Part two of the ‘What to Build’ series. We did consumer AI first because that was where the anxiety was loudest. We are doing fintech second because that is where the opportunity is least understood.
The “fintech is over” reflex is wrong, and quite badly.
Here is the conventional wisdom in any founder WhatsApp group in April 2026. Payments are commoditised; UPI killed the market. Lending is over-funded and the RBI is choking the consumer book. Neobanks have failed. Insurance is impossible. Wealth tech is a Zerodha and Groww duopoly. The conclusion: fintech is done.
This take is wrong on every clause. It conflates the death of one consumer fintech playbook with the death of fintech itself. The previous wave was about layering one feature (UPI rails, BNPL, P2P lending, low-cost broking) on top of an underdeveloped consumer market. That wave is genuinely tapped out. The next wave is being built on top of an entirely different stack, and almost no one has noticed.
Consider what India shipped in the last twenty four months. The Account Aggregator framework now has more than 110 million linked accounts and consent volumes growing at three percent week on week. The Unified Lending Interface began moving from agricultural pilots into MSME and personal credit, with disbursal times collapsing from four to six weeks down to under ten minutes in early production deployments. The new Digital Personal Data Protection Act, the revised co-lending norms, and 100 percent FDI in insurance all landed in the last eighteen months. India received 137 billion dollars in remittances in 2024, the most of any country in history. The 63 million MSMEs in India still represent a 530 billion dollar credit gap. Twelve million gig workers have less than fifteen percent formal credit penetration. Retail wealth is one third of GDP and the average Indian household still allocates two thirds of net worth to gold and real estate.
Read those numbers slowly. India in 2026 has more usable financial primitives than the United States. It has a larger underserved credit population than any country on earth. It has a diaspora that sends home more money than the FDI book and almost no fintech that serves them as customers rather than as remittance pipes. The “fintech is done” take is just an artefact of having looked at the wrong layer of the stack.
The list below is the layer we think is open. Same principles as the consumer AI piece. Twenty ideas, India-first and globally relevant where the unit economics travel, written for founders who actually want to build, not for decks. Each idea passes a four-part test: a real cohort with budget, a wedge that compounds with use, a why-now that did not exist eighteen months ago, and a non-obvious watch-out. None of these are easy. All are buildable today. We have tried to be specific about who wins.
If you are building one of these, or a sharper version of one of these, come talk to us.
1. The MSME underwriter on GST, AA, and ULI
There are 63 million MSMEs in India. Only 14 to 16 percent have ever received formal credit. The credit gap is approximately 530 billion dollars. The reason is not capital scarcity. The reason is that the marginal cost of underwriting an MSME for a 5 lakh working capital loan was, until recently, higher than the lifetime expected interest income. Banks could not justify it.
ULI broke that equation. With GST returns, bank statements via Account Aggregator, and credit bureau data flowing in real time through a single consent layer, an underwriter can now assess a small business in minutes for a fraction of the previous cost. The infrastructure is there. The product is not.
Build a vertical-specific MSME underwriter that combines the new data sources with proprietary cash flow signals from a specific industry. Start with one vertical (kirana, restaurants, salons, automobile workshops, pharmacies) where you can build a deep pattern library of revenue and stress signals. Lend off your own balance sheet via an NBFC partnership initially, then graduate to co-lending with banks under the new RBI norms.
Why now: ULI plus AA plus GST plus DPDPA is finally a closed loop. Two years ago, the data was either not consented, not standardised, or not real time. All three are solved.
Who wins: a founder pair with one credit person who has actually run a portfolio through a bad cycle, and one technical founder who can build the data pipelines. Not a marketplace founder who underestimated what underwriting actually means.
Watch-outs: do not underwrite at scale before you have lived through one cycle of stress in your chosen vertical. The losses on month 18 will define whether you are a real lender or a vintage-2026 statistic.
2. Vertical embedded credit for B2B software
A B2B SaaS company in India sees the entire transaction history of its customers. A pharmacy management software knows how much each pharmacy bills, what it owes its distributor, and what its working capital cycle looks like. A logistics platform knows which fleet operator has consistent payments coming in next week. None of them currently lend, because lending is hard and they are software companies.
