Post to StrataLens
Send the JSON your system already produces to an address we give you, with a key in the header. You get an immediate receipt and the decision follows. Repeats are recognised and processed once.
Where StrataLens fits in your business
Six use cases, written for the people who own the work: underwriting operations, delegated authority, broking operations, and the risk teams who have to sign it off. Each one says what arrives, what your underwriters set themselves, what goes out, and how it connects.
- 01Submission triage for carriers and MGAs
- 02Delegated authority bordereaux
- 03One broker feed, many carriers
- 04Delivery into the systems you already run
- 05MRC v3 and the digital contract
- 06Evidence for the risk committee
Then: how underwriters set the rules themselves, what a Chief Underwriting Officer asks of a technology vendor, and how it connects. Sources are listed at the end.
Use case 01 · Carriers and MGAs
Submission triage: a decision with reasons, in seconds
The scenario
A marine cargo team receives forty broker submissions a day from six brokers, in six shapes: a placing platform's JSON, two different API formats, a PDF, a spreadsheet and an email. An underwriter reads each one just to find out whether it is in appetite before any real work starts.
With StrataLens
Each broker's system sends its submission exactly as it produces it. StrataLens maps it to the London Market model, applies the team's own appetite rules, and returns accept, refer or decline with the reasons and the values behind them. Anything referred lands under Referrals in Microsoft Teams, showing the London Market data and exactly why it was referred. The underwriter accepts or declines, adds an internal note, and the broker gets the decision on their own system.
What changes for the business
- Underwriters spend their time on referrals, not on reading everything to find them.
- Brokers get a decision with reasons the same day, so out-of-appetite risks stop clogging both inboxes.
- Appetite is applied the same way by every underwriter, every day, and every decision is on the record.
Why it matters now. In the LMA's April 2026 survey, over 60% of Lloyd's market firms require mandatory human review of AI-generated outputs, and 93% have or are developing an AI governance framework [2]. StrataLens is built for that posture: decisions come from rules your underwriters approved, not from a model, and a person decides every referral.
Use case 02 · Delegated authority
Bordereaux: mapped once, checked every month
The scenario
An MGA reports monthly risk and premium bordereaux to four carriers, and each carrier wants its own column layout. On the carrier side, the delegated authority team re-keys or re-formats each coverholder's file into its bordereaux system, and discrepancies surface by email weeks later, or at audit.
With StrataLens
The coverholder's feed is mapped once to the London Market model. StrataLens' dictionary recognises how real systems name their fields, so Insured Name, insuredName and insured.name are all read as the same thing, and a list of risks becomes one record per risk with its own lineage. The carrier's binder rules run on every row. Problems are reported against the row and field they came from, and the carrier receives clean records in its own system's format.
What changes for the business
- One mapping per coverholder replaces a monthly re-formatting exercise.
- Out-of-binder risks are referred on arrival, not found at the quarterly review.
- Every value on the carrier's record traces back to the coverholder's row and column, which is what an auditor asks for.
Why it matters now. Lloyd's 2026 Market Oversight Plan keeps a continued focus on the oversight of delegated business [3], and the London Market Group's Data Council consulted in 2026 on extending the Core Data Record to delegated authority [7]. Firms that already hold bordereaux as governed, traceable records will have the least work to do when that standard lands.
Use case 03 · Brokers
One broker feed, many carriers
The scenario
A Lloyd's broker places a risk with a leader and three followers. Each carrier wants the submission in a different shape, so the broking team prepares four versions of the same risk and chases four decisions through four channels.
With StrataLens
The broker sends once. Each carrier the broker invites onto the exchange receives the risk as a London Market record, never the broker's raw message, applies its own rules privately, and its decision goes straight back to the broker's system, signed so the broker can prove where it came from. Either side can suspend or end the exchange, and both see every step in their own audit log. No other organisation sees anything.
What changes for the business
- The broker builds one integration and reaches every carrier on the exchange.
- Every carrier sees the same, consistent data, so queries about “which version” stop.
- Decisions arrive in one place, with reasons, on the broker's own system.
Why it matters now. 46% of London market firms name legacy system integration as a key barrier to AI adoption, and 49% name data quality [1]. Point-to-point integrations are how both problems are created. One record, mapped once, is how they are removed.
