Our AI transparency statement
In4ra is committed to using artificial intelligence responsibly, openly, and in a way that keeps people in control. This page explains how we use AI, where we do not, and the safeguards we have in place.
Last reviewed: August 2026
Our guiding principles
These principles underpin every decision we make about adopting or deploying AI tools, whether internally or in client-facing services.
Transparency
We are open about where and how AI tools are used within our services and operations. We do not use AI in ways that are hidden or deceptive to our clients or staff.
Human oversight
AI-assisted outputs are always reviewed by a qualified In4ra team member before being acted upon. No automated AI decision directly affects a client without human review.
Data protection
We do not input personal data, confidential client information, or sensitive business data into third-party AI tools unless covered by an appropriate data processing agreement.
Fairness & non-discrimination
We actively monitor AI-assisted processes for bias. We do not use AI in recruitment, performance management, or client-facing decisions without appropriate safeguards.
Security
AI tools used internally are subject to the same security vetting as any other third-party software. Access is restricted to authorised staff on a least-privilege basis.
Continuous review
Our AI use policy is reviewed at least annually, or sooner when significant new tools are adopted or when regulatory guidance changes.
Where we currently use AI
We maintain a register of AI tools and use cases. The table below reflects our current position as of the last review date.
Client-facing chatbot (Steve)
In useSteve uses a hybrid design: routine queries (packages, pricing, contact details, requests to speak to a person) are handled by deterministic, in-house logic, while free-text queries are answered by a third-party large language model (Anthropic’s Claude), reached via our own Cloudflare Worker rather than embedded directly on the page. Conversations are not stored beyond the browser session.
Internal documentation drafting
In useStaff may use AI writing assistants to draft internal documents such as proposals, reports, and knowledge-base articles. All outputs are reviewed and edited by a human before use.
Cyber threat intelligence
In useAI-powered threat detection tools are used within our managed security services to identify anomalous behaviour. Alerts are triaged by our technical team before any action is taken.
Automated ticket categorisation
PilotWe are evaluating AI-assisted categorisation of support tickets to improve response times. This is currently in a controlled pilot and not yet in production.
Predictive maintenance
ExploringWe are exploring AI tools that can predict hardware failure and network degradation for managed IT clients. No client data is used in external model training.
How Steve guides you around our site
In4raGP1 lets Steve show you the part of our website he is talking about instead of only describing it. When his answer relates to a specific section, a button appears beneath his reply — press it and the page moves there and highlights it, so you can check the answer against the page rather than take his word for it.
What In4raGP1 can do
- Scroll the page to a section of our website
- Take you to another page on our website where that section lives elsewhere
- Briefly highlight the section, so you can see which part Steve meant
- Offer a “Back to where you were” button to return you to your previous position
What In4raGP1 cannot do
- Click buttons or links on your behalf
- Fill in or submit any form
- Add anything to your basket
- Start or complete a purchase or payment
- Change any information you have entered
- Send any information anywhere on your behalf
You stay in control
In4raGP1 is the navigation feature only. It runs entirely in your browser, makes no network requests of its own, and stores nothing — it reads Steve’s reply, matches it to a registered section, and moves the page. It is separate from the rest of Steve: when you ask for a quote, leave your details, or request a callback, the information you give is handled by Steve’s enquiry features and covered by our Privacy Policy.
In4raGP1 Policy v1.0, published August 2026, applies to In4raGP1 v1 as deployed on in4ra.co.uk. Any future version able to do more than move you around the page will be described here before it goes live.
How AI agents can access our site
Beyond how we use AI internally, we have built our own platform to be properly understood by AI agents and automated tools through open, documented standards, rather than by leaving the door open to indiscriminate scraping.
Independently scored
100
Discoverability
100
Content
100
Bot Access Control
86
API, Auth, MCP & Skill Discovery
Scored by isitagentready.com, Cloudflare’s independent agent-readiness checker. Commerce is not yet assessed, since automated purchasing is not enabled on our site — see below.
View the live reportMachine-readable discovery
We publish llms.txt — a clear, structured summary of who we are, what we do, and who we serve — alongside explicit content-usage signals in robots.txt on our API subdomain. Search indexing is welcomed; bulk AI-training scraping of our content is not.
Open API catalog
Our API subdomain (api.in4ra.co.uk) exposes a standards-based catalog (RFC 9264 linkset) pointing to our OpenAPI specification, documentation, and a live health check. We advertise this via standard HTTP Link headers (RFC 8288) on our main site, so agents do not need to know our API subdomain in advance.
Documented agent authentication
Where an agent needs to act on a specific client’s behalf, it must authenticate through a documented identity flow (OAuth 2.0 discovery and token exchange) rather than being granted anonymous access to account data. Access can be revoked at any time.
In-browser AI tools
Compatible AI browser assistants can query a small, safe set of live tools directly on our site — our services, the sectors we support, our contact details, and a simple support Q&A — instead of guessing from scraped page content.
Published agent skills
We publish a Model Context Protocol server card and an agent-skills index describing exactly what an authorised agent can request from us today, starting with raising a support enquiry. This list grows deliberately, not by accident.
What this means for you
What we do not use AI for
The following activities are explicitly excluded from AI-assisted or AI-automated processes at In4ra. These boundaries are reviewed as part of our annual policy review.
- Automated hiring or rejection of job applicants
- Fully automated client billing or contract changes
- Unsupervised generation of legal, financial, or medical advice
- Processing special-category personal data through third-party AI tools
- Replacing human engineers on critical infrastructure changes
Human oversight commitment
In4ra operates a “human-in-the-loop” policy for all AI-assisted processes that affect clients or staff. This means:
Data protection & AI
In4ra processes personal data in accordance with UK GDPR and the Data Protection Act 2018. When AI tools are used in connection with personal data, we apply the following controls:
- We carry out a Data Protection Impact Assessment (DPIA) before deploying any AI tool that processes personal data at scale.
- We do not use personal data to train third-party AI models unless we have explicit contractual protections in place confirming the provider does not use our data for training.
- Special-category personal data (health, ethnicity, biometrics, etc.) is never processed through third-party AI tools.
- Where AI tools are hosted outside the UK or EEA, we ensure appropriate transfer mechanisms are in place under UK GDPR Chapter V.
For questions about how your data is handled in connection with AI, please contact our Data Protection lead at [email protected].
Questions about our AI use?
If you have concerns or questions about how In4ra uses AI in delivering your services, please get in touch. We are committed to answering any reasonable query openly and promptly.