How to actually use AI better at work (skills, not slogans)

Everyone has met both kinds of teams by now. One uses AI and ships more, faster, with fewer errors. The other — sometimes with the same tools and licenses — produces work that looks done and isn't: the polished-but-hollow draft, the confident wrong citation, the memo someone has to rewrite at 6pm. Stanford/BetterUp put a number on it: 40% of US workers received “workslop” in the past month (HBR, Sept 2025).

The difference isn't talent or the model. It's six trainable habits — the ones our coaching engine builds into practice areas. Here they are, in the order they pay off.

Habit 1: Provision context before the ask

Weak drafts are almost never model failures — they're context failures. Before “write the update,” feed it the audience, the goal, the constraints, and what good looks like: “Draft the client update for a CFO who has five minutes; lead with the number that changed; no jargon; 150 words.” A sourcer pasting a résumé into a chatbot without the job order gets a generic screen — and a hiring manager who can tell.

Habit 2: Pick the right task (respect the fit)

AI is a first-draft engine, a summarizer, a transformer of formats, a role-player for hard conversations. It is a bad oracle for facts it can't verify and a terrible place for your only copy of anything. Match the task to the trade: draft = yes; decide = draft the decision memo, then decide yourself.

Habit 3: Attribute and verify by default

Anything that arrived with a claim gets a check before it leaves your name: numbers to the source, quotes to a person who said them, citations to real documents. Nothing torches client trust faster than a confident fake. The rule of thumb our coaching uses: if you didn't verify it, don't send it as fact.

Habit 4: Keep the data rules visible while you work

The habits above fail silently when the workflow invites over-sharing — the fastest prompt is the one that pastes the whole file in. Good teams make the data rule present at the moment of use: what's green to paste, what needs stripping. (That's what in-browser guardrails do — the rule shows up as you type, not in a training from last quarter.)

Habit 5: Iterate out loud, then keep the winner

Brief it, critique what came back, re-ask sharper. The second prompt after a miss is where quality lives — and a team prompt library of what worked outgrows every tips article ever written.

Habit 6: Practice where the work happens

Courses can't carry these habits into a Thursday afternoon; moments of use can. Free daily practice: CoaPro Daily gives one work scenario a day — write the prompt, see it scored, get the lesson. In-product: CoaPro's coaching nudges land while the employee writes, tied to your own policy.

The owner's takeaway: these six habits are the whole difference between AI that costs and AI that pays — and none of them require a new license. They require practice, in the flow, with guardrails that make the safe way the fast way. See the coaching live.