Building AI Into Your Workflow

Until now you've used AI for individual tasks. The shift now is to treat it as a system woven through your week, not a tool you grab in a panic. Map what you do: open your calendar and a blank table, list your recurring tasks, and score each for repetition and time:

The tasks that rise to the top share three traits: they happen often, follow a predictable shape, and are more assembly than judgement. Those are your prime candidates. Tasks needing your accountability or sign-off stay human — AI can draft, but you decide. You can even hand the list to AI: "Rank these by how suitable each is for AI help, and say whether it should draft, assist, or stay out."

A prompt template is a saved, fill-in-the-blanks request for a task you do repeatedly. Instead of reinventing the wording each week, you paste a proven template and swap in today's details — the highest-leverage habit in this course. A good one has four parts: a role, the task, the context (with a gap to fill), and the format:

> You are my project communications assistant. Write a weekly status
> report for [PROJECT NAME], for a busy executive sponsor on their phone.
>
> This week's raw notes: [PASTE YOUR ROUGH NOTES HERE]
>
> Sections: Headline (are we on track?); Progress (3-4 bullets); Risks
> (each with an owner); What I need from you. Under 200 words, no jargon.

The bracketed placeholders are the only parts you change each week. Build a small library for your top tasks — status report, variance explanation, stakeholder reply, meeting summary, proposal draft — kept in one document, each clearly titled.

AI starts every conversation knowing nothing about you — your company, team, or style. A personal knowledge base is a short reference you paste in (or load into your tool's memory or "custom instructions" feature) so AI fits your world without you re-explaining it. Think of it as the briefing you'd give a new contractor — a page or two covering:

Paste it at the start of a session:

> Here is background about me. Use it to inform everything you draft.
> ROLE: Operations manager for the EMEA logistics team.
> ORG: Mid-size distributor; our audience is warehouse leads.
> STYLE: Warm but concise. British spelling. No exclamation marks.
> AVOID: buzzwords like "synergy" and "leverage".
> KEY TERMS: "DC" means distribution centre; "PO" means purchase order.
> Confirm you've understood, then wait for my first task.

From there, every draft lands closer to ready. Many tools let you save this in a memory or settings area so you needn't paste it each time.

An "AI-ready" input is a little preparation you do before asking, so the request needs far less back-and-forth. A few minutes tidying inputs saves much editing later.

The pattern: give AI the raw material, the goal, and the audience. A few habits make every input AI-ready — keep notes in plain text, paste data as a table (not a screenshot), name the audience in one line, and hand over the source rather than describing it.

A touchpoint is a planned moment where AI does part of the work. The skill is placing them where they help, not where they add risk. For each task, ask how costly is a mistake, and how easily can I catch one? That gives four zones:

A first draft of an internal email is low-cost and easy to check — automate it. A figure in a board report is high-cost but easy to verify — let AI draft the narrative, but check every number. A message to an upset client is hard to verify by tone — AI can suggest, but you own every word.

Now lay your high-opportunity tasks, templates, and knowledge base onto a real weekly plan, each touchpoint paired with a review step sized to its risk:

Three principles keep it sustainable: anchor AI to moments you already have so there's no new habit; keep a human first and last — you set the intent and approve the result; and review monthly, dropping touchpoints that didn't earn their place. You can even paste the draft back to AI and ask for a leaner version you'll actually stick to.