Openkoda AI
AI that does the work - with a human in control
Openkoda puts AI across the whole platform: configure products, price, report, read documents and support underwriting in plain language. AI drafts only typed, valid changes; your team approves - on a platform you own.
Six ways AI is boosting your insurance business
Each one is practical, scoped and reviewable - built into the product, not bolted on.
AI Product Builder
Describe a product or a change in plain English; AI drafts the configuration - coverages, pricing, rules and workflow - for you to review and approve.
Use it to: launch a new product or change pricing and rules from a plain-language brief.
Value: launch and iterate without developers or a release cycle.
Reporting AI
Ask questions about your book in plain English and get a report or dashboard back, with your data kept private.
Use it to: answer “loss ratio by product this quarter?” without writing a query.
Value: self-serve insight for every team, no analyst queue.
AI document reading
Extract details from uploaded PDFs, emails and images and pre-fill the case - with a human confirming before anything is used.
Use it to: read a submission attachment or policy document and populate the application.
Value: faster, cleaner intake with fewer manual-entry errors.
Underwriting copilot
Surfaces the factors that matter on a referred case - risk signals, pricing context and suggested next steps - inside the workbench.
Use it to: give underwriters a quick, consistent read on each referred risk.
Value: faster, more consistent underwriting decisions.
AI steps in your workflows
Call AI at a step in an authored workflow - to summarise, classify or draft correspondence - as part of your process.
Use it to: auto-summarise a claim or draft a customer email mid-process.
Value: automate judgement-heavy steps without custom code.
Explain this
Turn any report or metric into a plain-language explanation, so the numbers are readable by everyone.
Use it to: explain a change in a KPI to a non-analyst stakeholder.
Value: insight the whole business can act on.
From plain English to a live product change
Describe the change
Type what you want - a new coverage, a pricing tweak, a new rule - in plain English.
Review a clear diff
AI proposes a typed, valid change-set; you see exactly what will change, before and after.
Approve & ship
Apply it - safely versioned so in-force policies are untouched, and fully audited.
Run your PAS from an LLM client
Connect an assistant like Claude straight to Openkoda over the Model Context Protocol - ask questions and take confirm-gated actions against your live book, in plain English. Every call is permission-scoped and audited.
Safe to use on real insurance decisions
One shared, guard-railed service powers every feature above - so you get the same safety, auditing and privacy everywhere, and can meet your obligations when AI touches consequential decisions.
Business users lead, humans approve
AI drafts; your team reviews a clear before-and-after and approves. For underwriting, claims and other decisions that affect people, a qualified person signs off - nothing ships on its own.
Typed changes only
AI proposes valid, typed change-sets from a fixed catalogue - never free-form code - so it cannot invent something your platform will not accept.
Audited every call
Every request and result is logged, attributed and timestamped. Where AI helps produce a decision that reaches a policyholder, you can disclose it and show the trail.
Private & tenant-scoped
Each call is scoped to your organization with strict tenant isolation; a deterministic mock mode is available for testing. Your book stays private, managed or on-premise.
One model, routed for you
Openkoda AI selects the right model per task behind a single service - you never manage keys or models.
Metered and transparent
A monthly credit allowance you spend per action - a report query, a document read, a product change - on any plan.
How Openkoda AI is priced
Openkoda AI is a separate monthly subscription on top of any plan, with a monthly credit allowance you spend per action - a report query, a document read, a product change. Extra credits are billed per use, and every plan and credit rate is published.
AI you can actually trust in production
Natural-language configuration is emerging across the market - a few closed platforms have begun to offer it. Openkoda's difference is that it's guard-railed, audited, and runs on a platform we operate for you - or deploy inside your own environment.
See Openkoda in your context
Book a live, personalized demo with our product team - tell us your use case and see the platform work with your data. No commitment.