How to Choose and Support an AI Champion in a Small Team

Choose the person colleagues already go to when something's awkward, who's careful and patient, not automatically the most technical or the youngest. Then support them with protected time (two to four hours a week to start), a written remit, a small budget, a direct line to you and a monthly check-in.

An AI champion is the colleague who helps everyone else use AI well: answering "how do I…" questions, keeping shared prompts current, and spotting problems before they reach customers. It isn't the same as being accountable for AI in the business; that stays with you, and who should own AI in a small business covers that split. Below you'll find what the champion does and doesn't do, a scoring sheet for candidates, the picks that look right but aren't, a one-page remit, a worked example with time costs, and how to stop the role burning someone out.

Follow me on Instagram@sagnikteaches

What the champion does, and what stays with you

Keep the champion out of enforcement. If colleagues think the champion is watching what they paste into ChatGPT so they can report it, nobody will ask them for help, and the role stops working. Their job is to make the right way easy; making sure people follow the rules is yours.

Connect on LinkedInSagnik Bhattacharya

Scoring your candidates

Most owners pick whoever seems most excited about AI. Excitement helps, but it isn't what makes colleagues change how they work. Score each candidate from 1 to 3 on each line, then multiply by the weight:

Subscribe on YouTube@codingliquids

The maximum is 45. Anyone scoring above about 32 is a strong candidate. Technical confidence carries the lowest weight on purpose: current AI tools don't need technical skill to use well, and the most technical person is often not the one colleagues feel comfortable asking.

Here's the sheet filled in for two candidates at an illustrative ten-person independent hotel, where the owner is choosing between the reservations lead and the marketing assistant who already uses AI every day for social posts:

The marketing assistant wins on three lines and still loses clearly, because the heavily weighted lines are about trust. The reservations lead's weak spot is time, and that's the one line an owner can change: here, by moving her Sunday shift to a colleague so the champion hours had somewhere to live. A low time score is a scheduling problem; a low trust score isn't something you can fix by rearranging the rota.

Then ask them. Don't assign the role. Explain what it involves and the time you'll protect for it, and let them say no without consequence. A reluctant champion is worse than none.

Wrong picks that look right

• The loudest enthusiast. They'll promote AI for everything and may skip checks. Their energy is useful for testing new ideas, but they're often not the person sceptics trust.
• The most technical person nobody asks. Skill without approachability means questions go unasked.
• The most junior person, because "they're young and good with tech". They may well be good with it, but asking a new starter to change the habits of colleagues with ten years' experience sets them up to fail. How it shows up, in an illustrative dental practice that gave the role to a trainee nurse in her first year: her prompts were good and she was generous with her time, but after three months every entry in her help log came from the three youngest staff. The two senior hygienists, whose treatment-plan letters were the task most worth changing, had never asked her once.
• You. In most small businesses the owner is already the bottleneck for decisions. Be the first visible user and the one accountable, and let someone else be the helper.
• Someone already at capacity. If their week is full, the AI work will be the first thing dropped in a busy period.

The remit on one page

Write it down, even in a team of six. A written remit protects the champion's time and stops the role expanding into "anything with a computer in it". Adapt this:

AI CHAMPION: REMIT
Name: [ ]          Start date: [ ]          Review date: [ + 6 months]

PURPOSE
Help everyone here use our approved AI tools well and safely.

TIME
[3] hours a week for the first 8 weeks, then [1.5] hours a week.
Scheduled on [day/slot]. Protected: not the first thing cut
when we're busy.

YOU WILL
- Help colleagues on their own tasks (show, don't do it for them)
- Keep our shared prompts and assistant instructions up to date
- Keep a short log of wins, problems and near misses
- Run a 5-minute AI slot at the [weekly/fortnightly] team meeting
- Do a first check on requests for new AI tools, using our form

YOU CAN DECIDE ALONE
- Changes to shared prompts and instructions
- Trials of free features within our approved tools

BRING TO [OWNER]
- Any new tool or paid feature
- Anything involving customer, pupil or staff personal data
- Any mistake that reached a customer

BUDGET
Up to $[ ] a month for trials, agreed in advance.

NOT YOUR JOB
Enforcing rules, handling conduct issues, doing other
people's AI work for them, or being on call outside your hours.

CHECK-IN
30 minutes with [owner], first [weekday] of each month.

The "bring to owner" list matters most. It tells the champion where their authority ends, which makes them more confident using it. For the tool-request part, a simple form and a set of questions make the first check quick; an AI tool approval process gives you both.

A first check should take the champion 20 minutes, not an afternoon. Suppose a teacher at the music school asks to use a free AI meeting-notes app for parent consultation evenings. The champion reads the app's privacy page and pricing page and writes three lines to the owner: "It records and transcribes the call. The free plan's terms allow it to use recordings to improve its service. Parents' comments about their children would go into it. My view: no on the free plan; if we want this, we'd need a business plan with training off, and parents told in advance." The decision stays with the owner, but it's now a two-minute one, because the champion found the one fact that mattered.

