Every solo founder eventually hits the point where email stops being correspondence and starts being a second job. This is a practical order of operations for getting it off your plate — including the parts you should refuse to automate.
Disclosure: we build Mailient, which does several of the things described here. The sequencing below holds regardless of what you use, and the first two steps cost nothing.
Almost Everyone Starts in the Wrong Place
The instinct is to automate writing first, because writing is the part that feels like work. You notice yourself typing a similar reply for the fourth time this week and reach for a template or an AI drafting tool.
It helps less than you expect, because writing was never the bottleneck. The bottleneck is that you have to read everything to find the things worth writing about. Speeding up the writing leaves the reading — and the reading is the part that consumes the day.
The reframe: your inbox does not cost you time because replies are slow to write. It costs you time because deciding what deserves a reply requires reading all of it.
The Four Jobs Hiding in “Email”
Separate them, because they automate differently and in a specific order:
- Filtering. Removing what should never have reached you — newsletters, receipts, notifications, cold outreach.
- Deciding. Of what remains, working out the handful that genuinely need a human today, and why.
- Drafting. Writing the replies.
- Coordinating. Booking the meeting, chasing the thread that went quiet, remembering the thing you promised on Tuesday.
Most tools do exactly one of these and are marketed as though they do email. Knowing which one you are buying is most of the purchasing decision.
The Right Order
1. Filter, ruthlessly and for free
Before buying anything: unsubscribe properly rather than deleting, filter receipts and notifications straight out of the inbox, and turn off every product notification that is not an outage. This is an afternoon and it removes a surprising share of the volume. Do not pay a subscription to have software cope with noise you could have stopped at the source.
2. Automate deciding
This is the highest-leverage step and the one people skip. Something has to read everything that survived step one and tell you what actually matters — with a reason attached, because a shortlist you do not trust is just a second inbox.
Filters cannot do this. They match on sender and keyword, so they cannot tell that this particular message from a familiar domain is the one that matters, and they need you to have anticipated the message before it arrived. This is the job that needs something that reads.
3. Automate the repetitive drafting
Now drafting is worth automating, because it is pointed at the right messages. The bar: a draft is only useful if you would send it after editing one sentence. If you would start over, it cost you time rather than saving it — and a mediocre draft is worse than a blank page because you have to read it before rejecting it.
4. Automate coordinating
Scheduling and follow-ups last, because they need the most context — your calendar, what was agreed, who owes whom. Done well this is the step that quietly recovers the most revenue, since the thread that went silent is the one nobody notices losing.
What Never to Automate
Being fast is worth nothing on these, and being wrong is expensive:
- Apologies. A templated apology is worse than a late one.
- Pricing negotiations. Judgement, not throughput.
- Anything legal. Obvious, but tools will cheerfully draft into it.
- Your biggest customer, on anything that matters. The time saved is not worth the fraction of a chance it reads as automated.
- Anything needing what is only in your head. No tool can produce a fact it has never been told.
A good assistant should decline some drafts. One that confidently writes a reply to a legal notice does not understand what it is looking at, and that is worth testing for deliberately before you trust it with anything.
The Delegation Test
When unsure whether a task should be automated, ask what you would do if a competent new assistant had handled it on their first day.
- “Fine, thanks” → automate it.
- “Check with me first next time” → automate the drafting, keep the approval.
- “Please never do that again” → never automate it.
Almost everything in a founder's inbox lands in the middle category, which is why approval-by-default matters more than any feature comparison. The value of delegation is that it stops being your job; that only works if you are not double-checking everything, and you will double-check everything if the thing can act without you.
A Setup That Works
- Gmail filters for predictable noise. Free, an afternoon.
- Something that reads and decides — the step that actually returns your day.
- Drafting in your own voice, aimed at what step two surfaced.
- Calendar connected, so scheduling stops taking six messages.
- One review window a day. Automation fails if you keep checking anyway — the tooling removes the work, the habit has to remove the checking.
Mailient does steps two through four on a schedule and hands you one briefing — how it works, or start free and read real drafts to your real threads before paying. Step one is still yours, and still worth doing first.