Dr Sahibzada Ali Mahmud, Chief Digital Public Infrastructure Officer at the Pakistan Digital Authority (PDA), has argued that WASL, Pakistan’s National Data Exchange Layer, can become the nervous system for the country’s artificial intelligence (AI), writing in The News. It is the third article in his series on Pakistan’s digital public infrastructure, and he is also a former managing director of the KP Information Technology Board. He opens with a scenario in which a provincial finance officer must tell her minister how many pensioners in three flood-hit districts are still unpaid. Today that takes letters, calls, and a week of reconciling spreadsheets, but with a government AI workspace and a pensions agent querying each district’s own system, she could get a traceable draft briefing within minutes and sign it off herself.
His central point is that capable AI models already exist, and what Pakistan has lacked is a safe, standard way to reach authoritative government data without copying or leaking records. He describes two kinds of knowledge for an AI. The first is what the rules say, such as laws, notifications, and open data, which a model can look up, and the PDA, guided by the Ministry of IT and Telecommunication, has already issued the first Machine-Readable Legislation and Policy as Code. The second is what is true today about a person, a plot, or a business, which stays with custodians such as NADRA, the Federal Board of Revenue, and provincial revenue authorities. Under WASL, an AI asks the relevant office a specific question with permission, and the office answers from its own system, while WASL keeps no copy and records a tamper-proof receipt.
Mahmud says this could support a sovereign Pakistani AI model tuned for Urdu and regional languages, along with a marketplace of specialised agents built by departments, universities, and start-ups. Because providers would serve data in the same national format, an agent built for Punjab’s land records could work with any province that follows the standard. He says every agent would be treated like any other user, having to prove its identity, obtain consent where personal data is involved, and leave a record of which model acted, for whom, and why, with no special back doors. Citizens would benefit through departments and regulated institutions connected to WASL, with consent given through PAK-ID. He imagines assistants on phone lines or WhatsApp in Urdu, Punjabi, Sindhi, Pashto, and Saraiki that check a live record and tell people where their case stands.
He lists further uses, including scholarship and support applications filled in from verified records, doctors’ assistants summarising a patient’s records from different hospitals, and bank AI checking a shopkeeper’s land holding and tax status for a loan with his consent and nothing more. In many cases a verified yes or no is enough, such as confirming that income is below a subsidy threshold without reading a tax return. He also says provincial departments could compute their own figures and share only aggregated results, so dengue spread or maternal health gaps can be studied while data stays with its owners. He argues that consent rules and audit trails are what will make institutions trust AI, that an AI under WASL cannot consent for a citizen or see more than a human officer in the same role, and that personal AI agents acting for citizens remain switched off until liability is settled. His conclusion is that the model is not the lasting advantage, trusted access to data is. The piece gives no rollout dates, and the draft National Data Governance Policy that proposes WASL still needs federal cabinet approval, according to Arab News.
Follow the SPIN IDG WhatsApp Channel for updates across the Smart Pakistan Insights Network covering all of Pakistan’s technology ecosystem.
