AI your people use well, with engineers behind it.

How Managed AI works

• Deploy. We set up ChatGPT Business or Enterprise, Claude Team or Enterprise, or Microsoft 365 Copilot: single sign-on, user provisioning, permissions, admin settings, data retention and security options. Copilot gets particular care, because it can see whatever a user can see in Microsoft 365, so we check the permissions underneath first.
• Govern. Which tools are approved. What people can and can't put into them. How sensitive information, client data and personal information are handled. Who can create shared assistants and connect them to company data. Written in plain language your staff will read, and enforced through settings where the platform allows it.
• Train. Practical sessions, by team, using their own work: a finance team reconciling, a sales team preparing proposals, an operations team writing procedures. The aim is daily use, not a certificate.
• Build. Custom GPTs, Claude Projects, agents, Power Automate flows, API integrations, internal knowledge tools and scripts. Small requests (up to two hours each) are included from Managed Secure, regular building comes from Managed Complete's engineering capacity, and larger work is scoped as a project.
• Support. When an employee gets stuck on something they've built, they raise a request, the same way they would for a broken laptop, and an engineer helps them finish it.
• Improve. Each quarter we look for repetitive work worth removing, with automation, AI, an integration or a better process, and bring a short list to your review.

ChatGPT, Claude or Microsoft Copilot

You don't have to pick one, but most organisations should start with one. We'll help you decide based on how your teams work.

• Microsoft 365 Copilot works inside Outlook, Teams, Word, Excel and SharePoint, with your Microsoft 365 data. It suits organisations that live in Microsoft 365, and it needs your file permissions in order first.
• ChatGPT Business or Enterprise is a strong general assistant, with custom GPTs teams can build and share.
• Claude Team or Enterprise is strong for long documents, analysis and writing, with Projects that keep a team's context together.

Whichever you choose, we connect it to your identity platform, set it up to your governance rules and support it.

Help when your people get stuck

This is where most AI rollouts stall, and where Managed AI is different. Your staff will build things. That's the point. And they'll get stuck on the parts that need an engineer:

• "I built a workflow but can't get authentication working."
• "I need this Claude workflow to read documents from SharePoint."
• "My Power Automate flow fails every few days and I can't see why."
• "I've built a custom GPT and need it to call our internal API securely."
• "This prototype works. Can we make it reliable enough for the whole team?"

On Managed Secure and Managed Complete, those are support requests. An engineer helps your employee get it working, safely, with the right permissions and without secrets pasted into the wrong place. Each request is included up to two hours of engineering.

When it's bigger than a support request

Some requests turn out to be a build: a new integration, an agent that takes actions, a workflow that several teams depend on. When that happens we say so before starting.

• On Managed Secure, small builds are included, up to two hours each: a simple flow, a script, a custom GPT set up for a team.
• On Managed Complete, bigger builds come from your engineering capacity (10 minutes per user per month, used across the quarter).
• Larger work is scoped as a fixed-fee AI and automation project.

Whatever we build, we hand it back documented, with an owner, logging and a way to switch it off. Once it's live, keeping it running is part of your plan.

What's included in each plan

Managed AI is part of each plan. It isn't sold on its own, because AI that touches your identity, data and security needs the people who run them.

Licences and what you pay for

AI licences are separate from your plan price. You buy them directly from OpenAI, Anthropic or Microsoft, or through your licensing provider; Microsoft licences can come through North Ark. Either way they appear as their own line, so you always see what the software costs and what the service costs.

Your plan price covers deploying, governing, training, supporting and improving. It doesn't change with the AI platform you choose.