dutira
Book a Demo

Module A · Governed AI Workspace

The AI your people want.The control your firm needs.

One familiar workspace with approved models. Every request is checked against your policies before it reaches a model — and every decision is recorded.

Book a Demo Free AI governance check Phase 1 · available now

One workspace for approved AI

OpenAI, Anthropic and Google models in one familiar interface, chosen by role — so people use AI productively, inside company-approved controls.

Policies and safeguards on every request

Organisational AI usage policies applied automatically; sensitive data masked or blocked, risky requests held for a person, and staff see why.

Auditable by default

Employee activity logged and every decision recorded, subject to the logging and retention your firm configures.

See it work

Try it.

An interactive demonstration with a fictional firm’s data.

Illustrative demo · fictional data
AI Workspace — Halden Row WealthClaude Sonnet 5.5Illustrative
Summarise the changes to our complaints handling procedure for the compliance team.

Policy checks · AI Acceptable Use v14

Permitted · 5 of 5
  • Model access — Claude Sonnet 5.5 is approved for your role
  • Company AI policy — Internal summarising is a permitted activity
  • Sensitive information — No client data found
  • FCA-linked controls — No advice, promotion or complaint decision
  • Human approval — Not required for internal use
AN

Three changes take effect from 1 November [1]:

  • Acknowledge complaints within two working days.
  • Senior handler review before final responses.
  • Vulnerability flags recorded on every case.

Source [1] Complaints Handling Procedure v3.2 · section 4

Decision recorded to the audit trail · DEC-20481 Included in monitoring and compliance reports

Ask anything. Your request is checked against firm policy first.
Decision log — compliance viewIllustrative
  • DEC-2048109:12ComplianceClaude Sonnet 5.5Allowed
  • DEC-2049309:40Client ServicesGPT-5MaskedRule: Phone number
  • DEC-2050210:05AdviceClaude Opus 5.5Held for approvalRule: Personal recommendation
  • DEC-2051110:31Client ServicesGPT-5BlockedRule: NI number
  • DEC-2051711:02InvestmentGPT-5Allowed
  • DEC-2053211:47RiskClaude Opus 5.5Allowed

People pseudonymised. Reading content needs a recorded reason.

Content retentionContent older than 90 days is deleted daily; each purge is audited.

Capabilities

What it does today — and what is coming.

In the product today

  • One workspace for approved models from OpenAI, Anthropic and Google, called server-side — staff never hold API keys
  • Centralised user access and permissions: single sign-on with OpenID Connect, model access and data limits by role
  • Organisational AI usage policies: versioned, acknowledged by staff, and applied to each request
  • Sensitive-data safeguards: rule-based detection of client identifiers and your own patterns, with mask, warn, hold or block
  • Human approval for high-risk requests, with four-eyes and escalation
  • Employee activity logging and a searchable, hash-chained record of AI interactions — subject to configurable logging and retention
  • Secure workspace administration: roles, permissions, models, policies and retention managed centrally

Planned

  • SAML and SCIM provisioning
  • A reviewed machine-learning classifier for names and free-text client details
  • Email, Teams and Slack notifications for approvers
  • OCR for scanned documents

FCA AI governance expectations

How it relates to the FCA’s October 2026 publication.

The FCA’s AI in financial services: good practice and areas for improvement (19 October 2026) shares examples from industry. In its own words: “They do not create new requirements.” There is no standalone FCA AI rulebook; existing rules apply.

Below, each example it describes, and how this module supports a firm’s own approach.

Using approved source material and guardrails to reduce hallucinations, with checks and escalation routes

Dutira: Answers can be limited to approved company documents with citations; requests that need a person are held and routed.

FCA AI good practice §2.2 — Consumer understanding Good practice

Maintaining auditable records of prompts, outputs, model versions, filters applied and escalation decisions

Dutira: Each decision records the rules applied, the model, the outcome and any approval, in a hash-chained log.

FCA AI good practice §2.2 — Consumer understanding Good practice

Logging and audit trails of AI interactions to monitor outcomes and deal with complaints

Dutira: Decision log and audit trail, searchable by authorised roles; content access needs a recorded reason.

FCA AI good practice §3.2 — Monitoring and oversight Good practice

Human-in-the-loop as a risk-based control, with clear thresholds for intervention

Dutira: Rules decide which activities are allowed, warned, held for approval or blocked, by role and model.

FCA AI good practice §3.4 — Human oversight Good practice

The rules behind it

Existing rules continue to apply when firms use AI. Which apply depends on each firm’s permissions and business.

  • Rule Binding on the firms it applies to
  • Guidance How the FCA expects rules to be met
  • Good practice Illustrative examples; no new requirements

Your responsibility. Your firm decides its AI policy, which models and activities are allowed, and who approves. Dutira applies and records those decisions; it does not decide them. Dutira is not approved or endorsed by the FCA and does not guarantee regulatory compliance.

One platform, launched in phases

See it with your own policies.

A 30-minute walkthrough, tailored to your firm.

Book a Demo Free AI governance check