Communication channels: the unglamorous infrastructure of AI adoption

Here's a statistic that should worry anyone running an AI programme: Microsoft's Work Trend Index found that 52% of people who use AI at work are reluctant to admit using it for their most important tasks^[1]. Over half. In firms everywhere, the most instructive AI usage is happening in silence — no one learns from it, no one improves on it, and governance has no idea it exists.

Silence is the default state. People assume they're behind everyone else, worry that using AI looks like cutting corners, and aren't quite sure what's sanctioned. The cure isn't a policy PDF. It's communication infrastructure — visible, low-friction, running on a rhythm. In my experience it's the single cheapest lever in the whole adoption stack, and the most commonly skipped.

01The channel stack

What good looks like, concretely — most of this can live in whatever chat platform the firm already runs:

On leaderboards: they work, with one caveat. Celebrate outcomes and generosity — best shared skill, most useful win posted, best answer in #ai-help — rather than raw message counts, or you'll incentivise noise. The point is to make the behaviour you want visible and high-status. Public recognition in a monthly note costs nothing and reliably outperforms formal incentive schemes at this.

02Why this works: the flywheel

The research says the wheel spins on peers, not memos. MIT's study of enterprise deployments found adoption succeeded when driven through line managers and the people doing the work, not broadcast from a central lab^[2]. BCG's 70%-is-people finding points the same direction: communication and change management aren't the wrapper around an AI programme, they're most of its substance^[3]. A channel where a respected associate posts "this saved me two hours" is line-manager-driven adoption, operationalised — one post at a time.

Nobody was ever converted to a new way of working by a policy PDF. They were converted by watching someone they rate do their job faster — and being able to ask, in a low-stakes channel, "how did you do that?"

03The governance dividend

There's a compliance argument hiding in here too, and for regulated firms it may be the strongest one. That 52% of hidden usage^[1] isn't just a lost learning opportunity — it's shadow AI: unsanctioned tools, unknown data flows, zero audit trail. Microsoft found 78% of AI users bringing their own tools to work^[1], and a firm with poor comms channels has no idea which 78% of its staff that is.

Good channels reverse the flow. When it's easy to ask "can I use X for Y?" and the answer arrives in minutes — from a bot grounded in policy, or a champion in #ai-help — people ask before acting instead of guessing. When usage is celebrated in the open, it happens in the open, where it can be governed. The same infrastructure that spreads adoption gives risk and compliance real visibility. It's one of the rare investments that the CTO, the COO and the compliance officer should all want for their own reasons.

04Getting started, in order

1. Open #ai-help and #ai-wins this week. Two channels, zero budget. Seed them: have your champions post three real wins before announcing anything.
2. Name the champions. One respected person per team, formally recognised, with time carved out. They are the immune system of the rollout.
3. Ship newsletter #1 within the month. Short, honest, with a leaderboard and one great prompt. If it reads like corporate comms, rewrite it.
4. Stand up the Q&A bot when the docs are ready. A bot grounded in your own policy and training material — this one's a genuine build, and worth doing properly. (Conveniently, it's also a first-rate hackathon use case.)

None of this is glamorous. All of it decides whether the expensive parts — the licences, the builds, the training — actually convert into changed behaviour. Comms design is part of every embedding engagement I run, because a tool with no conversation around it is just an unused tool waiting to be declared a failed project.

If your rollout has gone quiet, this is usually why. Happy to talk it through.