Use case 04 · IT and architecture
Delivery into the systems you already run
The scenario
A carrier wants triaged submissions in its underwriting workbench and bound risks in its policy administration system. Every new sender has meant a new adapter, written by a developer who has since moved on, and none of them survive the next core upgrade.
With StrataLens
The carrier's admin gives StrataLens the OpenAPI document of the receiving system and picks the operation to call. StrataLens suggests which London Market field fills each of the system's fields, the admin reviews the table and previews a real record, and from then on every record arrives as a native record in the system's own format. There is no adapter to build. Deliveries are signed, retried until the system accepts them, and visible in the app with every attempt and response.
What changes for the business
- Adding a sender is a mapping, not an integration project.
- Your systems never wait on StrataLens: sending is asynchronous and repeats are recognised and processed once.
- Everything can be stopped in one click and is audited: pause an endpoint, revoke a key, switch off a delivery.
Why it matters now. Lloyd's Principle 12 expects managing agents to maintain robust and resilient operations, embedding cyber resilience and effective third-party risk management [4]. The PRA's SS2/21 expects documented exit plans and full audit and access rights for material third-party arrangements [8]. StrataLens' canonical model is documented, your mappings belong to you, and on exit your data is exported and deleted within 30 days.
Use case 05 · Placing data
MRC v3 and the digital contract
The scenario
The risk on the placing platform, the risk in the pricing tool and the risk in the policy system have drifted apart. The slip says one thing, the PAS another, and when a query comes in nobody can say which field came from where.
With StrataLens
The London Market model is organised like a Market Reform Contract and built from the public MRC v3 standard: placement (UMR, stage, placing basis), parties, risk details, period, limits, deductibles and premium, market and security, contract terms, and fiscal and regulatory. Coded fields are resolved through StrataLens' own reference lists, so H&M and Hull & Machinery are the same class, and Y, Yes and true are the same answer. Every value carries its source field, the conversion applied and the raw value it came from.
What changes for the business
- One governed, MRC-shaped record per risk, shared by placing, pricing and policy teams.
- Field-level lineage answers “where did this number come from?” in one click.
- A record already structured like the MRC is the natural starting point for the market's Core Data Record.
Use case 06 · Risk, compliance and audit
Evidence for the risk committee
The scenario
The Chief Risk Officer asks a simple question about an AI-assisted process: “This risk was accepted automatically. Show me why, and show me who approved the logic.” The honest answer, for most tools, takes a week and a developer.
With StrataLens
Every decision records which rules fired and the values behind them. Every mapped value records its source field, the conversion applied, the raw value and whether it was valid. Mappings and rules are versioned, only one version is live at a time, and every approval is in an audit log that cannot be edited or deleted, enforced in the database rather than in application code. AI is used only to draft mappings and rules from field names and short examples; it never sees whole submissions, never decides, and nothing it proposes takes effect until a person approves it. Processing stays in Azure UK South, and customer data is not used to train any model.
What changes for the business
- The answer to “why was this accepted?” is a click, not a reconstruction.
- Model governance is already documented: which AI is used, where, for what, and what it can never do.
- Third-party risk questionnaires are answered from the same evidence, not written fresh each time.
Why it matters now. The LMA's 2026 survey found data privacy, cybersecurity and third-party risk are now the leading AI concerns in the market, and that around one in four firms still rely on general third-party risk frameworks rather than AI-specific provisions [2]. Evidence that already exists is cheaper than evidence assembled under pressure.
Underwriters set the rules
No developer. No ticket. No waiting for the next release.
Appetite changes faster than IT backlogs. In StrataLens the people who own the appetite write it, test it and switch it on themselves, in the Teams app they already have open.
If information a rule needs is missing, the submission is referred, never guessed. If a request is unclear, StrataLens asks a question rather than inventing an answer. Old versions are kept for the audit trail.
- Write it in plain English. “Refer if the maximum value per conveyance is over USD 100,000; decline if any exposures are declared.”
- See exactly what it checks. StrataLens drafts a precise rule against London Market fields and shows it next to your words. It only uses fields that exist and make sense for their type.
- Test it on real history. Run the rule set against recent submissions and see how many would have been accepted, referred or declined.
- Approve and apply. That version goes live. Every new submission gets a decision with the reason and the values behind it.
- Decide the referrals. Anything referred appears in Teams with the data and the reason. Accept or decline, add a note, send a message to the broker.
