کے ساتھ
A working prototype of the $1 Billion national AI plan by 2030: the vision A-Z, and a concrete delivery model to train 1,000 PhDs and 1,000,000 professionals. Announced by PM Shehbaz Sharif at Indus AI Week, 9 February 2026.
The youngest large nation on earth, meeting the most powerful technology of the century.
A demographic dividend with a deadline
64% of Pakistanis are under 30. That window is open now. AI skills are the highest-leverage way to turn 240 million people into a productive, exporting digital workforce.
From IT outsourcing to AI value
Pakistan already exports IT services. The plan moves the country up the value chain: from doing tasks cheaply to building AI products the world pays a premium for.
Leapfrog, not catch up
Countries without legacy systems can adopt AI-native services directly. Pakistan can skip a generation of infrastructure in health, education and government.
Not a slide about AI. AI, working.
This is what the plan delivers on the ground: AI that speaks Urdu and serves a farmer in Sahiwal, a clinic in Loralai, a citizen at a counter, a child in Gilgit. Press a tab and watch.
The region is spending billions. Pakistan's edge is 240 million people.
Capital is flooding into AI across the neighbourhood. Pakistan cannot out-spend petro-states, but it can out-talent them. The $1B, aimed at people, is the highest-leverage move on the board, if it is delivered, now.
The young win
AI rewards whoever can deploy and operate it at scale. That is a labour game, and Pakistan has the largest, youngest talent pool in the region.
The window is ~24 months
Standards, supply chains and talent flows are being set right now. Move in 2026 and Pakistan is a builder; wait and it becomes a buyer.
Talent is the export
Gulf and Western demand for AI-skilled workers is exploding. Train at home, earn abroad, remit back: a foreign-exchange engine.
Pakistan, A to Z, with AI integrated.
Pick a sector to compare where it stands today against where the 2030 plan takes it. Agriculture, mining and minerals are the government's stated priorities; the same model extends across the economy.
Agriculture
23% of GDP, 37% of jobs, but yields per acre trail regional peers. Decisions on water, seed and pesticide are made by guesswork.
Satellite + soil-sensor models advise every farmer in Urdu and regional languages by SMS and voice. Crop-disease detection from a phone photo. Water optimised field by field.
A $1B input. Move the sliders. Watch the output.
This is not aid spending, it is the highest-return investment the state can make. Today Pakistan exports about $3.2B in IT services; this plan targets $10B+ by 2030, and that is only the export slice.
Conservative model for discussion. Even modest per-person gains compound across a workforce of this size.
The National AI Talent Pipeline
One coherent ladder, three tiers. It turns the headline numbers into a delivery model with curriculum, duration and outcomes, run through regional hubs so no province is left behind.
Non-IT professionals: teachers, clerks, farmers' co-ops, nurses, SME owners, civil servants.
Confident daily use of AI tools in their own job. A certified, searchable national talent registry.
- ▸AI literacy and safe, ethical use
- ▸Prompting and everyday AI copilots
- ▸Job-specific tooling (sector tracks)
- ▸Data privacy and Pakistani context
- ▸Capstone: automate one real task at work
Existing IT technicians and graduates converted into applied AI builders.
Can build, fine-tune and deploy AI solutions. Feeds startups, government and the export sector.
- ▸Python, data and ML foundations
- ▸Applied LLMs, RAG and agents
- ▸MLOps and responsible deployment
- ▸Urdu / regional-language NLP
- ▸Industry capstone with a real employer
Top graduates funded through PhDs at home and at partner universities abroad.
A world-class research base anchoring National AI Centres of Excellence and bonded to serve Pakistan.
- ▸Frontier ML and foundation models
- ▸Low-resource-language AI
- ▸AI for agriculture, health, minerals
- ▸AI safety, policy and governance
- ▸Publish, patent and supervise Tier 2
Delivery model
Train-the-trainer multiplies reach: a small expert core certifies thousands of instructors, who run hybrid cohorts from regional hubs and a national online campus. Employers co-design capstones so graduates are hired, not just certified. Every learner lands on a public, verifiable talent registry.
