How I Use AI to Save 6 Hours a Week (After 15 Years in Fintech)

I've spent 15 years in regulated finance — banking, fintech, the kind of work where a sloppy summary of a compliance document can cost real money. For most of that time, "productivity" meant working later. Then I changed how I use AI, and I genuinely got six hours of my week back. Not from a magic tool. From a handful of prompts I treat like saved templates, and one shift in how I think about the whole thing.

This isn't a "10 prompts to 10x your life" listicle. It's what actually moved the needle for me, the mistakes that wasted my time first, and the system I now keep so I never start from a blank page again.

When I started using AI seriously, I did what most people do: I typed short, vague requests. "Summarize this." "Write an email about the project delay." The output was always the same — technically fine, completely generic, and useless for my actual work. I'd spend more time rewriting the AI's bland draft than if I'd written it myself.

The mistake wasn't the tool. It was me treating it like a search engine instead of a colleague. A new colleague who's brilliant but knows nothing about your context will give you mush if you give them mush. Tell them who they are, what you need, what to avoid, and how to deliver it — and suddenly they're useful.

The day I understood that, my prompts went from one line to four parts, and the output became something I could actually send.

Every prompt I rely on now has the same skeleton. I don't deviate from it, because it's the reason the output works:

• Role — who the AI should act as ("a compliance analyst", "a skeptical investor", "a product manager")
• Task — the specific thing to do
• Context — the messy reality I paste in
• Format — exactly how I want the answer back

That's it. It looks simple. It's the difference between a tool that wastes your afternoon and one that hands you back your afternoon.

Here are the ones I reach for most. I've stripped out anything confidential, but the shape is exactly what I use.

1Turning a 30-page document into something I can act on

In my world, you're constantly handed dense documents — regulatory updates, vendor contracts, internal specs — and expected to know what matters in them by tomorrow. This used to eat half a day.

"You are a compliance-minded analyst. Read the document below and give me: (1) the 3 things that change what we have to do, (2) any new obligations or deadlines, (3) anything ambiguous I should flag to legal. Be concise and specific — no generic summary. Document: [paste]."

What used to take me three hours of careful reading now takes about twenty minutes — I read the AI's flagged points, then go deep only where it matters. The key is the role ("compliance-minded") and the demand for specifics, not a bland recap.

2Drafting the email I'm dreading

Every job has the difficult message — the delay, the pushback, the "no." I used to procrastinate on these for an hour, then write them badly anyway.

"You are a calm, professional communicator. Write a short email to [recipient] about [situation]. Be honest but tactful, acknowledge their concern, give one clear next step. No corporate filler. Keep it under 120 words."

Two minutes, and I have a draft that's 80% there. I tweak the tone and send. The "under 120 words" and "no corporate filler" constraints are doing the heavy lifting — without them, AI writes like a press release.

3Pressure-testing my own thinking

This is the one nobody talks about, and it's the most valuable. Before I commit to a decision — a product direction, a vendor choice — I have the AI argue against me.

"Act as a skeptical [investor / head of risk]. Here's what I'm planning: [paste]. Give me the 3 strongest reasons this could fail, and the one thing I'm probably underestimating. Be blunt."

It's like having a sharp colleague on call who isn't afraid to tell you you're wrong. Half the time it confirms my plan. The other half, it catches something that would've cost me weeks. That's worth more than any time it "saves."

Here's the counterintuitive part. For a while I hoarded prompts — hundreds of them, screenshotted from Twitter, saved in random notes. I almost never used them. The collection itself became the procrastination.

What actually works is the opposite: a small set of prompts you've shaped for your own work, that you reuse until they're second nature. I use maybe fifteen regularly. They live in one place, organized, so I never dig through chaos to find the one I need. That's the entire system — not more prompts, but the right ones, ready to go.

If you take one thing from this: stop typing one-line requests, and start giving the AI a role, your context, and a format. That single change will do more for your output than any tool you could buy.

Either way: give the AI a real brief, not a vague wish. That's the whole game.

Written by the founder of ProAI Hub — practical Notion systems for people who actually use AI to get work done.