Playbooks · 5 min read
Should a Six-Person Startup Automate Its Instagram DMs?
Somebody on the founding team has "set up DM automation" on a list. It has been on the list since March. Nobody has done it, and every so often it gets moved to the next sprint with a slightly guilty note attached.
Here is the version of this post that would sell more software: your inbox is leaking leads, automate it today. Here is the honest one. At six people, pre-product-market-fit, this is usually not worth a week. It is sometimes worth an afternoon. What decides which is a single question about your own answers, and most teams have never actually asked it.
The condition: are you typing the same sentence twice?
Not similar sentences. The same one.
Open your DMs and read the last thirty. If most of them are asking things whose answer has been identical for two months — is there a free tier, do you have an Android app, is this only for India, where do I sign up — then you are hand-typing a FAQ and you should stop, and it will take you an afternoon.
If instead most of them are different questions that need a real answer, or the same question whose answer keeps changing because you keep changing the product, then automating is not a saving. It is a maintenance job you are volunteering for at the worst possible moment.
Templates are cheap to write and expensive to keep true. At the stage where the answer changes monthly, that trade goes the wrong way.
The DMs are the research, and automating them destroys the sample
This is the argument specific to early-stage and the reason to be careful rather than fast.
At six people the founder is answering DMs personally, and that is not overhead. It is the cheapest user research available. The way somebody phrases their question tells you what they think you are. The follow-up they ask tells you what your landing page failed to say. The thing they assumed you did tells you what to build or what to stop implying.
Automate that away and you get a clean inbox and a slightly stupider company. So the split is not "automate the inbox". It is: automate the logistics, keep the interesting ones, and know which is which.
Automate — it is logistics
Answer yourself — it is signal
The one automation worth building at seed stage
If you build exactly one thing, build the launch-post comment automation, and build it before the launch post rather than during.
You post the announcement. The caption says comment EARLY and I will send you the access link. People comment, the automation sends the link in a DM, the conversation is captured. That is a real use of the mechanic, because a launch post is the one time an early-stage account gets more inbound than a founder can type through, and because the answer is genuinely identical for everyone.
Two details that matter more than the template:
- Pick a word that is not already in your comments.
LINKandINFOappear in half of any comment section. A word specific to the launch fires only when somebody means it. - Say the word in the caption. No line, no comments, no flow. The automation is downstream of the ask, and the ask is your job.
- Set a rate budget before you post, not during. A launch that travels is exactly the scenario where an uncapped automation sends a large pile of near-identical DMs in an hour and starts to look like spam to Meta.
- Put a retirement date in the calendar the same day. Early access closes. The link changes. The reel stays on the grid forever.
What it actually costs, since you are counting
Templated replies are free and unlimited, so a fixed launch reply costs nothing whatever the volume. A credit is spent only when the AI writes a new reply, one credit per reply. The free tier is 100 credits with no card, and packs start at ₹499.
The real cost is not the credits. It is the half-day of writing replies, the half-hour a month of reading them back against a product that has moved, and the discipline to switch things off. Budget the last one honestly, because it is the one nobody does.
What it does not do, and what that rules out
It does not schedule or publish anything. No content calendar, no captions, no hashtag tool. Growth is not a feature here.
It cannot message people who have not messaged you first, which rules out DMing your followers about a release entirely. There is no cold outreach, no broadcast, no nurture sequence, no analytics product, no CRM sync — captured conversations leave as a CSV. No WhatsApp.
Official Instagram Graph API only: seven days to reply to a comment, twenty-four hours inside a DM thread, restarted only when the person messages again. That last one kills the drip campaign somebody on your team is imagining right now.
The honest recommendation
Do not make this a project. Spend one afternoon: read thirty DMs, write two templates for the two questions you type most, build one automation on one post, test it on yourself, go live, and put a reminder a month out to read them again.
If in a month the templates are still true and you are still typing the same sentences by hand, do the second one. If they have already gone stale, you have learned something more useful than a full setup would have taught you, and it cost you an afternoon. What the free tier actually gets you covers the rest of the maths.
