SELF-HOSTED AI DOCUMENT REVIEW
StructLex checks document sets against regulations, internal policies and review criteria, and returns findings linked to the exact source passage. It deploys into your Azure, AWS or GCP subscription or your data centre, and runs on the models you choose.
We deploy it with your platform team and tailor it to your document types, systems and workflows.
The plan is tested yearly, but no test scenario covers a switchover from the primary ICT infrastructure to redundant capacity, which Art. 11(6) requires.
A finding with its verdict, reason and quoted passages. Illustrative DORA review, sample data.
WHY TEAMS SWITCH
The usual options
Reading documents against rulebooks by hand takes most of a compliance team's time, and external findings still arrive late.
Regulated documents go to a vendor cloud and a third-party model, and procurement stalls there.
Model usage is metered inside the subscription, marked up, and moves every month.
A generic model produces verdicts without the passage they rest on, so the team re-checks everything by hand.
With StructLex
It extracts the obligations and checks each document against them. Your reviewers accept or challenge the result.
Installed in your cloud subscription or data centre, behind your identity, storage and logging.
It calls the endpoint you already govern. Usage is billed by your cloud provider, with no margin in between.
Verdict, reasoning and the quoted passages it relied on, linked to the page, so an auditor can follow the chain without asking anyone.
WHAT STRUCTLEX DOES
Every step is a screen in the product, shown here on an illustrative DORA review of a fictional reinsurer. Reviewers stay in control of every finding.
Upload a regulation, an internal policy or a checklist in its original language. StructLex reads it into a scope: one obligation per requirement, with the evidence it would expect to see and the external references it resolved along the way.
A coverage profile narrows the scope to one subject: your organisation, or a counterpart you review. An applicability questionnaire, pre-filled from what the platform already knows about that subject, decides which obligations bind it and shows why.
The workflow generated from the profile checks each obligation against the documents you point it at, in whatever language they are written. Every automated review comes with a verdict, a reason and the quoted passages it relied on, linked to the page.
Failed obligations become tasks with owners, priorities and due dates on a board your team already understands. Reviewers can challenge a finding in a conversation with the AI, and the report exports when the run is signed off.
WHERE TEAMS RUN IT
One workflow covers the regulatory gap analysis a regulator asks for, the internal review that runs every quarter, and the checks your own systems trigger a thousand times a month.
Regulatory gap analysis
Solvency II, DORA, CSRD, the EU AI Act, or your national regulator's rulebook. Upload the regulation and your documentation. StructLex extracts each obligation, checks your evidence against it and produces a report with every finding linked to source.
Internal audit and policy compliance
Check operations against internal policies, control frameworks and vendor requirements on a schedule, with the same evidence trail you would show an external auditor.
High-volume checks through the API
Fraud case reviews, claims against policy terms, loan files against lending criteria, vendor due-diligence packs. The criteria are defined once in StructLex and every new file is checked against them from your own systems.
Works with any structured rule set
RegulationsInternal policiesInsurance termsLending criteriaQuality standardsFraud frameworksDue diligence checklists
DEPLOYMENT
Security teams ask first where the documents go. StructLex installs into your cloud subscription or data centre and connects to the model endpoint your organisation has already approved. Nothing in the diagram sits outside your environment.
Your documents and systems
StructLex core
Your LLM endpoint
Azure, AWS and GCP subscriptions, or your own data centre. We deploy with your platform team and hand over the runbooks.
Azure OpenAI, Amazon Bedrock, Google Vertex AI, or open-weight models served on your hardware. You choose the model and you keep the usage logs.
For organisations that prefer not to host it themselves, StructLex runs as a managed deployment in a region they choose, under the same licence and services model.
SECURITY
Because StructLex runs in your environment, most questions on a vendor questionnaire have the same answer: it sits under your controls.
Documents, intermediate data and outputs stay in the region and tenant you choose.
Runs behind your network and identity controls, with the permissions your platform team sets.
Uses the model endpoint your organisation has approved, with usage visible in your own logs.
None by default. Tailoring work happens under access you grant and can revoke at any time, or on samples you approve.
StructLex credentials
Built in EuropeMicrosoft AI Cloud Partner ProgramRegistered Microsoft CSPAzure Marketplace listingICO registeredHOW A ROLLOUT RUNS
A generic setup produces findings your reviewers have to redo. So the StructLex core stays one product, and the integrations, workflows and model tuning around it are built for your document types and rulebooks. Our engineers do that work with your team, from the first pilot to steady-state operation.
We map your document types, rulebooks, reviewers and the systems your findings need to reach.
A real workload runs in your environment, and we go through the results with your reviewers.
StructLex is installed into your tenant and connected to identity, storage and your model endpoint.
Integrations with your ERP, DMS or case systems. Custom workflows. Prompt and retrieval tuning on your documents. Help fine-tuning models you own, where it pays off.
Versioned releases, maintenance and a named engineer who knows your deployment.
You keep the deployment, the integrations and any models tuned on your data. The knowledge stays in your organisation.
SharePoint and file stores, document management systems, ERP, CRM and case systems. Findings flow back out through the API. Where a connector does not exist, we build it and maintain it as part of the rollout.
COMMERCIAL MODEL
AI subscriptions break down in two places: usage fees finance cannot forecast, and model costs marked up inside the price. StructLex keeps the two apart.
Available on Azure Marketplace for organisations that procure through Microsoft. There is no public price list; every deployment is quoted.
QUESTIONS
In your environment. StructLex installs into your Azure, AWS or GCP subscription or your data centre and calls the model endpoint you already govern. Documents, intermediate data and findings stay inside a boundary you control.
The endpoint you choose: Azure OpenAI, Amazon Bedrock, Google Vertex AI, or open-weight models served on your own hardware. Usage shows up in your own logs and on your own cloud bill.
Model usage is billed by your cloud provider at your rates, with no margin added. StructLex itself is a fixed annual licence plus scoped services, so the two numbers stay apart.
Because it is tuned to you before go-live. Prompts, retrieval and workflows are adjusted to your document types, rulebooks and terminology during the pilot, and kept adjusted with your team afterwards. Every finding also carries the passage it rests on, so a reviewer checks the source rather than the model.
Any regulation, internal policy or structured checklist, in its original language. Rules and documents can be in different languages. Teams run it on Solvency II, DORA, CSRD, the EU AI Act, national rulebooks and their own control frameworks.
Yes. Define the criteria once and trigger checks through the API from your case, workflow or pipeline system. Each result comes back with the same verdict, reasoning and source links.
We do, as part of the engagement: versioned releases, maintenance and a named engineer who knows your deployment. You own the deployment, the integrations and any models tuned on your data.
StructLex runs for you as a managed deployment in a region you choose, under the same licence and services model.
SEE IT ON YOUR OWN DOCUMENTS
Bring one regulation and the documents it applies to. In thirty minutes you see the obligations extracted, the findings with their evidence, and the deployment options for your cloud or data centre. Bring your platform or security lead if you can.
WHAT COMES WITH IT