Build an embedded credit infrastructure that lets vertical SaaS companies offer credit to their customers without becoming lenders themselves. The product is a B2B platform. You handle the underwriting using the platform’s data and AA, you handle the regulated entity (NBFC partnership or in-house licence), the SaaS company handles the relationship and the distribution. Revenue split. The SaaS company gets a new monetisation lever. The customer gets credit that actually understands their business. You get scale through the SaaS company’s existing distribution.
Why now: the new RBI co-lending norms make these arrangements far cleaner than they were even a year ago. Vertical SaaS companies are now mature enough (10 to 100 crore ARR) to want a credit revenue stream.
Who wins: a founder with both fintech and B2B SaaS DNA. Pure fintech founders underestimate how hard distribution is. Pure SaaS founders underestimate how hard credit is.
Watch-outs: pick three verticals and go deep. The temptation to be a horizontal embedded credit player kills companies. Stripe’s lending product took a decade to expand; you do not have that runway.
3. Healthcare lending at the point of care
Indian households spend roughly 50 percent of healthcare costs out of pocket, the highest share among large economies. A hospitalisation or major procedure routinely wipes out savings or pushes families into informal debt. Hospital tie-ups with NBFCs exist but are clunky, slow, and limited to chains. The point-of-care moment, where a family is being told they need to pay 2 lakh in the next 24 hours, is one of the most acute willingness-to-pay moments in the entire Indian economy and almost no fintech serves it well.
Build a point-of-care lending product that lives inside hospitals, diagnostic chains, and IVF centres. Approval in under five minutes using AA. Repayment plans that align with cash flow rather than calendar months. A back-end that integrates into hospital billing software so the loan is invisible to the patient until the conversation. Credit life insurance bundled.
Why now: every major hospital chain in India has gone digital with billing in the last two years. Account Aggregator coverage of the salaried middle class crossed a usable threshold in 2025. Together they make in-the-moment lending operationally viable.
Who wins: a founder pair who can actually sign hospital chains. This is half product, half enterprise sales. Without the relationships, the product never reaches the patient.
Watch-outs: this is a category where collections are the entire business. A patient who took a loan for cancer treatment is a different collections psychology from a personal loan default. Build the empathy into the recovery process from day one or you end up on the wrong end of a Mint expose.
4. The study abroad financing product
One million Indians apply to study abroad every year. The average US graduate program costs 60 to 80 lakh rupees. The current education loan market is dominated by HDFC Credila, Avanse, and Auxilo, products built for a more analog era, requiring co-applicants, collateral, weeks of paperwork, and rigid disbursal schedules. The market is begging for a digital-first product.
Build an education loan product designed entirely around the student journey. Pre-approval at the application stage based on the student’s profile and target school. Co-applicant flow that uses AA rather than physical paperwork. Disbursal directly to the university. Tuition paid in dollars at preferential rates through the cross-border layer. Optional living-cost top-ups. A repayment structure that defers principal until graduation plus six months. The full product is the financial companion across the eighteen-month admission-to-arrival journey.
Why now: PA-CB licences from RBI now make legitimate cross-border tuition disbursal possible without the friction of the previous correspondent banking flow. The Indian middle class is sending students abroad at unprecedented rates, and the willingness to pay for a clean financial product is high.
Who wins: a founder with strong credit DNA paired with someone who has either gone through the process themselves or worked at one of the existing lenders.
Watch-outs: the political environment in destination countries (US visa rules, UK student work rights) materially affects default rates. Build the model with a real understanding of how cohort default behaves under macro stress, not under steady-state assumptions.
5. Working capital for Indian exporters
India’s services exports crossed 350 billion dollars in 2024. Goods exports added another 450 billion. Roughly 200,000 small Indian exporters are sitting in a structural cash crunch: they ship product or deliver services, get paid in 30 to 90 days, and need bridge capital to fulfil the next order. The current options are restrictive bill discounting from banks, slow LCs, or expensive private working capital. Wise and Skydo solved the inbound payment leg. The financing leg is open.
Build a working capital product for the Indian exporter. Underwrite the receivable using verified buyer data and the export documentation. Finance against the verified invoice in 24 hours. Settle in INR or hold in USD as the exporter prefers. Recover from the inbound payment when it lands. The product is invisible if done right. The exporter ships, draws, and repays as cash flows in.
Why now: PA-CB licences and Skydo, Payoneer, Wise, and the new RBI cross-border framework have collectively opened up the data layer required to underwrite an Indian exporter. Platform-based exporters (Amazon Global, Etsy, Upwork, Toptal) have full transactional visibility that did not exist five years ago.