What to hand over on day one

Champions often stall in their first fortnight because they can't actually change anything. Before they start, give them:

• Access: edit rights to the shared prompts, projects or assistants, and whatever admin role your AI tool offers for managing shared content (not billing).
• The current picture: which AI tools the business pays for, who has seats, and any usage figures from the admin reports.
• The rules: your AI usage policy or, if there isn't one yet, the three or four data rules you've agreed.
• A baseline: how long the main target tasks take now, so the champion can show progress later.
• An announcement: tell the team yourself who the champion is, what they'll help with and that their time is protected. It gives them standing from the first day.

The wording of that announcement shapes how colleagues treat the role. A common first version, sent as a one-line message: "[Name] is our new AI person, so ask her if you need anything." That sounds like a help desk with no hours and no limits. A better version, said at a team meeting and then sent round (illustrative):

"[Name] has agreed to be our AI champion for the next six months. She has three hours a week set aside for it, on Tuesday afternoons, and that time is protected. She'll help you use AI on your own tasks, starting with lesson notes, and she'll keep our shared prompts up to date. She won't do the work for you, and she isn't checking up on anyone. Questions about what's allowed still come to me."

The last two sentences do the most work. They tell the sceptics she isn't policing them, and they tell the champion where her authority ends.

Worked example: a music school's first champion

Take an illustrative music school with nine teachers, a part-time administrator and an owner who also teaches. The school uses AI for lesson notes to parents, practice plans, timetable emails and its social posts, but only three teachers use it regularly.

Choosing. The owner scores three people. The keenest AI user, a younger guitar teacher, scores 28: very curious, but colleagues rarely ask him for help. The administrator scores 30, held back by time: she's already stretched. A piano teacher who's been there eight years scores 36: everyone asks her about the booking system, she's meticulous with reports, and the two most sceptical teachers respect her. She agrees, on condition that the time is real.

Time and cost. Three hours a week for eight weeks, then 1.5 hours a week, all in her non-teaching hours and paid at her usual rate. At an illustrative $30 an hour, the first eight weeks cost about $720, then roughly $195 a month. The owner adds a $40-a-month trial budget.

Her first eight weeks:

• Weeks 1 to 2: 15 minutes with each teacher to find their most repetitive writing task. Five of the nine say lesson notes to parents.
• Weeks 3 to 4: writes and tests a shared lesson-notes assistant on 20 real notes, and a practice-plan prompt that asks for the pupil's grade level and the pieces they're learning.
• Weeks 5 to 6: sits with the six teachers who weren't using AI, one at a time, on their own notes. Starts a five-minute slot at the fortnightly staff meeting.
• Weeks 7 to 8: checks a sample of notes, finds that two teachers' notes still include pupils' surnames, and raises it with the owner rather than with the teachers, as agreed in her remit.

At the week-eight check-in: eight of nine teachers are using the lesson-notes assistant weekly. Teachers estimate they each save about an hour a week, so roughly eight hours a week across the school, against the champion's 1.5 hours a week from then on. The owner reminds everyone of the first-names-only rule at the next meeting, and the champion adds a line to the assistant's instructions: "If a surname appears, remove it and say so."

The monthly check-in: a 30-minute agenda

1. What people used AI for this month: two or three examples, including one that surprised you.
2. Problems and near misses: anything wrong that was caught, and anything that wasn't.
3. Tool requests: what colleagues asked for and the champion's first view.
4. Prompts and instructions: what was updated, and what nobody uses and can be retired.
5. What the champion needs: time, access, a decision from you, or support with a colleague.
6. One thing to stop doing: every role accumulates tasks; drop one.

Items 1 and 2 are only as good as the champion's log. It needs four columns and a line per entry, kept in a shared sheet. A few rows from the music school's second month (illustrative):

The "Question" rows are worth counting. If the same question turns up three times, it belongs on the rule sheet, not in the champion's inbox.

The notes from a check-in needn't be long. The music school's month-three notes might look like this (illustrative):

CHECK-IN, month 3 (30 min)

1. Used for: end-of-term report first drafts (all 9 teachers).
   Surprise: brass teacher turned the grade syllabus into a
   12-week practice calendar for his exam pupils.
2. Problems: a practice plan listed a piece that isn't on the
   current syllabus. Caught by the teacher. Fix: prompt now
   says "use only the pieces I list".
3. Tool requests: one teacher wants an AI notation tool.
   Champion's view: free tier, one teacher, no pupil data.
   Owner: yes, 30-day trial.
4. Prompts: retired the social-post prompt (unused 6 weeks).
5. Champion needs: her Tuesday slot was lost twice to cover
   lessons. Owner: cover comes from the supply list, not her.
6. Stop doing: the fortnightly meeting slot goes monthly.

Item 5 is the one to act on straight away. Champion time that gets borrowed "just this once" is how the role quietly disappears.

Every quarter, add a sixth question: is the remit still right? The role in month one (getting people started) is different from month nine (keeping things tidy and assessing new tools), and the time you protect should change with it.

