Enterprise AI,
on the record.
Six connected products to use, build, govern and run enterprise AI. One identity, shared policies and a tamper-evident audit trail, on your infrastructure with the models you choose.
One platform. Six connected products.
From an employee's first question to a production workflow, each product has a clear role and its own access controls. Choose a product to walk through the live screens, point by point, or let the tour play.
One front door for every employee.
The live product: the same screen your staff open on day one. Ask for answers, give Work an objective, and find approvals, knowledge and AI Solutions in one place.

What AIRA Home does
Where every employee uses AI. Ask for answers; Work for objectives. One inbox for approvals, running and failed work, with AI Solutions a click away.
- Ask: governed chat across approved models, grounded in knowledge, with citations and abstention
- Work: give AIRA an objective and follow it through approvals to a result
- Knowledge Hub: access curated enterprise knowledge collections, from policies to compliance guidance, with collection-level permissions and answers grounded in source material
- Document Insights: upload, OCR and chat with a document
- The AI Solution catalog, with a run page for each type and replayable run history
Start from banking templates.
Copy a template, wire in your knowledge and systems, and publish, with sign-off designed in. What a builder tests in Studio is exactly what Runtime runs.


What AIRA Studio does
Where teams build. One path from idea to production: Build → Configure → Assign AI → Test → Evaluate → Publish. What a builder tests in Studio is exactly what Runtime runs.
- Four kinds of capability: AI Assistant, AI Agent, AI Flow and AI Solution
- A Solution Advisor that drafts a capability from a plain-language description
- Typed-node workflow canvas: prompt, knowledge, HTTP, code, database, agent, MCP tool, branching, error handling and subflows
- Prompt Lab, a run debugger, and evaluation scenario libraries graded by programmatic checks plus an LLM judge
- Templates, data sources, an encrypted connections vault and MCP servers
Every model call, on one screen.
The live state of the model plane: traffic, errors, spend and policy, per time window, through one governed door to every model.

What AIRA Gateway does
One controlled, audited entry point for on-premises and cloud LLMs, agents and AI Solutions. AIRA Nexus Runtime and external applications access the same governed catalog through application subscriptions.
- One catalog: approved LLMs, agents and AI Solutions for Nexus and external applications
- Application subscriptions: subscribe to catalog offerings with controlled access grants and limits
- Policy controls: models, fallbacks, routing, guardrails, data policy and security level
- Audited access: identity-bound requests, usage tracking and a record of every call
Published once, live by environment.
What Studio publishes, Runtime runs: versioned, approved and isolated per environment.

What AIRA Runtime does
Where published capabilities run. Runtime deploys versions, governs execution and gives your team full control over environments on Kubernetes and OpenShift. Access to models, agents and AI Solutions is controlled through AIRA Gateway.
- Environment control: manage environments, versioned deployments and schedules on Kubernetes and OpenShift
- Governed access: each deployed capability uses its own identity and access grants through Gateway
- Execution governance: approvals, run history and a durable record of every execution
- Triggers in and deliveries out with retry; usage and cost per deployment
- Tenant isolation enforced in the database
Evidence your auditors can verify.
Every action lands in a hash-chained ledger, checked in place and exported in one click.

What AIRA Signal does
One answer to "what is our AI doing?" Signal observes runs, cost and health across the estate, and gives investigators the thread to pull when something looks wrong.
- Overview of AI Solution runs, failure rate, node latency and spend, from one hour to ninety days
- Activity, LLM activity, journeys, flagged events and incidents
- Health of AI Solutions, knowledge, the workflow engine, the platform and the estate
- Cost, spend and security views, with export
Feed AI from your systems, governed.
The governed data plane: connect enterprise systems, ingest documents exactly once, and keep every extraction reviewable against the content it came from.
What AIRA Data does
The governed data plane. Connect enterprise systems, ingest documents exactly once, and keep every extraction reviewable against the content it came from.
- Governed connectivity to PostgreSQL, SQL Server, MySQL, Oracle and IBM Db2, read-only by default, every query audited
- Idempotent ingestion jobs that report incomplete attempts and never create duplicates
- Document review bound to the content hash and the extractor version
- A bounded tabular extraction service for CSV, TSV and JSON
One gateway. Every interaction governed.
AIRA Nexus brings the platform together around AIRA Gateway. AIRA Home users and external applications connect through this single entry point to approved LLMs and to agents and AI Solutions executed by AIRA Runtime. Every request follows the same access policies, limits and audit controls.
Send a request through the Gateway.
A simulation with illustrative rules. In AIRA Gateway, these rules are your own policies.
Built to be operated by your team.
On your premises, in your private cloud or in an isolated environment. Keep your data and AI under your control.
Shape your deployment.
- Flexible hosting
- Run in the environment that suits your organisation.
- Connected identity
- Single sign-on and shared access controls across the platform.
- Model choice
- Use private models, with approved cloud providers when needed.
- Operational visibility
- Monitor health, usage and activity from one place.
Prove it on your infrastructure, against your own criteria.
A scoped pilot runs AIRA Nexus inside your perimeter, on your models and your data, with success criteria written down before we start. Your security team reviews the real system, not a slide.

