AI readiness assessment
A proper look at your tenant, your data estate and your risk appetite. We tell you what you can safely switch on this quarter, and what needs work first.
Services
AI that makes it to week three.
Microsoft 365 Copilot, Claude, and custom agents, rolled out with the boring bits done. Data hygiene, governance, training, measurement. The reasons most AI pilots stall.
Why it matters
Most AI pilots die in week three.
Not because the tech does not work. Because the data is messy, the policy is missing, the licences are wrong, or the team was never shown what to type. We have seen the pattern. Every time.
We run AI rollouts like a proper programme. Readiness assessment, data hygiene, governed pilots, measurement, training, and an operating rhythm afterwards. You end up with hours back, not a cancelled subscription.
Copilot in anger Book a Copilot readiness workshop
AI that earns its seat. Not just another chatbot.
Copilot done properly is hours back per week, not a pretty demo. We handle governance, data readiness, rollout and adoption so AI actually sticks.
For your people
- Copilot readiness check before any rollout
- Guided pilots with the teams who need it most
- Training that covers real work, not features
- Adoption reviews at 30, 60 and 90 days
For your governance
- Sensitivity labels and DLP configured first
- Data oversharing audits, fixed before go-live
- Usage analytics by team and licence tier
- Custom agents built on vetted sources
Adoption up Week 12
Finance hit 74% weekly active. Avg 6.4 hours saved per user. No data-leak flags.
Hours saved per user per week
6.4
What we deliver
Six engagements that cover the ground.
Pick one, or stack them. Most clients start with readiness and a Copilot pilot.
Microsoft 365 Copilot rollout
Licence planning, SharePoint data hygiene, pilot groups, measurement, and the change comms that actually get adoption past the 20 percent cliff.
Claude and custom agents
Anthropic Claude deployed against your knowledge bases with proper access controls. For when Copilot is not the right shape for the job.
Agentic automation
Agents that raise tickets, draft quotes, triage inboxes, reconcile invoices. Built on Power Automate, n8n or custom, governed from day one.
AI governance and policy
Acceptable use policy, data classification, audit logging, DLP, and a named owner. AI that passes your auditor the first time.
Training and enablement
Short sessions by role. Prompt libraries. Internal champions. The difference between paying for a Copilot seat and getting work done with it.
How we run it
Four phases, measurable at each gate.
Adoption is a behaviour change project. Treat it like one and it lands.
Readiness
Two weeks. Tenant review, data estate audit, risk register, licence map, a prioritised backlog of use cases.
Pilot
Four to six weeks. Governed pilot group. Measured adoption, measured time-saved. Early wins, then the harder cases.
Rollout
Wider deployment with proper change comms, training, and a prompt library. Weekly stand-ups until stable.
Govern and iterate
Usage reporting, quarterly policy review, new use cases added in batches. AI as a capability, not a one-off project.
Frequently asked
The questions we get every week.
Is our M365 tenant actually ready for Copilot?
Probably not quite. The common blockers are oversharing in SharePoint, missing sensitivity labels, and a sprawl of duplicated documents. We find them in a two-week readiness review and give you a remediation plan before a single Copilot licence is bought.
Copilot or Claude?
Both, usually. Copilot is the right answer for anything tightly coupled to Word, Excel, Teams and Outlook. Claude tends to win for longer-form reasoning, custom agents against your own data, and anywhere you want tighter control of the model and the prompts.
How do we stop staff pasting customer data into public AI?
A combination of an acceptable use policy, Purview DLP rules, and giving them a sanctioned alternative that is actually good. Banning tools does not work. Providing a safe, fast, well-trained option does.
What does a realistic first year look like?
Month one: readiness and policy. Month two to three: pilot with 20 to 40 users on two or three use cases. Month four to six: wider rollout and a first custom agent. Month seven onwards: quarterly batches of new use cases as the business finds them.
Can you build a private agent on our data?
Yes. We build on Anthropic Claude or Azure OpenAI, with retrieval against your SharePoint, file shares or apps, proper auth, proper logging, and a decent UI. Scoped, priced per agent, delivered as a project.
We are regulated. Does that change things?
It changes the evidence trail, not the outcome. Financial services, healthcare and legal clients all run Copilot and custom agents with us. We handle the DPIA, the audit logs, and the retention policy up front.
Free · no slide deck
Thinking about Copilot? Start with readiness.
Two weeks, fixed fee, and a plan you can take to your board. Whether we do the rollout or not, you will save yourself a false start.