Keeping the champion from burning out or becoming a bottleneck

A successful champion faces two risks: too much work, and too much knowledge held in one head. Watch for these signs:

• They're doing colleagues' AI tasks for them instead of showing them how.
• They're the only person who knows how a shared assistant or automation is set up.
• Their champion hours creep into evenings, or disappear entirely in busy weeks.
• They start to sound resentful ("everyone just comes to me").

The second sign tends to reveal itself at the worst moment. Picture a small gym whose champion built the automation that sends new members their induction booking link. She went on a fortnight's holiday; a password change in the booking system broke the link on day two, and nobody else knew where the automation lived or which account owned it. Eleven new members joined that fortnight and none got a booking link until she came back and fixed it in five minutes. A one-page note of what she'd built, where it ran and whose login it used would have let anyone fix it on day two.

And the fixes:

• "Show, don't do" as a rule, written into the remit. Doing it for someone once is fine; the second time, sit beside them while they do it.
• Everything written down in shared places. Prompts, settings and notes live in a shared library, not a personal account. Building a shared prompt library shows how to structure it.
• A named backup who knows the basics and can cover holidays.
• A clear end date: review at six months, and consider rotating the role after a year so it doesn't become a permanent extra job.
• Visible recognition: credit at team meetings, the role on their job description, and a proper conversation about pay or title if it grows.

If the team is struggling to use AI at all, a champion helps, but can't fix everything alone. The blockers covered in getting staff to actually use AI tools are worth clearing first, and as the number of tools grows, the decision rights set out in AI governance for a small business keep the champion's role from quietly turning into yours. For a first team session the champion can run, a 60-minute AI workshop is a ready-made format.

More questions about the champion role

Should the AI champion be paid extra for the role?

At minimum, the hours must be paid and protected, not squeezed into lunch breaks. Whether to add an allowance depends on how you reward other extra responsibilities, such as first aid or key-holding. If the role grows into something substantial after six months, recognise it in pay or title; an unrewarded champion tends to quietly stop.

What if two people both want to be the champion?

In a team of ten or more, having two can work well: split by area (one for customer-facing tasks, one for back-office) or make one the backup. In a very small team, pick one using the scoring sheet and give the other a named part, such as testing new tools or maintaining one set of prompts.

What happens if the champion leaves?

That's the test of whether the role was set up well. If prompts, settings and notes live in shared places rather than in one person's head or personal account, a successor can take over within a few weeks. Make handover notes a standing part of the monthly check-in so there's never much to reconstruct.

Further reads

• AI Change Management for Small Teams: A Practical Plan — Where the champion fits in a team-wide rollout.
• How to Train Staff to Use AI in a Small Business — Training the champion can run for colleagues.
• Shadow AI: Is Your Team Using AI Without Telling You? — What the champion should notice and bring to you.
• AI Error Log: Track Mistakes and Stop Them Happening Again — A simple log the champion can keep up to date.
• Automation Audit: Find the Zaps and Scenarios Nobody Owns — A useful job for a champion once automations exist.
• How to Redesign Job Roles Once AI Handles Routine Tasks — Think about how the champion's own role changes.
• Example AI Roadmap for a 12-Person Business, Month by Month — A 12-person driving school's first AI year, month by month: five workflows, about $30 a month in new tools, one idea postponed and why.
• How to Talk to Staff Who Fear AI Will Take Their Job — Prepare your honest answer, then use a six-part conversation outline, better phrasing and a physiotherapy clinic example to talk it through.
• How to Survey Your Staff Before an AI Rollout (With Questions) — Fifteen ready-to-use questions, an invitation email and anonymity settings for surveying a small team before you choose any AI tools.
• Reduce Owner Dependency: Use AI to Capture What Only You Know — Get what's in your head into a form staff can use: an interruption log, an AI interview prompt, decision rules with thresholds, and a shared assistant.
• How Much Time and Money Does AI Staff Training Take? — Training hours by role, free and paid course options with current prices, a costed plan for an eleven-person decorating firm, and how to check it paid off.
• How to Use AI on Your Own Admin First, Then Roll It Out — A six-week owner-first plan: pick three of your own admin jobs, log what AI saves and gets wrong, write a playbook, then hand one task to one person.
• Why Your AI Pilot Stalled, and How to Get It Live — A one-hour diagnosis for a stuck AI pilot, the fix for each of five causes, a 30-day restart plan, and when stopping is the better call.
• Barriers to AI Adoption in Small Businesses and How to Clear Them — Eight practical barriers that keep small firms from adopting AI, the cheapest way past each, and the few barriers you should respect rather than clear.
• How to Keep Up With AI in 30 Minutes a Week — A timed weekly routine for owners: scan your own tools' release notes, filter hard, test one change on real work, and log what you decide.
• Why AI Adoption Stalls in Accounting Firms, and How to Restart It — The five reasons AI stalls in accounting practices, a twenty-minute diagnosis, and a 30-day restart timed around the deadline calendar.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.