M.A.I. Consulting
Services

AI training that changes the work, and proves it.

Each department leaves with a working playbook built on its own tasks and documents, and the rules that make it safe to use. Five services, each with the situation it addresses, the result you hold at the end, worked examples and a price.

Why this works quickly

Controlled studies keep finding the same thing: the gain is large, it arrives fast, and it is largest for the people who were struggling.

Published field experiments, not our results. They matter because they isolate the variable that decides whether an organisation gets the gain at all: which tasks are handed over, and how the output is checked.

40%

less time on professional writing tasks, with output quality rated 18% higher, in a randomised experiment with 453 college-educated professionals. Science, 2023

14%

more issues resolved per hour across 5,179 customer support agents, rising to 34% for the least experienced staff. NBER, 2023

25%

faster on 18 realistic knowledge-work tasks, with 12.2% more tasks completed at higher quality, among 758 consultants. HBS working paper, 2023

The same study found the opposite effect on the wrong task. On one task chosen to sit outside what the tool does well, consultants using AI were 19 percentage points less likely to reach the correct answer than colleagues using none. HBS working paper, 2023 That single finding is the argument for training: the difference between a large gain and a costly error is knowing which tasks to hand over and how to check what comes back. It takes one session to teach and it is the whole of what we teach first.

We do not publish a return-on-investment percentage for our own engagements. We have not measured enough of them to state one, and a figure quoted before your documents have been seen has no basis. What we do instead is time the tasks on each department's page before the lab and ninety days after, so the number you end up with is your own.

The situation, with sources

Adoption is already high. The method around it is not.

Use is routine and individual

92% of nonprofits report using AI, 81% use it individually without shared workflows, and only 7% report major improvements in organisational capability. 2026 benchmark of 346 nonprofits

Few staff have a working grasp of it

Across 215 foundations and 451 nonprofits, almost two-thirds report that none or just a few of their staff have a solid understanding of AI and its applications. CEP, 2025

The barrier named is training

Among UK charities, 76% now use AI tools, up from 61% a year earlier. Small charities name limited digital skills (48%) and lack of training (44%) as their biggest barriers. Charity Digital Skills Report 2025

Five services

Bought singly, combined, or as one of three packages.

Method, worked examples and price sit on each service's own page, which opens in a new tab.

01
2–3 weeks

AI Readiness Assessment

SituationThe board or a funder has asked where the organisation stands, and no measured answer exists.

ResultA score across ten dimensions, an inventory of the tools in use and a ranked 90-day plan. You answer with a figure and a course of action.

Method, examples and price →

02
3–4 weeks

AI Use Policy

SituationStaff use AI on donor reports and proposals without agreed rules, or under a prohibition that moved the use out of view.

ResultA policy the board approves and staff follow, because it states what is permitted, with which data, on which tools.

Method, examples and price →

03
2–6 weeks

Team AI Training

SituationStaff have completed general courses and still ask what is appropriate in their own role.

ResultA working playbook for each department, built on its own tasks and documents, plus the attendance record and the material you keep.

Method, examples and price →

04
3–12 months

Fluency Support

SituationSix months after training the tools have changed, the champions have moved on and nobody has re-measured.

ResultMonthly office hours, updated pages, coached champions and a re-score that shows the board what moved.

Method, examples and price →

05
2–6 weeks

Custom Agents, Skills & Tools

SituationThe same task is performed by hand each week, differently by each person, with results that depend on who did it.

ResultThe repetitive part configured once on the platform you already pay for, tested on your documents and cleared against your policy.

Method, examples and price →

The method

One framework underpins all five.

Delegation, Description, Discernment and Diligence: the AI Fluency framework published by Anthropic with Rick Dakan and Joseph Feller. source It is public, so the method can be checked against it.

Delegation

What to hand to AI, what to keep human, and who decides.

Description

How to ask, with your organisation's context already in the request.

Discernment

How to check an output before anyone relies on it.

Diligence

What to log, what to disclose, and who owns the result.

The same four words are the axes of your score, the sections of every playbook, the clause groups of your policy and the modules of every session, so a gap found in the assessment maps directly onto a policy clause and a training lab.

Three packages

Ready-made scopes. Adoption covers the full situation.

Foundation

Up to 50 staff · a first step

ResultA baseline score, one shared page everybody works from, and a first session that puts the same rules in front of all staff.

ContainsReadiness assessment in short form, one organisation-wide session, one shared playbookPlaybookThe working reference a department leaves the lab with: the tasks to hand to AI and the tasks never to, the prompt patterns tested on that team’s own files, the checks before anything is sent, the red lines for its data, and what has to be logged. Written with the team, cleared against your policy, and revised as the tools change.. About three weeks.

Estimate this scope

Institution

Regulated, audited or multi-site

ResultA documented file mapped to the frameworks your counterparts cite, and trainers inside the organisation who can run the sessions without us.

ContainsEverything in Adoption, plus a governance gap plan, an evidence fileEvidence fileThe assessment report, the policy, the tool inventory, the attendance record and the two scores, kept together. It is what a funder or an auditor asks to see. It is documentation, not a certification., train-the-trainer, twelve months of support, and delivery in English, French or Arabic. About twelve weeks, then twelve months.

Scope on a call

Individuals and representatives

Courses for fewer than ten participants

The general module taught on the participants' own tasks, in person or online, priced per person. Each participant leaves with a working playbook for their own role and an attendance record. In-person courses include accommodation and hospitality arrangements.

See the per-person price

Scope and limits

What this practice does not do

No model fine-tuning, no data pipelines, no custom software. We configure agents, skills and tools on platforms you already use, and stop there.

We do not offer regulatory compliance services, certification or legal advice. Our documents use the vocabulary of recognised frameworks so that they are legible to your counterparts, and where a legal or compliance opinion is needed we will say so rather than provide it.

A thirty-minute call establishes which of the five you need, or that you do not need us yet.

Book it, or run the estimator first and bring the range with you.