I was a sceptic. Every other product update for the last two years has been “now with AI,” and most of it has been wallpaper — a chatbot bolted onto something that didn’t need one. So when AI email drafting landed in my workflow, I assumed it would be the same: a button I’d click once, find unimpressive, and never touch again.

Then I actually started using it. And then I did the maths on the time it was saving me, and the number was so ridiculous I had to check it twice.

The bottleneck I didn’t know I had

Every quote I send needs a covering email. So does every invoice. So does every follow-up. So does every “thanks for the chat, here’s the next steps” message after a sales call.

I never thought of these as a problem. They take a few minutes each. But “a few minutes each” times “twenty of these a week” is over an hour. And the cognitive cost is bigger than the time cost — I’d be deep in a quote, finish it, and then have to context-switch into “polite professional email mode” to write the cover. That switch costs more than the typing does.

What templates didn’t solve

Before AI, my answer to this was templates. I had four or five canned emails: quote sent, quote follow-up, invoice attached, payment received, project complete.

Templates work, but only halfway. The skeleton is fine, but every email needs editing — the client’s name, a reference to the specific job, a line about whatever we just discussed on the phone. By the time I’d opened the template, pasted in the variables, and tweaked the tone, I’d saved maybe 30% of the time over writing it fresh. Better than nothing, but not transformational.

What changed

The AI drafting in stedd.io is doing something different. It’s not picking a template and substituting variables. It’s reading the actual quote — the line items, the prices, the client’s name, the project description — and writing an email that references all of that naturally.

So instead of “Dear [name], please find attached your quote for [project],” I get something like:

Hi Sarah,

Thanks for the chat earlier. I’ve put together a quote for the new shop signage covering the design work, the printing of the two main panels, and the installation. The total comes to £4,650 with a validity of 30 days.

If you’ve got any questions or want to adjust anything, just give me a shout.

I’d have written almost exactly that email by hand. The difference is I didn’t have to. I clicked one button, skimmed the result for thirty seconds, made sure it didn’t say anything weird, and hit send.

The compounding effect

I tracked it for two weeks. In a typical week I send around 25 quotes and follow-ups, plus maybe 10 invoice covers and another 15 various business emails (project updates, payment confirmations, a few “got time for a call?” messages).

Before AI drafting, the average was somewhere between 4 and 7 minutes per email — including the context switch, the typing, the second-guessing of tone, the proofread. Call it 5 minutes on average. Fifty emails a week × 5 minutes = ~4 hours.

After AI drafting, it’s around 30-60 seconds per email. Skim, tweak, send. Fifty emails × 45 seconds = ~37 minutes.

That’s the bit that surprised me. Not the speed of any individual email — the cumulative weekly total. Almost a full working day, given back.

Where it’s not magical

Two caveats, because I don’t want to oversell this.

First, you still have to read what it generates. AI is excellent at sounding right and occasionally wrong about details. I’ve caught it confidently misstating a line item once or twice. Skim-then-send isn’t optional, even when you trust it.

Second, the most personal emails — bad news, a difficult negotiation, a sensitive client moment — I still write by hand. AI is great at the 80% of professional emails that are routine. The 20% that need genuine thought are still on you.

The bonus: project summaries

The other place AI quietly saves me time is summaries. After a long project — dozens of jobs, hundreds of messages, weeks of back-and-forth — generating a clean “here’s what we did and what it cost” summary used to take an evening. Now it takes a click. The AI reads the project, summarises the work and the spend, and gives me something I can paste straight into a wrap-up email or share with the client as a record.

It’s the kind of work nobody enjoys doing but that absolutely needs doing. Removing the friction means it actually gets done.

I went from “AI is overhyped” to “I genuinely don’t want to work without this” in about three weeks. The pitch isn’t that AI is doing my job for me — it’s that the tedious 20% of my day is now 2%, and the time it gives back compounds week after week.

See how stedd.io’s AI features fit into your workflow.