Microsoft for Startups PartnerAI Training for Teams
Your team is already using AI to write reports. Copying a paragraph out, pasting it in, tidying it up, with no agreement on what's allowed and no check on what comes out. It's inconsistent and not standardised. That's a training problem, and it's the one we fix.
Book a training call to learn more →By the end of session one, we'll have produced a draft report in your house format, from your notes, in front of your team. Not a demo — your template, your standards, your language.
By the end of the programme, your people can take site notes to a first draft, use AI to catch the carried-over paragraph and the wrong-project reference before the client does, and write to one house style across the firm.
Your reports, your style, your standards.
8 hours across three sessions to make sure it sticks
Run the numbers for your firm.
| Time per report | 1 hour |
| Reports per week | 15 |
| Hours per year | 700 |
| Recover a third | 230 hours |
That's roughly an extra part-time technician — without the salary. Even if you halve those figures, training still pays for itself in a month.
Scottish Building Standards Hub
Fife Council
Scottish Futures Trust
RSE Water Technology
Microsoft for Startups
Crown Commercial ServiceYou finish the report, the client receives it — and uploads it straight into their own AI to summarise, cross-check and query.
Mistakes that used to sit quietly in appendix C now get surfaced in seconds. Inconsistencies between sections get flagged. The tolerance your work used to enjoy is gone.
Templates work very well but as AI is being used more and more to read documents, summarise and integrate data the old problem with templates is further exacerbated. It becomes easier to spot inconsistencies, errors, typos and mistakes.
AI can use those same templates to help improve standardisation, consistency and reduce mistakes across a team but you need to understand how AI works and get a system in place to use it.
In the rush to adopt AI, lots of firms just started with whatever they had—ChatGPT and Gemini. Teams used it, saw the benefits, and now you know it works well but want to take it to the next level. It's crucial at this point: you can trip yourself up. Here's what we discovered.
Claude, ChatGPT, Gemini, Copilot, Grok — it depends on the work, and you may already be paying for the right one. Most firms have Copilot or licences sitting unused. You need the tool to work for the report, not another project.
Someone's on the free tier on a personal phone, pasting in site notes — weakest protections, no audit trail. A sanctioned tool and five minutes of instruction ends it.
On consumer tiers, what you type may train the model. On business tiers, it generally doesn't. Same logo, different contract — and almost nobody has checked which they're on.
Site addresses, client names, sensitive findings. We cover what can go in, what must be stripped, which settings matter, and how to record the decision.
Clients, tenders and PI insurers now ask how AI was used and controlled. Every firm we train leaves with a one-page usage policy a client or underwriter will accept.
We train your team on your own reports. Send us one anonymised report before the first session and we open by drafting one in your house format, live, in front of your people. Your template, your language, your standards. Not a demo on a made-up example.
Eight hours across three sessions of around three hours each, spread across five weeks — with practice on live jobs in between. The spacing is deliberate. People only keep what they use, so between sessions your team works on real reports and brings the problems back.
We start with your own report. Scribbled site notes, photos on a phone, and voice memos — we show your team what captures cleanly and what needs a different approach, then produce a structured first draft in your house format by the end of the session. Your team leaves able to do it themselves on the next job.
Using AI as a second reader. Catching the section carried over from a previous report, the photo from a different site, the wrong-project reference, the US spelling. We work on real drafts your team produced in the fortnight between sessions.
Standardising output so the report reads the same whoever wrote it. Referencing the standards and regulations you work to, with citations you can check. And the boundary: judgement, rating, sign-off, liability and authority stay with your professional, every time.
The training ends. These don't.
You don't just learn how — we build it with you, live, and hand it over.
That's it. No IT project, no procurement, no new system to roll out.
The training doesn't end when the third session does. You get four follow-up sessions, one every three months for a year — 30 minutes each, with the same people.
They exist because of what happens after training. New starters join. Someone hits a report type we didn't cover. A tool changes and the thing that worked in March stops working in June. A quarterly check-in catches all of that before it turns into the team quietly drifting back to old habits.
Each one is short and practical: what's working, what's slipped, what's new, and one thing to fix before the next call.
No extra cost. It's part of the programme.
The instructor
AI experts don't understand the Built Environment. Built Environment experts don't understand AI.
We know both.
Founder, Skirr AI
Paul has spent his career in the property and infrastructure world — which is why this training sounds like your site notes and house templates, not a tech pitch. Skirr AI is trusted by Scottish Futures Trust, Fife Council and the Scottish Government's building standards programme. He teaches the same practical methods used with those clients: live on your documents, with clear boundaries on where AI stops and professional judgement begins.
Same method, your documents and your standards.
Condition surveys, dilapidations, schedules of condition
FRAs and fire door inspections under the Building Safety Act
Factual and interpretative reports, AGS data
Condition and remediation reporting
Inspection, assessment and design reports
BS5837, protected species, BNG assessments
We also work with asbestos and damp surveyors, drainage and subsidence specialists, M&E consultants, energy and retrofit assessors, architects, and building control.
No, and you shouldn't want it to. AI drafts and checks; your professional signs off and carries the liability. We train the boundary as carefully as we train the tool.
Often not. We start by auditing what you already pay for.
Session one produces a usable draft in your house format. The QA and standardisation gains build over the five weeks.
Tell us what kind of reports you produce and how many people write them. We'll come back with a straight answer about whether training is worth it for a firm your size.
Book a training call to learn more →