What actually changed in January 2025, and why it matters to product teams
The IRDAI (Regulatory Sandbox) Regulations 2025 were notified in early January 2025, repealing the 2019 regulations that had run the first cohorts. The headline shift is scope. The 2019 framework limited the sandbox to specified categories, mostly insurance solicitation, distribution, products, underwriting, policy servicing and other functions named by the Authority. The 2025 regulations replace that closed list with an open one: an applicant may seek relaxation from any provision of any regulation, notification, master circular, guideline or circular issued by IRDAI, for innovation anywhere across the insurance value chain in India.
The only carve-out is prudential. The sandbox cannot relax requirements touching capital, liquidity, investment, solvency and reserving, plus any other area the Authority adds later. Everything else is, in principle, on the table: product design, rating, wording, distribution, claims, even the use of new data sources or settlement mechanisms.
For a product team this is the difference between asking permission to tweak a process and asking permission to run a genuinely new commercial cover live with real policyholders. A parametric flood trigger that does not fit the indemnity logic of existing fire wordings, an AI-priced cyber cover that uses external telemetry, or a blockchain-verified cargo cover that settles on confirmed sensor events: each previously sat in a grey zone. The 2025 sandbox gives them a defined door.
The timing matters because the sandbox has already proven appetite. The first two cohorts drew well over three hundred applications, which tells you the market wants supervised room to experiment rather than a binary approve-or-reject filing. The widened scope channels that demand toward harder, more original product ideas rather than incremental process tweaks. For commercial lines specifically, where the interesting problems are around speciality risks, data-driven pricing and faster settlement, the open scope is the most consequential regulatory change of the cycle.
The inter-regulatory provision is the quietly significant part
Buried in the 2025 regulations is an enabling clause for proposals that span more than one financial sector. Innovation in commercial insurance rarely stays inside one regulator's perimeter. A parametric crop cover bundled into a farm loan touches RBI's lending rules. An embedded cyber cover sold through a payments app touches the payment-system framework. A surety or trade-credit construct linked to a capital-markets instrument touches SEBI. The 2025 regulations recognise this and provide that the process for sandbox applications cutting across sectors will follow what the competent authority specifies, opening the path to coordinated testing with RBI, SEBI and PFRDA.
This matters because the most interesting commercial products of the next few years are bundles, not standalone policies. Embedded covers on lending and payment rails, parametric triggers tied to commodity indices, and warranty constructs sitting inside e-commerce flows all need more than one regulator comfortable with the design.
The practical caveat: the detailed procedure, timelines and the mechanics of inter-regulatory coordination were not fully specified in the regulations themselves. IRDAI signalled that these would come through a master circular. Until that circular lands, treat the inter-regulatory route as available in principle but operationally immature. If your product is single-sector, file it now. If it is genuinely cross-sector, sequence your insurance application first and keep the cross-regulator angle ready to activate.
One more thing to watch: several procedural details, including final timelines that the 2019 regulations had specified but the 2025 text omitted, were left to be operationalised by that same master circular. Confirm the current procedural position with IRDAI before you commit a launch calendar, because the published text alone will not tell you how long a cohort approval now takes end to end.
Who can apply, and the eligibility gate brokers underestimate
Eligibility under the 2025 regulations is broad. Any person other than an individual can apply, provided it held a minimum net worth in the range mandated by the regulations (set at a modest floor, around ten lakh rupees of net worth in the previous financial year) and meets fit-and-proper and data-protection conditions. That floor is deliberately low so that insurtechs, brokers, web aggregators, intermediaries and technology firms, not only insurers, can bring proposals.
This is the structural opening for brokers. Under the old reading, a broker waited for an insurer to design a product and then distributed it. The 2025 sandbox lets a broker or a broker-backed insurtech be the named applicant, or co-applicant, on a novel cover. A broker who knows a client segment cold (say, rooftop-solar EPC contractors, or cold-chain logistics operators) can design the trigger logic, bring the loss data, and partner an insurer for capacity and the balance sheet.
The gate brokers underestimate is not net worth. It is the obligation set that comes with a live experiment: policyholder protection, system integrity, data security, a defined experiment population, clear consumer disclosures and an exit plan for when the cohort ends. You must show how genuine policyholders are protected if the product underperforms, how their data is handled, and how cover runs off cleanly.
- A named applicant with the requisite net worth and clean compliance record.
- An insurer partner carrying the underwriting risk (the sandbox relaxes conduct rules, not solvency).
- A bounded experiment: defined geography, segment, sum-insured caps and policyholder count.
- A documented underwriting basis and pricing logic the Authority can interrogate.
- A policyholder-protection and exit plan for when the experiment period closes.
Mapping the route: sandbox versus use-and-file
Brokers and insurers conflate two different doors, and choosing wrong wastes months. The ordinary route for a new commercial product is the product-filing process. Under the prevailing regime most general-insurance products move on a use-and-file basis: the insurer's product management committee signs off, the product launches, and the filing follows. That route is fast and needs no IRDAI pre-approval, but it assumes the product fits inside existing regulatory boundaries. See our note on the use-and-file route and its broker impact for how that works in practice.
