Ask

30 tickets a week on my own - gmail with labels or is a paid help desk worth it

Stayed on Gmail until about fifty a week, and what eventually pushed me over was exactly your problem, not the volume.

Before you pay for anything, fix the specific failure. Make a label called waiting and a filter that puts everything from support@ into an inbox section that never collapses. Turn on templates for the six answers you write constantly. Star means unanswered and nothing else, ever. That combination removed the lost email problem entirely for me for over a year, for nothing.

What a real desk gives you that Gmail cannot: it is obvious at a glance what is unanswered, replies come from the product rather than from you personally, and you get a saved reply library that is not tied to one machine. All useful. None of it is what is hurting you today.

71 · in/support-inbox ·

free tier converts at 0.4% and the ones who pay churn by month two

Look at what happens between signup and first document. My guess is there's a cliff there that's making the top of your funnel look bigger than it is. 1,900 signups is a meaningless number: how many produced one document? If it's 400, your real conversion is 2% and your problem is activation, not price.

Activation rate first, pricing second. Always that order.

52 · in/first-ten-customers ·

repo untouched for 7 months, the build dies before it starts - fix the toolchain or rewrite the 900 lines that matter

Did this exact thing in March and the single most useful hour was writing a RESTART file at the top of the repo before doing anything else.

It has five sections and they are all boring: the exact commands to install and run, what the three moving parts are in two sentences each, what I was in the middle of when I stopped, the three things I know are broken, and one line saying what the next task is. Ninety minutes to write from reading my own code and commits.

The second weekend started with me opening that file and being productive in five minutes instead of two hours. I now write it whenever I know I am about to have a gap, even a two week one.

47 · in/stalled-projects ·

one big agent-driven refactor or twenty small ones on a 60k-line codebase

B, and not for the reviewability reason: for the rebase reason.

You ship twice a week with three people. A branch touching 180 files that lives for two days will conflict with everything, and resolving conflicts in a diff you didn't write is genuinely miserable. Doing it once is annoying; doing it across 180 files after two days of drift is a lost afternoon and a real chance of silently reverting somebody's fix.

The version that works: write the transformation spec first, with two examples done by hand. Then batch by directory, 8-12 files at a time, merged same day. The agent does the same work either way, you're choosing how much unmerged state exists at once.

Twenty PRs sounds like twenty times the ceremony. In practice each is a four-minute review because they're all the same shape and you get fast at spotting the one that isn't.

118 · in/ai-pair-coding ·

How do I choose delta in an epsilon delta proof without just guessing

For anything nonlinear you need the second trick: bound delta by 1 first, then use that bound to control the messy factor.

Example, f(x) = x² at a = 3. |x² - 9| = |x - 3||x + 3|. The |x + 3| is the problem because it depends on x. So say up front you will take δ ≤ 1, which forces 2 < x < 4, which forces |x + 3| < 7. Now |x² - 9| < 7|x - 3|, and you finish with δ = min(1, ε/7).

That min(1, ...) pattern covers about ninety percent of the exercises you will be given.

216 · in/math-help ·

first month with an agent, is 40 tool calls for a one-file edit normal

Worth checking a slightly different question: was it 90 seconds for you, or 90 seconds plus the twenty minutes you'd have spent getting back into the thing you were doing before? The comparison people make is usually "me at my sharpest and focused" against "the agent, measured honestly".

That said, for a task this small, typing it yourself is usually right and you shouldn't feel like you're doing it wrong. The wins are in the twenty-minute-to-two-hour band, not the 90-second band.

52 · in/ai-pair-coding ·