For the Chief Underwriting Officer
What a CUO asks of a technology vendor, and how we answer
These are the questions we are asked in the first meeting. We would rather answer them here, with the status of each, than imply a maturity we have not reached.
| The demand | How StrataLens answers it | Status |
|---|---|---|
| “Do not touch my core systems.” | StrataLens sits alongside your estate. Nothing is migrated or replaced, and switching it off leaves your systems exactly as they were. | In place |
| “Show me why every decision was made.” | Decisions come from a deterministic rules engine. Each one records the rules that fired and the values behind them; each value records its source field. | In place |
| “Keep my underwriters in charge.” | Underwriters write, test and approve the rules. AI only drafts; a person approves. Referrals are decided by a named underwriter and audited. | In place |
| “Keep my data in the UK and out of anyone's model.” | Azure UK South only, enforced by policy. The AI service is a UK regional deployment reached over a private endpoint and does not train on customer data. | In place |
| “Do not make my people learn a portal.” | StrataLens is a Microsoft Teams app. Your people sign in with your own Microsoft identity, with your MFA and conditional access. | In place |
| “I am not funding a nine-month integration.” | Three ways to connect, none of which needs a new message format. A sender is set up from a sample message in about ten minutes; its developers change a base address and add a header. | In place |
| “My risk team will want evidence, not promises.” | Append-only audit log; access and audit rights in the contract; a drafted DPIA; a CREST-accredited penetration test with results shared. | Audit in place Pen test before first paid pilot |
| “I need to be able to leave.” | The canonical model is documented, so your mappings are not locked in. Exit plan and data return are agreed before contract; data is exported on request and deleted within 30 days. | Policy in place Contract terms before production |
| “Which certifications do you hold?” | Cyber Essentials, Cyber Essentials Plus and IASME Cyber Assurance Level 1, all certified in October 2026 (verified badges on our trust centre). We build against the SOC 2 and ISO 27001 control sets, with 43 controls mapped and evidenced; we do not hold those two yet and will not claim otherwise. | CE, CE Plus, IASME CA L1 SOC 2 / ISO 27001 roadmap |
How it connects
Connect in minutes, not months
Your developers do not reformat anything and do not adopt a new standard. They choose one of three ways to connect, and mapping to the London Market model happens inside StrataLens, not in your code.
Keep your own API
Give us your OpenAPI (Swagger) document. StrataLens answers at the same path, method and body your system already calls. Your team changes only the base address and adds a key.
Tell us a record is ready
Your system publishes a “record ready” event on your Azure Service Bus or as a web notification (CloudEvents or plain JSON). StrataLens fetches the record from your API, with listen-only access you grant.
What “minutes” means
Setting up a sender takes about ten minutes once a sample message is to hand: paste the sample, issue a key, click Suggest mapping, review, approve. The time after that is your developers changing an address and a header. Systems with no API and no file export are the exception, and we say so before a pilot starts.
What your IT team will check
HTTPS only. Keys shown once and stored hashed. Signed deliveries with replay protection. Retries that back off and stop. One fixed outbound address to allow-list. Rate and size limits per key. Full detail is in our integration guides and the CTO and architect FAQ, available on request.
References
- Lloyd's Market Association, Over one-third of London market firms now actively using AI (2025 AI survey): integration with legacy systems 46%; data quality and availability 49%; cost and uncertain return 48%.
- Lloyd's Market Association with Barnett Waddingham, AI adoption more than doubles across the Lloyd's market in 12 months, 16 April 2026: 93% of respondents have or are developing an AI framework; over 60% require mandatory review of AI-generated outputs; respondents represent over 60% of Lloyd's market stamp capacity.
- Lloyd's, Market Oversight Plan 2026: continued focus on oversight of delegated business.
- Lloyd's, Principles for doing business at Lloyd's, Principle 12: Operational Resilience.
- Lloyd's, Core Data Record: the relationship between the MRC v3 and the CDR.
- London Market Group, Data Council launches Core Data Record v3.3, supporting treaty reinsurance, May 2026.
- Lloyd's Market Association, 2026 “year of significant transition” in handling of market operations, December 2025: Core Data Record to extend to claims and delegated authority during 2026.
- Bank of England, Prudential Regulation Authority, SS2/21 Outsourcing and third party risk management.
Product capabilities and status on this page are as of October 2026 and describe our pre-production environment. We update this page as capabilities reach production.
See one of these running on your own file
Bring a bordereau, a submission or a slip. Thirty minutes, no preparation needed at your end.