After the Kickoff: Keeping AI Momentum With a Monthly Clinic

AI adoption usually stalls a few weeks after training, because the session ends and the old way of working is still there. A monthly clinic keeps it moving: a standing hour where people bring real work that got stuck, fix it together, and add what they learn to the shared setup everyone's AI reads.

Why does AI adoption stall after training?

Adoption stalls because training is an event and habits are not. People leave a session keen, try AI on a few tasks, hit a poor result or a busy week, and go back to the old way. Without a regular place to ask questions and fix problems, early enthusiasm fades into occasional use by a few people.

The numbers show how thin the support usually is. According to Microsoft and LinkedIn's 2024 Work Trend Index, only 39% of AI users have received AI training from their employer. So many teams start from almost nothing, and one session cannot carry them.

Use also plateaus. BCG's 2025 AI at Work survey found that regular use among frontline employees stalled at 51%, and only about a quarter of them say they get leadership support. Training gets people started; it does not keep them going.

There is a quieter reason too. After a kickoff, each person improves their own prompts in private. What one person learns stays with them, so the team never gets better together.

What is a monthly AI clinic?

A monthly AI clinic is a standing session, usually an hour, where people bring work that AI did badly or that they could not get started. The group fixes it together on screen. Unlike training, there is no curriculum: the agenda is whatever got stuck that month, so every session deals with real work.

The format borrows from medicine's drop-in clinic. People come with a problem, not to sit through a lecture. That keeps attendance honest, because the session is only useful if people bring something.

A clinic differs from a workshop in one more way. A workshop teaches a method; a clinic repairs the setup. When the group finds that every proposal draft gets the pricing wrong, the fix is not "write a better prompt". Instead, it is correcting the shared product facts that every draft reads.

That shared setup matters. It is the business knowledge, written workflows and approved facts that sit underneath everyone's AI. The guide to AI training for employees explains why training sticks better when that layer exists.

How do you run a monthly AI clinic?

Run it on a fixed day, keep it to an hour, and ask people to submit their stuck work a few days before. Spend most of the time fixing two or three real examples on screen. Then close by agreeing what changes in the shared setup and who will make each change before the next session.

A simple agenda that works for teams of 10 to 50:

1. Last month's changes (5 minutes). What was fixed, and did it help?
2. Stuck work (35 minutes). Two or three submitted examples, worked through live.
3. Wins worth copying (10 minutes). One person shows a task AI now does well for them.
4. Changes and owners (10 minutes). Each fix gets a name and a date.

Keep the group small enough to talk. If 40 people want in, run it by team. Ask a champion in each team to collect examples during the month, so people do not arrive empty-handed. The guide to choosing AI champions covers who suits the role.

Choose examples that will recur. A one-off task fixed live helps one person once. However, a weekly report or a standard client email fixed live helps everyone who writes one.

What should come out of each clinic?

Every clinic should leave a record of what changed: an updated shared instruction, a new or corrected written workflow, a fact added to the approved list, or a question only a leader can answer. If a clinic ends with good conversation and no change, the same problems will come back next month.

Decisions for leaders deserve a special mention. Clinics surface the questions nobody has authority to answer: which data AI may see, which tools are approved, who signs off what. Collect them and send them up the same week, because an unanswered question stalls a whole team.

Keep the record short and shared. One page per clinic, in a place everyone can find, is enough.

How do you measure AI adoption after training?

Measure the work, not the tools. Count how many recurring tasks now start with AI, how long a first draft takes, and how much of each draft gets rewritten. Licence counts and logins only show access. Then review the numbers at each clinic so the group can see whether the fixes are working.

Be careful with self-reported speed. According to METR's 2025 randomised trial, experienced developers took 19% longer with AI tools while believing they had been 20% faster. So time tasks rather than asking people how they feel. The guide to AI productivity metrics covers measures that hold up.

Leadership attention moves the numbers too. According to BCG, employees who feel positive about generative AI rise from 15% to 55% with strong leadership support. A leader who turns up to the clinic, even briefly, sends that signal.

When should a monthly clinic change or stop?

Change the clinic when the stuck work dries up or repeats. If people bring nothing, adoption has either settled or stalled, so check the measures to tell which. Once most recurring tasks run on shared workflows, move to a quarterly session and keep a channel open for questions in between.

Repeated problems point to a gap the clinic cannot fix. If the same task fails every month, the shared knowledge is probably missing something, or the workflow needs rebuilding properly.

Skein offers adoption as one of three ways in, alongside strategy and implementation, and a business can begin at any one; the AI facilitation page shows how the adoption side works. That suits a business where AI licences were bought for everyone and are used by a few. A team that has not yet chosen its AI tools needs those decisions first.

Frequently asked questions

Why does AI adoption drop after training?

Training is a single event, while habits form over weeks of daily work. People try AI, hit a poor result or a busy week, and return to the old way. Without a regular place to fix problems and share what works, use narrows to the few people who were already keen.

What is an AI clinic?

An AI clinic is a regular working session, often monthly, where people bring tasks AI handled badly or could not start. The group fixes them together on screen and records what changes in the shared setup. It has no fixed curriculum, so every session deals with real work the team does.

How long should an AI clinic be?

An hour is usually enough for a team of up to about 20 people. Spend most of it on two or three real examples, then agree the changes and who owns each one. Longer sessions tend to drift into general discussion, which helps less than fixing specific work.

Who should attend an AI clinic?

Anyone who uses AI for recurring work, plus at least one person who can update the shared instructions and workflows. A champion from each team helps by collecting examples during the month. A leader should attend at least occasionally, because many stuck problems need a decision only they can make.

How do you know if AI adoption is working?

Look at the work rather than the tools. Track how many recurring tasks start with AI, how long a first draft takes and how much of it gets rewritten. Time real tasks instead of asking people, because self-reported speed gains are often wrong, as controlled studies have shown.

Related reading