Acceptance criteria become test sets and graded rubrics. If you can't measure "good", you can't ship a probabilistic product.
Featured MVP · Live on Hugging Face
Ask SEBI circulars. Get cited answers — or an honest "not in the corpus".
SEBI Circular RAG is an open-source, AI-native product I built to answer questions over 1,490 SEBI circulars. It cites every claim, flags circulars that have been superseded, answers "as of" a past date, and refuses to guess when the evidence isn't there.
1,490SEBI circulars indexed
0.943Recall@10 on 260 test queries
0.881Citation recall
0.981Abstention accuracy
Metrics are from a full run on 15 Aug 2026 against the 260-query golden_v7 set (728-circular corpus at the time). A 15 Sep 2026 check on the 1,490-circular index still cleared every quality gate. Not legal advice.
Field guide · updated 24 September 2026
AINPM.in is my practical guide to AI-native product management, built from India. It covers what changes for PMs, the Indian AI and regulation landscape, and a playbook — with the SEBI RAG MVP as the working example.
01 · What changes
Six shifts that make a PM "AI-native"
AI-native product management isn't adding a chatbot to a roadmap. It changes what a spec is, how quality is proven, and what a unit of work costs.
Design the job to be done end-to-end — tools, memory, hand-offs and where a human approves — not just the UI.
Every task has an inference bill. Model choice, context size and caching are product decisions with margin impact.
Models improve monthly. Plan in short bets, instrument everything, and re-run evals whenever the model underneath changes.
Consent, labelling, audit trails and kill-switches belong in the first sprint — regulators in India now expect them.
Voice, vernacular and low-bandwidth are defaults, not localisation tasks — Indic models make this practical today.
02 · India lens
Why India is a distinct place to build AI products
Sovereign compute, Indic foundation models, population-scale digital rails and the world's largest pool of capability centres — a combination no other market has.
IndiaAI compute
Subsidised national compute under the ₹10,372 crore IndiaAI Mission, with 20,000+ more GPUs announced in February 2026.
Source ↗Indic models
BharatGen Param2 (17B) spans 22 Indian languages; Sarvam launched 30B and 105B models plus speech and vision at the AI Impact Summit.
Source ↗Capability centres
2.36 million professionals in India's GCCs (FY2026) — increasingly owning global AI products, not just delivery.
Source ↗Public infrastructure
UPI, Aadhaar, DigiLocker, Account Aggregator and ONDC give AI agents identity, consent and payment rails to act on.
03 · Regulation radar
What an Indian AI PM must track
India has chosen principles-based guidelines plus sector regulators over a single AI Act. That means your obligations depend on your domain — and several dates are close.
RBI FREE-AI framework
Seven "sutras" for responsible AI in regulated finance — trust, people first, fairness, accountability, understandable by design, safety and resilience.
MeitY India AI Governance Guidelines
Principles-based, techno-legal approach relying on existing law and sector regulators rather than a standalone AI statute.
DPDP Rules notified
Data Protection Board provisions active; the rest of the Digital Personal Data Protection regime phases in.
IT Rules amendment on synthetic content
AI-generated media must be prominently labelled; takedown windows cut to 3 hours (2 hours for deepfake intimate imagery).
SEBI AI/ML guidelines Pending
SEBI's chairman has said guidelines will mandate human-in-the-loop, kill-switches and data governance for AI in securities markets.
DPDP consent managers go live 7 weeks
Registered consent managers become operational — design your consent UX to interoperate.
DPDP full compliance Deadline
Consent, itemised notices, security safeguards, breach reporting and Significant Data Fiduciary duties apply; penalties up to ₹250 crore per instance.
This summary is for orientation only and is not legal advice.
04 · Playbook
An India-first AI-native PM loop
Eight steps I use to take an AI product from problem to production. See them applied in the SEBI RAG case study →
Frame the job, not the model
Start from a user outcome and a baseline. Is AI 10× better here, or just novel?
Write evals before prompts
Golden datasets in the languages and dialects your users actually speak; rubric-graded, versioned, owned by the PM.
Choose models by ₹ per task
Compare frontier APIs, open-weight and Indic models on quality, latency, cost and data residency.
Ground it in your data (RAG)
Retrieval over trusted sources with citations — essential for BFSI, health and government use cases.
Design agents with guardrails
Scoped tools, approval gates, audit logs and a kill-switch. Human-in-the-loop where money or rights move.
Consent & privacy by design
Purpose-limited data, itemised notices and consent-manager readiness for DPDP.
Ship small, observe everything
Trace every call, sample outputs for review, track cost and quality drift per cohort.
Measure outcomes, then scale
Task success, time saved and trust signals — not just DAUs. Re-run evals on every model change.
05 · Start here
Where to begin, depending on who you are
- Build one eval set for a feature you own this month — even 50 labelled examples change the conversation.
- Learn the cost model: tokens, context windows, caching and model routing.
- Prototype yourself with AI coding tools — PMs who can ship a working demo move faster than specs.
- Map your data flows against DPDP before May 2027.
- Show, don't tell: publish two small AI products on GitHub with a written eval and a cost breakdown.
- Pick a Bharat problem — vernacular, voice or low-connectivity — and use an Indic model.
- Learn the basics of RAG, agents, prompt design and evaluation; skip hype, keep notes.
- Target GCCs and AI-first startups — both are hiring for AI product skills.
- Fund evaluation infrastructure as seriously as feature work — it is your quality moat.
- Price for inference: make unit economics visible per workflow, per customer.
- Use IndiaAI compute and Indic models where sovereignty, cost or language fit matter.
- Assign an owner for AI governance — RBI, SEBI, MeitY and DPDP obligations overlap.
- Ask for evals, not buzzwords: how did the candidate define and measure "good"?
- Probe trade-offs: quality vs latency vs cost vs risk, with numbers.
- Look for shipped artefacts — prototypes, agents, write-ups — over certificates alone.
- Check regulatory fluency for BFSI, health and public-sector roles.
06 · The portal
What's coming to AINPM.in
This page is the front door. The portal will grow into open, practical resources — built in public on GitHub.
Eval templates
Reusable rubrics and golden-set formats for multilingual AI features.
Regulation tracker
A living timeline of DPDP, RBI, SEBI and MeitY AI obligations for product teams.
More BFSI & mutual fund case studies
RAG and agentic patterns for AMCs, distributors and fintech platforms, following the SEBI RAG case study.
Agentic PRD
A spec template for agents: tools, autonomy levels, approvals and failure modes.
₹-per-task calculator
Estimate inference cost across models before you commit to an architecture.
Community notes
Write-ups from Indian PMs shipping AI — contributions welcome via GitHub.
About the author
Ian Pinto
Technical Product / Program Manager working on Mutual Fund AMC products and platforms, cloud-native systems, RAG and agentic AI. I started AINPM.in to share what I'm learning about building AI-native products in India — openly and iteratively.
- MBA
- BE (IT)
- PMP
- CSM
- AI Product Management — IIT Kharagpur (upGrad)
- AI Product Management — ISB
- Certified AI & cloud-native engineering