Who wins: a founder with trade finance experience or a deep payments operator. This is not a generalist consumer fintech play.
Watch-outs: forex risk and counterparty risk are real and unforgiving. A few large bad debts can sink the book. The team that takes risk management seriously wins. The team that treats this as a software arbitrage does not.
6. AI-native, humane debt collections
The single ugliest part of Indian fintech in 2024 was collections. Aggressive call centres, public shaming on social media, harassment of family members, occasional violence. The RBI cracked down hard in 2024 and 2025. Most lenders are now scrambling to clean up their collections function while maintaining recovery rates. The category is broken and the regulator is watching.
Build an AI-native collections product that works at scale and behaves with dignity. Voice agents that genuinely listen, understand a borrower’s situation, and offer realistic restructuring. Personalised payment plans generated in real time based on cash flow patterns from AA. Multilingual outreach that respects regional norms. Escalation flows that are calibrated to financial stress, not to recovery KPIs. Sell as a SaaS plus revenue share to lenders.
Why now: voice LLMs in Indian languages crossed a usable bar in 2025. The regulatory cost of bad collections jumped sharply. Lenders are actively shopping for solutions.
Who wins: a founder who has either built a collections function inside a lender or is a domain operator who has seen the bad version up close. This cannot be built by people who think collections is a routing problem.
Watch-outs: do not over-promise on recovery rates. The honest pitch is that you maintain or marginally improve recovery while sharply reducing complaints, regulatory risk, and reputational damage. That is a real product. A product that promises higher recoveries through pressure is the old playbook in a new wrapper.
7. The next-generation credit bureau
CIBIL, Experian, Equifax, and CRIF dominate the Indian bureau market. Their data is bank-centric, lagging, and increasingly inadequate for the new credit cohorts: gig workers, new-to-credit borrowers, exporters, MSMEs with cash-heavy operations. The RBI’s tightening of unsecured retail lending in late 2023 exposed how thin the existing scoring models were when stressed. Lenders are paying for bureau pulls but underwriting on a parallel set of alt data they have hacked together themselves.
Build the next-generation bureau as a product, not as a regulatory body. Combine traditional bureau data with AA cash flow patterns, GST returns, platform earnings (Ola, Uber, Swiggy, Zomato, Meesho, Amazon, Upwork), telco signals, and verified employer data. Sell to lenders as an underwriting layer. The output is not a single score but a structured risk vector with explainability. The compounding moat is data.
Why now: AA volumes crossed a usable threshold in 2025. Multiple alt-data sources are now consented and clean. The bureaus have been slow to integrate them. The window is now.
Who wins: a founder with deep credit DNA paired with strong data engineering. Probably someone who has worked inside CIBIL or a major NBFC and seen the gaps from the inside.
Watch-outs: this is a regulated category and the existing bureaus will lobby aggressively. Build with a clear regulatory thesis, possibly via the existing CIC framework, and engage with the RBI early rather than late.
8. The financial OS for India’s gig workers
Twelve million Indians drive for Ola and Uber, deliver for Swiggy, Zomato, Blinkit, and Zepto, or run shifts for UrbanCompany. Less than fifteen percent have access to formal credit. Forty percent earn below 15,000 rupees a month. They are the most underserved consumer financial cohort in the country. KarmaLife and a handful of others have made a start, but the category is wide open.
Build a full financial OS for the gig worker. A neobank-style account that pulls earnings from multiple platforms. Earnings-linked credit that adjusts in real time. Health and accident insurance bundled at thin premiums. Auto-savings into a micro-SIP linked to busy days. Term life for the worker’s family. Emergency credit that disburses in fifteen minutes when a medical or vehicle emergency hits. Voice-first support in regional languages. Pricing simple, transparent, free at the base tier.
Why now: India Stack components (AA, OCEN, ULI, eKYC) plus platform API access plus voice LLMs in Indian languages plus the new gig worker welfare framework introduced in the 2026 Budget all combine for the first time.
Who wins: a founder who has lived alongside this cohort, not someone optimising on a TAM slide. The product trust is built by going to driver canteens, not corporate offices.
Watch-outs: the platforms (Ola, Swiggy) will sometimes try to build this themselves. The right answer is to be the worker-side product, with the platforms as data partners. Picking sides between the worker and the platform is the most important strategic choice in this category.
