We're building ADJST to rebuild claims for the AI era — starting with UK motor — with a small number of founding MGA and carrier partners. Not as customers of a finished product, but as the people who shape it, own the roadmap around their book, and get first-mover advantage on the biggest change coming to claims.
Claims is the largest cost on your book, and whichever way you run it today the technology is working against you. Outsource it and you lose sight of it — a monthly bordereau instead of a live view, no way to intervene while a claim is going wrong, and quarters to change anything. Keep it in-house and your experts spend their day in twenty-year-old software, doing work that should have been automated a decade ago. Either way, inconsistent reserving and quantum let 5–10% of indemnity leak away. And Consumer Duty keeps raising the bar on fair value, explainability and vulnerable-customer care — which legacy stacks, in-house or outsourced, struggle to evidence.
AI-native claims is coming to the whole industry — someone needs to lead it. Founding partners are those people: they shape the product around their book and become the case study others point to, known for it inside their own business and across the market.
These are target economics — modelled from published benchmarks (ABI; McKinsey/Accenture) and our own analysis, to be proven with founding partners rather than measured yet. The gains come from the platform and the operating model, so they hold however you run claims.
Plus capital release from reserve precision — a balance-sheet effect.
ABI 2025 average claim · McKinsey/Accenture leakage benchmarks · ADJST analysis · illustrative, validated with design partners
On the SaaS tiers — what we're recruiting for now — these gains land inside your own operation: your handling cost, your indemnity, your P&L, for a platform subscription. Nothing is shared; the savings are simply yours.
Where we perform the handling, the commercial model adds per-claim fees and a savings-share: the fee applies only to recovered leakage — fraud, supplier overcharging, quantum cushioning — never to underpaying a valid claim, and never where a complaint, a vulnerable customer, FOS or litigation, a below-protocol settlement, or a fair-outcome flag is in play. Against the legacy TPA model, our target is a +15ppts EBITDA uplift on this comparison.
Being a design partner doesn't mean handing over your claims. ADJST is one platform with a service catalogue on top, and you choose the setting — from running it yourself with the models tuned to your book, through our specialists on the complex work or surge capacity, to a full outsource.
Our founding partners start on the Managed SaaS tier — MGAs and carriers who keep their own claims team, with our specialists tuning the models to their book. Their book is what the calibration is built on.
See the full service catalogue →Not fully — the design is well advanced, we have a working preview, and we've started building the foundational platform. The claims system of record goes live H2 2026, the full AI-native platform in H1 2027, with calibration deepening after. Founding partners start on the system of record and layer the AI in as scale justifies it — you're early on purpose, which is why you get the terms and influence you do.
The platform — AI specialists, always-on checks, Day-0 reserves, live analytics, the modern workbench — running your claims operation, with our team evolving it around you. Tuning the models to your book happens on top of that, and founding partners' books are the ones it's tuned on first.
No. The SaaS tiers — what we're recruiting for now — keep your own team handling the claims, on our platform. The service tiers exist if you want them later; that's the whole idea of a service catalogue.
It isn't, and the architecture is the evidence: moving to another tier requires no migration, and that is true in both directions. If you build out your own specialist bench and no longer need ours, you step back down a tier and keep the platform. We'd rather earn a tier every year than lock you into it.
UK motor first, then property (starting with escape of water) and casualty.
DCA approval will be started once seed capital is raised. No DCA approval is required for the platform offering.
Tenant scope, row-level confidentiality and field-level PII masking are enforced in the engine, not bolted on — with UK data residency.
Calibration is tenant-scoped: your models are tuned on your data, and your book's specifics don't leak into anyone else's. What does travel is generalised model improvement — the same way any vendor gets better at its job. This boundary is contractual.
A design-partner agreement and an LOI for a motor book. We start small and prove it, claim by claim.
We're recruiting founding partners for the first release, including MGAs and carriers who only ever want the platform.