The sandbox is for the product that does not fit. You go to the sandbox when the design needs relaxation from an actual rule: a settlement mechanism that pays on a parametric index rather than assessed loss, a rating approach that uses a data source the standard framework does not contemplate, or a distribution model outside the normal intermediary structure. If your product can launch under use-and-file, do that. Do not use the sandbox to test something you are already allowed to do.
A simple decision test
- Does the product require relaxation from a specific IRDAI provision? If no, use-and-file.
- Is the relaxation in a prudential area (capital, solvency, reserving, investment, liquidity)? If yes, the sandbox cannot help; rethink the design.
- Does the design cut across RBI, SEBI or PFRDA? If yes, flag the inter-regulatory route and expect a longer runway.
- Can you bound the experiment to a defined population with policyholder safeguards? If yes, you have a fileable sandbox proposal.
The discipline of this test alone prevents the most common failure: insurers lobbing half-formed ideas at the sandbox that either belonged in use-and-file or breached the prudential carve-out.
Designing the experiment so it survives the cohort
A sandbox permission is time-bound. You get an experiment period (extendable once at the Authority's discretion), after which the product either graduates to a full filing, gets modified, or winds down. Design for that lifecycle from day one, because a product that cannot survive graduation is a product that should not enter the sandbox.
Start with a tight cohort definition. Pick one segment, one geography, and hard caps on aggregate exposure. A parametric heat-stress cover for textile units in one cluster, with a per-life and per-employer sum-insured ceiling, is fileable. The same cover offered nationally with open limits is not a controlled experiment; it is an uncontrolled launch wearing a sandbox label.
Build the data spine before you apply. The whole point of a parametric or AI-priced cover is that the premium and the trigger rest on data. The Authority will want to see the index source, its independence, the historical basis for calibration, and the settlement logic. If you cannot evidence the data, the experiment cannot be evaluated, and you will not be able to graduate it later either.
Think about basis risk and disclosure up front. Parametric covers pay on a trigger, not on actual loss, so a policyholder can suffer real damage and still receive nothing if the index does not breach. That gap must be disclosed in plain language and stress-tested. A sandbox approval will not protect an insurer from a conduct complaint if the basis risk was buried in the fine print.
Finally, draft the policy wording and the endorsement structure as if you were already filing the graduated product. Wording you can defend at graduation is wording you should be selling in the experiment. Treating the sandbox wording as a throwaway draft is how good ideas die at the cohort boundary.
Three commercial covers that fit the 2025 sandbox
Abstract eligibility is less useful than concrete candidates. Three product families are natural fits for the widened scope, and brokers sit close to the data for each.
Parametric covers tied to physical triggers
Flood depth, rainfall deficit, wind speed, solar irradiance and heat-stress indices all support covers that pay fast on a measured trigger. These suit commercial property, agriculture-linked corporates and outdoor workforces, and they relieve the dispute-heavy loss-assessment process that drags conventional claims. Our pieces on corporate parametric uptake and supply-chain trigger design go deeper on calibration. The sandbox is the right door because parametric settlement departs from the indemnity logic baked into standard wordings.
AI-priced cyber and liability covers using external telemetry
A cyber cover that prices off continuous external scanning of an insured's attack surface, or a liability cover that adjusts terms on real-time exposure signals, uses data sources and rating mechanics the standard framework did not anticipate. The sandbox lets an insurer run this live with a bounded cohort while keeping the model explainable to the regulator.
Blockchain-verified cargo and transit covers
A cover that settles automatically on a verified sensor or shipment event removes manual claim-triggering friction for high-frequency logistics flows. The relaxation sought is around the settlement and verification mechanism, not the prudential basis, which puts it squarely inside sandbox scope.
In each case the broker's edge is the same: proximity to the client segment, ownership of the loss data, and the ability to co-design the trigger with an insurer rather than wait to be handed a finished product.
How brokers co-design rather than wait
The strategic shift the 2025 regulations enable is that brokers move from distributors of finished products to co-designers of new ones. That is a different operating posture, and a few habits make it work.
First, treat your book as a research asset. The covers worth sandboxing are the ones where you already see unmet need and have the loss history to prove it. A broker with five years of claims data on a niche, say small hydropower operators or speciality chemical units, holds exactly what an insurer's product team lacks. Package that data into a proposal rather than a sales pitch.
Second, bring the insurer in as a capacity and balance-sheet partner early, and be explicit about the split: you own the segment insight and distribution, they own the underwriting and the prudential obligations the sandbox does not relax. Reinsurance support, including GIC Re's appetite for the novel risk, should be lined up before filing, because an insurer will not commit capacity to a parametric or AI-priced experiment without treaty comfort.
Third, build the compliance and conduct story in parallel with the product, not after. The data-protection obligations, policyholder disclosures and exit plan are not paperwork to bolt on at the end; they are part of what gets evaluated. Our overview of insurtech regulatory compliance challenges covers the common gaps.
For the distribution layer once a product graduates, align early with the Bima Sugam operating model so the cover has rails to scale on, and check the design against the 2024 insurance products regulations that govern the graduated filing.