9. The wealth coach for the UPI generation
A generation of Indians born after 1995 has grown up with UPI, Zerodha, Groww, and SIPs. They are saving and investing earlier than any previous generation. They are also making consistent, predictable mistakes: over-allocation to direct equities they do not understand, under-allocation to tax-advantaged products, near-zero allocation to insurance, no estate planning, no goal alignment. Zerodha and Groww built the rails. They did not build the coach.
Build a personal wealth coach for the salaried 25 to 40 cohort. Onboard via AA so you see the full picture of bank balances, mutual funds, stocks, EPF, and credit. Give honest advice, not product pushes. Optimise tax with a real understanding of the user’s bracket and instruments. Run goal-based planning for marriage, home, and children with real probabilistic models. Recommend term life and health insurance as the first product, not the last. Charge a flat fee, not a commission. Build trust by being the rare honest player in the category.
Why now: AA full-coverage, the SEBI investment advisor framework, and the maturity of direct mutual fund and ETF infrastructure together make a fee-only AI advisor feasible at retail prices.
Who wins: a founder who understands both the regulatory grain (SEBI RIA) and the product grain (consumer fintech). The credibility of the voice is the moat.
Watch-outs: do not optimise for AUM growth. Optimise for retention and Net Promoter. The wealth coach business compounds over decades. The team that thinks in years compounds. The team that thinks in quarters churns.
10. Wealth and decumulation for Indian retirees
India has 150 million people over the age of sixty and growing fast. Average household financial assets at retirement run between 25 and 75 lakh for the urban middle class. The product set serving this cohort is brutally inadequate: bank fixed deposits, postal savings, a handful of senior citizen schemes, and an LIC annuity book that is mispriced. The right product, decumulation planning that turns a lump sum into a multi-decade income with care for inflation, healthcare costs, and longevity, simply does not exist at scale in India.
Build a retiree wealth product. The first conversation is not a portfolio question; it is “how do you want to live for the next 25 years?” The product translates that into a structured income plan, allocates across instruments (bonds, debt funds, REITs, annuities, equity), layers in healthcare and long-term care planning, and maintains it. Charge a flat annual fee. Pay relationship managers to do quarterly check-ins, especially for users with no adult child managing their finances.
Why now: 100 percent FDI in insurance opened up annuity innovation. Bond and REIT retail availability has matured. The diaspora children of Indian retirees are willing to pay for their parents’ financial care.
Who wins: a founder with deep wealth advisory experience plus genuine empathy for an older Indian user. Most wealthtech founders are 28 and building for themselves. This product needs the opposite.
Watch-outs: do not let the children become the buyer and the parent become an afterthought. The product has to delight the seventy-year-old user. If the seventy-year-old does not log in, the product has failed regardless of who is paying.
11. The retail bond and private credit platform
Indian retail investors hold roughly 60 lakh crore in fixed deposits and another 40 lakh crore in small savings. The post-tax return is poor. The bond market is largely institutional. SEBI opened up retail access to corporate bonds in 2024 and to a wider private credit set in 2025. Wint Wealth, GoldenPi, and Tap Invest have started, but the category is still under-built.
Build the retail bond and private credit platform that India’s wealth-accumulating middle class deserves. Curate a clean shelf of corporate bonds, government securities, REITs, InvITs, and accredited private credit deals. Offer fractional access where the regulator permits. Provide credit ratings, default histories, and stress test outputs in plain language. Auto-allocate ladders for FD-style users who want a 7 to 10 percent post-tax yield without the lockup. Pricing flat, never commission.
Why now: SEBI’s revised framework on online bond platforms made retail access cleaner. The AA-driven income proofing for accredited investors is now operational. Distribution can finally scale without the broker call centre model.
Who wins: a founder with a real fixed income background plus consumer fintech distribution chops. This is not a category where you can fake the credit work.
Watch-outs: when the credit cycle turns, retail will get hit with defaults they did not understand. Your job is to over-disclose and to choose your shelf carefully. The first big retail default that lands on a platform without proper risk communication will set the category back five years.
12. The wealth product for the Indian SMB owner
The 1.5 to 2 million Indian SMB owners running businesses with 5 to 50 crore in annual revenue are uniquely under-served. Their wealth lives largely in business equity, real estate, and gold. Their financial advisors are the family CA, who optimises for tax compliance, not wealth creation. They are too small for a private banker and too big for a Groww account.