Hubs in every province, AJK and GB
Deliberately decentralised. The policy rollout stalled when provinces were not engaged; here each unit owns local delivery from day one.
An honest allocation, not a slogan.
An illustrative breakdown of the announced investment across talent, infrastructure and adoption. Figures are a starting point for discussion, designed to be defensible and measurable.
Not a someday plan. The first 90 days.
Give the mandate and delivery begins immediately. Inside three months there are real learners, real pilots and a public dashboard, not committees.
- ▸Constitute the delivery unit and name accountable owners
- ▸Sign MoUs with provinces, AJK and GB so no one waits
- ▸Lock curriculum v1 and select the first hub cities
- ▸Launch the national online campus (LMS) and first 3 hubs
- ▸Certify the first 500 master trainers (train-the-trainer)
- ▸Open Tier-1 enrolment and the public talent registry
- ▸First 25,000 Tier-1 learners in training
- ▸First PhD cohort shortlisted; agri + health pilots in the field
- ▸Public KPI dashboard live, money and milestones visible
A sequenced roadmap with public milestones.
- ▸Stand up the delivery authority and finalise AI Council
- ▸Launch first 10 regional hubs and the national LMS
- ▸First 100,000 Tier-1 learners; first PhD cohort selected
- ▸Tier-1 crosses 400,000; Tier-2 bootcamps live nationwide
- ▸Sector pilots in agriculture and health show measured gains
- ▸Sovereign compute phase 1 online
- ▸AI services embedded in federal citizen services
- ▸Minerals and agri models in production at provincial scale
- ▸AI/IT service exports growing toward target
- ▸1,000,000 professionals certified; 1,000 PhDs in the pipeline
- ▸Centres of Excellence self-funding through research and IP
- ▸Pakistan recognised as a regional AI talent exporter
Governance that fixes the current stall.
The policy exists; delivery stalled on unclear ownership and a missing provincial response. This structure assigns accountability and makes progress visible.
National AI Delivery Authority
A single accountable body that owns the roadmap, the budget and the KPIs, reporting to the PM and the AI Council. Ends the current stall caused by unclear ownership.
Provincial AI Units
Each province plus AJK and GB runs a unit that localises delivery, so the plan does not wait on a single centre. Designed to fix the missing provincial response noted in the policy rollout.
Public KPIs, quarterly
Learners certified, hubs live, pilots measured, exports earned. Open dashboards so Parliament, press and citizens can see progress and money spent.
Industry & diaspora board
Pakistani AI talent in the country and abroad advises on curriculum, hiring and standards, keeping the programme current with the frontier.
Every real objection, and how it is de-risked.
A serious plan names its own risks before anyone else does. Here are the ones that matter, and the answer to each.
Bonded service for funded PhDs, strong domestic demand, and a deliberate ‘train at home, earn abroad, remit back’ track. The diaspora becomes an asset, not a loss.
Provincial AI Units own delivery from day one, and funding is tied to participation. The plan does not wait on a single centre.
Milestone-based disbursement, public quarterly KPIs, open dashboards and independent third-party audit. Nothing is spent in the dark.
Pilot-first and measured. Scale only what shows results in the field, sector by sector, district by district.
One standardised national curriculum, a certified master-trainer corps, and employer-validated capstones so a certificate means real capability.
What we publicly commit to measure
Reported every quarterStraight answers.
Why one national talent pipeline instead of scattered schemes?+
A single ladder, Tier 1 to PhD, with one curriculum, one registry and one accountable owner, is auditable and avoids the duplication and gaps that fragment past programmes.
How is $1 billion enough for a country of 240 million?+
Because it is spent on leverage: people and pilots. Train-the-trainer multiplies reach, an online campus drops per-learner cost, and pilots prove return before any large rollout. The plan is designed to compound, not to be consumed.
