What I Handed to AI Last Week (and the 9 Hours It Gave Back)

Last Tuesday I finished a run of shifts, sat down at the kitchen table after the kids were asleep, and did something that would have swallowed my whole evening two years ago. I cleared a backlog of portfolio work, drafted three clinic letters, turned one blog into a week of content, and summarised two papers I’d been meaning to read.

It took me about forty minutes.

I’m not telling you that to show off. I’m telling you because the hours AI gives back are the whole point — and because most doctors I talk to are still using it like a novelty search engine, if they use it at all. So here is exactly what I delegated last week, how I did it, and what it actually meant for my time. No hype. Just the real workflow.

The rule that makes all of this safe

Before anything else, one hard line: no patient-identifiable data goes into a general AI tool. Ever. That single rule keeps me on the right side of information governance, and it shapes everything below. When I work with clinical material, I either anonymise it completely first, or I use approved, contract-covered tools inside the trust’s own systems. Everything I describe here either used fully anonymised inputs or wasn’t clinical at all. Get that boundary right, and the rest opens up.

Job 1: The portfolio backlog (saved ~3 hours)

Reflective practice is the tax every UK doctor pays in unpaid evenings. I had a stack of anonymised cases sitting in note form, waiting to become proper reflections for my CESR portfolio.

So I used AI as a scaffolding tool, not a ghost-writer. I fed it my own bullet points — what happened, what I did, what I’d change — and asked it to structure them against a reflective framework. The thinking was mine. The clinical judgement was mine. What AI did was turn my scattered notes into a clean first draft I could then edit into my own voice. That’s the ethical line, and it matters: it assists your thinking, it doesn’t fake your work. Three evenings of writing collapsed into one focused hour.

Job 2: Clinic and discharge letters (saved ~2 hours)

Documentation is where AI converts a sceptic in a single shift. Working only from anonymised, structured inputs, I drafted the skeleton of several letters — the standard scaffolding of history, findings, plan, and follow-up — then filled in and checked every clinical detail myself.

Here’s what surprised me. The time saved wasn’t mostly in the typing. It was in avoiding the blank page. Starting from a structured draft rather than a blinking cursor removed the friction that makes admin drag. As a result, letters that used to pile up got cleared the same day.

Job 3: One blog into a week of content (saved ~2.5 hours)

This is the one that touches the business side of Building Out directly. I write one long-form article; then I need it to become a week of LinkedIn posts, a teaser, and a recap. Historically that repurposing ate an entire evening on its own.

Now I hand the finished article to a set of prompts I’ve refined over months, and it drafts the micro-content for me. I still rewrite every hook in my own voice, because that’s where the value lives and where AI is weakest. But the mechanical work — pulling out the five best points, reshaping them for a different format — is done in minutes. What used to be a bottleneck stopped being one.

Job 4: Keeping up with the literature (saved ~1.5 hours)

Finally, the reading pile. I took two open-access papers I’d been avoiding and asked AI to summarise the methods, the headline findings, and the limitations, so I could decide fast whether they were worth a proper read.

One caveat here, and it’s non-negotiable: you verify anything factual against the source. AI is excellent at compression and terrible at knowing when it’s confidently wrong. So I use it to triage and orient, never to be the final word on a citation or a number. Even with that check built in, thirty minutes of skimming replaced two hours of grinding.

What this actually adds up to

Roughly nine hours, in one week. Not because AI is magic, but because I stopped doing by hand the parts of my work that don’t need a human, so I could spend my finite evenings on the parts that do — my family, my portfolio, the business I’m building outward from medicine.

That’s the reframe. The question was never “will AI replace me?” The real question is what you’d do with a reclaimed evening a week. Here’s how to start finding yours.

Start here this week

1. Set your boundary first. Write down your own rule: no patient-identifiable data in general tools, full stop. Everything else follows from that.
2. Pick your worst admin job. Choose the single task you dread most — letters, reflections, references — and delegate only that one this week.
3. Use it to draft, never to decide. Let AI beat the blank page; you keep the judgement and check every fact.
4. Build one reusable prompt. The payoff isn’t a clever one-off — it’s a workflow you run again and again with almost no effort.
5. Bank the time on purpose. Decide in advance what the reclaimed hour is for. Time you don’t consciously protect gets swallowed straight back into the job.

For more topics on building a life of time and financial freedom, sign up for our weekly newsletter at www.building-out.com

This post is for educational purposes only and does not constitute financial advice. Always do your own research and, if needed, ask for advice from a qualified financial adviser regulated by the FCA.