Build a wealth product specifically for the SMB owner. Personal balance sheet that integrates business equity, household assets, and liabilities. Tax planning that optimises across personal, business, and family. Succession and inheritance structuring (HUF, LLP, family office light). Portfolio allocation that recognises the concentration risk in their business and counterbalances. A service tier with a real human relationship for the moments that matter (acquisitions, exits, divorce, disputes).
Why now: AA, GST, and corporate filings make a unified wealth view possible for the first time. Demographic transition, the second generation taking over the business, has created a willingness to professionalise.
Who wins: a founder who has been a wealth advisor in a private bank or has come out of a CA practice that served this exact cohort. Credibility is the product.
Watch-outs: the buyer makes decisions slowly and emotionally. Do not over-engineer the onboarding. The first three meetings are about trust, not features. Build the product to be patient.
13. The full-stack NRI bank
India received 137 billion dollars in remittances in 2024, the largest remittance flow to any country in history. The Indian diaspora numbers 35 million, including 16 million NRIs. Most of them bank in their country of residence and remit to India. None of them are well served by either side. Indian banks treat NRIs as a low-touch deposit base. Foreign banks treat them as a marketing segment. Nobody has built the product the actual NRI wants: a single financial home across two countries.
Build a full-stack NRI bank. Multi-currency accounts (USD, GBP, AED, SGD, INR) with FX at near interbank rates. NRO and NRE seamlessly managed. Mutual fund and PMS investing in India with KYC, FATCA, and PFIC handled in software. Real estate investment with end-to-end legal and registration. Tax filing in both jurisdictions. Estate and inheritance planning across borders. Concierge for the parent in India who needs help with anything (from a hospitalisation to a property dispute).
Why now: the diaspora is wealthier, older, and more willing to pay for service than at any prior point. RBI’s revised NRI account rules and the new tax framework on foreign remittances both landed in 2025, opening up product space.
Who wins: a founder who is themselves a sophisticated NRI or is married to one. The product nuances are buried in lived experience.
Watch-outs: regulation in two jurisdictions is harder than founders expect. Do not start with twenty geographies. Pick US-India or UAE-India and own that corridor before expanding. Each corridor is a different product.
14. Cross-border payments for Indian SMBs
PA-CB authorisations from RBI in 2025 and 2026, granted to Wise, Payoneer, Skydo, and a handful of others, opened up the legitimate cross-border payments market for Indian businesses. Skydo has shown what is possible: tens of thousands of Indian service exporters on the platform, flat-rate pricing, zero forex markup. The category is no longer regulatory blocked. It is now a product and distribution race.
Build a cross-border payments product for a specific cohort that the current players underserve. Indian e-commerce sellers exporting on Amazon Global. Indian agencies serving global clients on retainer. Indian SaaS companies billing in dollars but operating in India. Indian creators monetising on YouTube and Substack. Each cohort has a slightly different set of needs around invoicing, recurring payments, and currency hedging. Pick one. Build the deepest product for that cohort. Expand later.
Why now: PA-CB regime is operational. Payment volumes are growing. The previous semi-legal corridors are shutting down, pushing volume onto legitimate rails.
Who wins: a founder with both payments operations DNA and a real understanding of one specific exporter cohort. Generalists lose.
Watch-outs: this is a high-volume, low-margin business. Unit economics matter from day one. A founder who plans to subsidise growth with venture capital will hit a wall when a more disciplined competitor underprices them.
15. Embedded insurance in commerce flows
The IRDAI’s 2025 framework and the rollout of API-driven insurance distribution have made embedded insurance commercially viable. PwC India estimates the embedded insurance market could exceed 2 billion dollars by 2026 and protect 100 million gig and mobility users in that period. Insurance bundled into a Swiggy delivery, an Ola ride, a Cleartrip flight booking, an Amazon order is now a real distribution channel.
Build an embedded insurance infrastructure that lets any consumer platform offer relevant insurance at the right moment in the customer journey. Mobility flows, e-commerce flows, travel flows, and utility flows each have different relevant covers. The product is a B2B platform that handles underwriting, regulatory compliance, claims processing, and fraud detection, while the consumer platform handles distribution. Revenue share with the platform.
Why now: IRDAI’s regulatory updates, the maturity of API-led insurance product design, and consumer comfort with thin, single-event insurance products all converged in the last 24 months.
Who wins: a founder pair with insurance DNA and platform partnership chops. This is a B2B sale to consumer platforms, then a regulated insurance operation underneath. Both halves are hard.
