Drafting slips of this kind occur in every legislature, and India catches and corrects them just like any other. When the Narcotic Drugs and Psychotropic Substances Act was amended in 2014, a renumbered clause left the penalty section pointing to the wrong provision; Parliament corrected it in 2021, backdated to 2014. Smaller fixes are routine, through corrigenda and a standing Repealing and Amending Bill; the tax department issued one such corrigendum to the new Income-tax Rules in April 2026. Repair often come after a bill becomes law, because a crowded calendar leaves little time to catch slips before it passes. The 17th Lok Sabha referred 16 percent of bills to a committee and gave 35 percent of them less than an hour of debate, indicating potential value from assistive tools to aid, and free up the time of lawmakers.

India has already moved quickly to bring AI into its legal institutions. The Supreme Court uses AI to flag defects in electronic filings, a tool built with IIT Madras, and to translate judgments into 18 languages. Parliament’s Sansad Bhashini transcribes and translates debates in real time. District courts in nine states record testimony through Adalat AI. A checker that reads the text of a draft bill would extend this work to the drafting desk.

At the same time, care must be taken to ensure assistive use and suitable guardrails. On 2 July 2026 the Supreme Court set aside a tribunal order that relied on AI-invented precedents, holding that citing fabricated judgments is professional misconduct and any decision resting on them is void. The Kerala High Court permits AI in the district judiciary for assistance alone, with every output verified by a judge.

An assistive checker could read a draft bill for broken references, wrong citations and impossible dates, and return them to the drafter before the bill is introduced. It would strengthen the scrutiny a committee brings and leave every question of policy to elected members. The engineering is within reach and locally available: Indian firms such as SpotDraft, Lucio, and Jhana.ai already build software that checks contracts for such flaws. The result would be a cleaner statute book, with lawmakers still answerable for every word, supported by AI to make the mundane aspects of their responsibility easier.

At Bharat AI Initiative, we help public institutions adopt AI where it strengthens the work and stops short of the decision. Talk to us: contact@bharataiinitiative.org.

Sources: WIRED (Estonia, 9 July 2026); PRS Legislative Research (NDPS (Amendment) Bill 2021; Repealing and Amending Bill 2025; committee-referral and debate-time statistics, 17th Lok Sabha); JurisHour (CBDT corrigendum to the Income-tax Rules, 2026, 16 April 2026); Press Information Bureau (Supreme Court use of AI, 20 March 2025; Sansad Bhashini); ThePrint (Adalat AI in district courts); Inc42 (SpotDraft); LiveLaw and Bar & Bench (Supreme Court, Pooja Ramesh Singh v. Jammu and Kashmir Bank, 2 July 2026; Kerala High Court AI policy, 19 July 2025).

A platform conceptualized in 2014, and first launched in 2017 already provides a solution to reduce the gap between invoice recognition, and payment. TReDS, the Trade Receivables Discounting System, is an electronic platform regulated by the Reserve Bank of India.

The financier’s risk is the buyer’s default, so the discount is priced on the buyer’s credit. A two-year-old start-up selling to a solvent city borrows, in effect, at the city’s rating.

Already at scale

Financiers put Rs 3.5 lakh crore through TReDS in FY 2025-26, financing 16 lakh invoices across five licensed platforms: 4.5 times the volume of three years before.

The missing buyer

TReDS is open to private corporates, PSUs, and government entities. Government’s share of the TReDS flow is not published by RBI, which publishes volumes by platform, not by buyer type. What the record shows instead is government being ordered to participate.

The centre has directed its own enterprises onto the platform three times in eight years: in 2018, in 2024, and again in the 2026-27 Budget, which makes TReDS the compulsory settlement route for central-PSU purchases from MSMEs. An order that has been issued three times suggests friction or bottlenecks in actual implementation. No mandate is addressed to the states; a parliamentary standing committee’s recommendation to put state-government dues on the platform remains a recommendation.

Impact on collaboration with startups

The companies cities want to buy from are young, tech-focused, asset-light and pre-profit: the least able to carry a 60-day receivable and a bank guarantee at once. For a young AI or tech startup, the working capital cost of selling to government is the recurring barrier. A city that wants innovative AI suppliers, has to be a customer they can afford to serve.

None of this is automatic. Financiers favour highly rated buyers, so invoices drawn on weaker ones can go unfunded; the 2026-27 Budget adds a credit guarantee for invoice discounting. The platform reaches small and medium firms more readily than the smallest micro-enterprises. And it works only if the buyer confirms invoices promptly, every time.

The system is built, regulated and moving Rs 3.5 lakh crore a year. What remains is the ability and willingness of state and city governments to use it.

At Bharat AI Initiative, we help cities and public institutions become customers that young start-ups can afford to serve, from first pilot to lasting supplier. Talk to us about your city’s first pilot: contact@bharataiinitiative.org.

Bharat AI Initiative. Figures are RBI’s official Entity-wise TReDS Statistics: value of factoring units financed each month, 48 monthly statements (April 2022 to March 2026) summed by financial year, verified July 2026. Other sources: RBI TReDS FAQ, Notification of 7 June 2023 and (Trade Receivables Discounting System) Directions, 2026; PRS Legislative Research (Standing Committee on the Factoring Regulation (Amendment) Bill, 2020); Press Information Bureau (Union Budget 2026-27); Swarajya on MSME payment cycles (January 2026).

The two don’t usually connect. Standard procurement is built for established suppliers, so promising young companies and forward-looking cities often miss each other.

A handful of programmes around the world have bridged that gap.

Be the first buyer. The best programmes commit, up front, to buying what works. India's iDEX signed contracts with 430 of 619 startups: a 69% rate. Dubai's Future Accelerators converted 74% of participants into signed agreements, with no equity and no grants.

Fund in tiers. No city backs a moonshot with a single cheque. The strongest schemes open with a small discovery grant and step up as the solution matures: from a few lakh for a prototype to crores for deployment.

Open the city as a sandbox. Helsinki, Seoul and Amsterdam give startups supervised access to live infrastructure and open data. A city that lets startups test gets first pick of what they build.

And a few more well-documented important supporting moves:

Pay fast. Prompt-payment targets and invoice discounting keep young firms solvent between payments.

Build the talent pipeline. University partnerships and intern support supply the talent, which startups need to deploy in your cities.

Give startups one door. A single-window desk gives founders one place to go.

None of this needs a big budget: just clear structure and the habit of treating startups as partners, rather than suppliers.

At Bharat AI Initiative, we want help cities and public institutions design collaborations like these from first pilot to lasting supplier. Talk to us about your institution’s pilots, or your efforts to engage with the government. contact@bharataiinitiative.org.