Why trust an individual to help lead this?+
On evidence, not titles. This entire blueprint, the live AI demonstrations, the site, the budget and the delivery model, was built solo, fast, in the open. The work itself is the credential.
What about women and remote regions?+
Hubs are placed in Balochistan, Gilgit-Baltistan and AJK, the online campus reaches anyone with a phone, and women-focused cohorts are built into Tier 1 from the start.
How quickly can it start?+
Within 90 days. The first 25,000 learners can be in training inside three months, with a public dashboard showing progress and spend.
Is this an official government document?+
No. It is a citizen’s proposal prepared for discussion. Figures are illustrative and meant to begin a serious conversation, not to bind anyone.
You just used four AI products. He builds those for a living.
Zaib Khan is a Pakistani software engineer who has been writing code since 2003. Over two decades he moved from web and mobile engineering into applied artificial intelligence, where he now works full time: designing, building and shipping production AI products end to end, from large-language-model apps and autonomous agents to voice assistants.
He has shipped dozens of products, solo, each taken from idea to live deployment. The demo you just played, this entire site, the plan, the budget, the domain and the email system behind it, was designed and built by him alone, in the open, fast. That is the exact operating speed this programme has been missing.
His signature is AI for his own people: tools for Pakistani and Pashtun communities, including an anonymous AI-mediated jirga that helps people reach a fair faisla in their own cultural frame, and tooling for local languages. He is convinced the next decade of AI must speak Urdu, Pashto and Sindhi, not only English. He is also a recording artist published on Apple Music and YouTube, a creative streak that shapes how he thinks about product and reach.
Builder, not bystander
Has already shipped the kind of AI this plan needs, at production quality.
Local to the core
Builds for Urdu, Pashto and Pakistani realities, not Silicon Valley defaults.
Delivers, fast
Idea to live in days, the antidote to the policy's current stall.
Anonymous AI-mediated jirga for Pashtuns — three voices, one faisla, in local cultural frame.
AI that drafts a reply to every Google review, on autopilot, for small businesses.
AI chase-letters and statutory-interest engine recovering late invoices for SMEs.
Production voice assistants that book, qualify and answer calls end to end.
Hollywood-grade original songs, composed and produced end to end with AI, released on YouTube, Apple Music and major streaming stores.
Dozens of products taken from idea to live deployment, solo, since 2003.
“Give me the mandate and the first 100,000 are trained before this time next year. Pakistan does not need to import this capability. It needs to be trusted to build it.”
You can copy a vision. You cannot copy delivery.
This site is deliberately the summary. It is meant to convince, not to be a manual. The detail that makes the programme actually work, and the person who can run it, are not on this page.
This page is roughly 5% of the work
What you are reading is the executive summary. The full implementation playbook, a 400+ page document covering the curriculum, master-trainer certification, hub operations, procurement, district-by-district rollout and the KPI framework, is held privately and shared only under NDA with a serious counterparty.
The systems are already built
The Urdu AI demonstrations on this page are not mock-ups of an idea. They reflect production AI this builder already ships. Reproducing that capability is months of scarce, specialist work, or a single conversation.
Execution is the moat, not the document
Plans do not deliver; people who have shipped do. The National AI Policy already exists. It stalled for exactly one reason, no one who could actually build it owned it. A copied summary inherits the same fate.
It moves as one package
The plan, the operating model, the talent network and the builder come together. Take the summary and you hold a pitch deck. Bring in the builder and you have a running national programme.
The blueprint is shared freely. The 400-page playbook, the operating model and the builder who has already shipped this kind of AI come together, under NDA, to whoever is serious about actually delivering it.
Request a briefing for your ministry or office.
Fifteen minutes. The vision, the delivery model, and exactly how the first cohort starts in 2026. For ministers, secretaries, the AI Council, universities and industry partners.
Download the formal application (PDF)This is a prototype for discussion, not an official government document. Figures are illustrative and meant to start the conversation.
Where figures are modelled or indicative, they are labelled as such on this page. This is a citizen's proposal prepared for discussion, not an official Government of Pakistan document.