Watch-outs: claims experience is the make-or-break. A platform partner whose customers have a bad claims experience will turn off the integration in a quarter. Build the claims layer with the same rigour as the underwriting layer.
16. Parametric crop and weather insurance
Roughly 85 million Indian farmers depend on agriculture. Climate volatility has made the last five monsoons increasingly variable. Traditional crop insurance, primarily PMFBY, is plagued by slow claims, dispute, and corruption. Parametric insurance pays out automatically based on triggers (rainfall, temperature, satellite-derived crop health) without the slow human-driven claim assessment. The technology is finally cheap enough to deploy at India scale.
Build a parametric insurance product for Indian farmers. Use satellite data, weather station inputs, and on-ground IoT (where available) to define triggers per crop and region. Distribute through agri-input retailers, FPOs, and rural NBFCs, the points where the farmer already has a financial relationship. Settle claims directly to bank accounts via UPI or AePS. Bundle with crop loans for retention.
Why now: Bajaj’s ClimateSafe and a handful of pilots demonstrated the technical and operational viability in 2025. Satellite data costs have collapsed. IRDAI is encouraging the category. Climate risk is rising.
Who wins: a founder with deep rural distribution DNA paired with insurance and remote sensing capability. The hardest part is rural distribution. The technology is the easier half.
Watch-outs: parametric basis risk (the gap between the parameter and the actual loss) is real. Educate the farmer up front. A product that pays out when the satellite says “drought” but the farmer’s specific field had rainfall is a credibility disaster. Build the trust by paying out fairly even at the edges.
17. Insurance for chronic disease cohorts
Roughly 77 million Indians have type 2 diabetes. 200 million have hypertension. Tens of millions live with PCOS, cardiovascular conditions, asthma, and other chronic illnesses. Standard health insurance treats these as risks to price out. The right product treats them as cohorts to manage actively, sharing the upside of better management with the patient.
Build a chronic-disease-first health insurance product. Continuous monitoring through CGMs, BP cuffs, and connected devices. A care team (nutritionist, coach, doctor) included. Premium discounts for verified clinical improvement (HbA1c down, BP under control, weight in range). Claims paid via cashless network. Integration with the AI health concierge from the consumer AI list.
Why now: 100 percent FDI in insurance, IRDAI’s openness to outcome-based health products, and the dramatic drop in CGM and connected health device prices over the last 24 months together make this category technically and commercially viable.
Who wins: a founder pair with one healthcare insider (preferably a doctor who has seen the chronic care population in scale) and one insurance operator who can stand up the regulated entity.
Watch-outs: the unit economics are tight at the start. You will lose money on the first cohort. The bet is that better-managed members generate dramatically lower claims after year two. If you do not have the conviction and the capital to ride that, do not build this.
18. AI tooling for chartered accountants
India has roughly 400,000 practising chartered accountants. The CA serving the 5 lakh small business owners and the 15 million salaried filers spends most of their time on data entry, reconciliations, GST, TDS, and form filing. The actual advisory work that justifies their fees is squeezed by the operational load. The right product is not consumer accounting software. It is the picks-and-shovels tool that makes the CA dramatically more productive at scale.
Build an AI-native CA practice management product. Auto-import bank, GST, and platform data via AA and OCEN. Auto-categorise transactions with pattern learning. Auto-prepare and auto-file ITRs, GSTs, TDS returns, and ROC filings, with the CA reviewing and signing. Built-in client communication that drafts the right reminder at the right time. Pricing per CA seat, plus per filing. Distribution through ICAI bodies and regional CA networks.
Why now: AA and the GST API combined with capable LLMs make end-to-end auto-preparation viable. The CA is overwhelmed by compliance load and is genuinely shopping for tools.
Who wins: a founder with either a CA background or deep experience in a regulated B2B tooling category. The trust of the practitioner is the product.
Watch-outs: do not try to bypass the CA. The dream of “consumer self-filing replaces the CA” has failed for thirty years in India for good reason. The CA is a trusted relationship. Make them more powerful and you have a customer for life.
19. The AI compliance product for fintechs
Indian fintechs spend a stunning percentage of engineering and ops time on compliance. RBI guidelines change frequently, IRDAI now ships material updates every quarter, SEBI moves on its own cycle, and DPDPA layers a privacy regime on top. Most fintechs run compliance as a manual operation with a spreadsheet and a worried compliance officer. The cost of a bad call is regulatory action, which can mean a 30-day pause or a permanent shutdown.
Build an AI compliance product for the fintech sector. Continuous monitoring of regulatory updates with auto-mapping to a fintech’s products. Pre-built control libraries for KYC, AML, transaction monitoring, customer disclosures, and grievance redressal. Audit-trail-ready evidence packages. Voice or text agent that the compliance officer can ask in plain English. Pricing per regulated entity per month.
Why now: the regulatory cadence has accelerated. The compliance burden has become a real engineering and finance line item. LLMs are finally good enough to map regulatory text to product controls reliably.
Who wins: a founder who has been a compliance head inside a regulated fintech, paired with strong product engineering. Domain depth is non-negotiable.
Watch-outs: do not over-promise on automation of regulatory judgement. The product augments the compliance officer; it does not replace them. Sell that frame. The fintech that thinks they bought a replacement and gets fined will be vocal.
20. The corporate card for India’s SMBs
Razorpay X, Cred Escrow, Open, and a handful of others are building corporate cards and spend management for the Indian startup. Most stop at the funded startup tier. The 1.5 million Indian SMBs running 5 to 50 crore businesses, almost all of them bootstrapped or family-owned, are a different segment with no real product.
Build a corporate card and spend management product for the bootstrapped SMB. Underwrite based on GST, AA, and bank statements rather than on equity funding. Issue cards with line discipline (per category, per vendor, per employee). Auto-reconcile against GST input tax credit. Built-in vendor and employee reimbursement workflows. Bundle with payment gateway and working capital. Pricing based on transaction volume, not subscription.
Why now: GST input credit auto-reconciliation needs the GST API maturity that landed in 2025. AA and corporate filings allow underwriting without equity collateral. The SMB market has digitised payment behaviour materially in the last 24 months.
Who wins: a founder with deep SMB distribution chops, ideally from a payments or accounting background. This category lives or dies on the ability to acquire SMBs at low cost.
Watch-outs: the larger banks and the existing fintech leaders will compete hard once the segment proves out. Your edge is the depth of fit for the bootstrapped SMB and the efficiency of acquisition. Premium pricing or premium positioning will not work in this segment. Build cheap, build accurate, build trusted.
Picking one
Twenty ideas is a menu, not a strategy. Here is how we would think about narrowing if we were sitting across from a founder next Tuesday.
First, fintech is a regulated category and the regulator is a co-author of every product. The teams that win in fintech are the teams that go to the regulator early, listen carefully, and design within the lines. The teams that try to operate first and ask permission later have a one to two year window before they hit a wall they did not see coming. Pick a category where you understand the regulatory thesis cold.
Second, the unit economics are different in fintech. There is no growth-at-all-costs path. A fintech that loses money on every loan, every premium, every transaction does not turn the corner with scale; it turns into a bigger losing fintech. We wrote in January that companies which controlled burn and proved unit economics raised cleanly in 2025. That is twice as true in fintech. Build a model that is profitable at the loan or policy level on day one. Subsidise distribution if you must. Never subsidise risk.
Third, distribution is the moat in Indian fintech. The product layer is being commoditised. The data layer is being democratised. What is left is who acquires customers cheaply and retains them honestly. The categories above all have a specific distribution advantage attached: a vertical SaaS partnership, a hospital chain, a CA network, a retiree relationship. Pick yours up front. Without one, you are buying customers from Google and Meta at a unit cost that will kill you.
Fourth, fintech rewards founders who think in decades. The wealth coach, the chronic disease insurer, the NRI bank, the retiree wealth product are all multi-decade businesses with compounding trust as the asset. The crypto-era playbook of “ship fast, raise fast, exit fast” is a fintech graveyard. The team that can hold the line for ten years wins. The team that needs an exit in three should pick a different sector.
Finally, India in 2026 is not the India of 2018 in fintech terms. The infrastructure is genuinely better. The regulator is genuinely more thoughtful. The customer is genuinely more aware. The cliche of “this could only happen in India” used to be a cope. It is now the actual differentiator. Build accordingly.
We will follow up with what to build in vertical AI SaaS next, then AI infra. If you are building one of the twenty ideas above, or a sharper version of one, we want to hear from you.
A note on intent: this is a thought piece, not an investment thesis. We write to surface ideas worth thinking about and to start conversations with builders.
