Three ideas I said no to, and the one I picked
Before building Autopilot I looked at a link-in-bio tool, an AI visibility tracker and SMS review requests. Why each one lost, and what made the winner different.

I look for opportunities worth building. Most of the work is saying no — so this series is a log of ideas I looked at, and why they did or did not make it.
No: a link-in-bio page
Every account that posts on Instagram needs a link page, and Clienzo already knows a user’s accounts and posts. It looked like an easy add-on.
The problem is that link-in-bio tools are a commodity. Free versions are good enough for most people, and the ones that pay are hard to win from established tools. Needed, but nobody pays — that is a feature, not a business.
No: an AI visibility tracker
The idea: show a business how often AI assistants mention it when people ask for recommendations. The interest is real.
I could not make it reliable. Assistant answers change with wording, location and time, and a number that moves for reasons nobody can explain is worse than no number — the same rule as Clienzo’s analytics.
No: SMS messages
Sending messages by SMS reaches people better than email. It also costs money per message, and at the prices small businesses pay for a tool like Clienzo, the margin does not survive. I turned it down on cost, and I do not reopen it without new numbers.
Yes: drafting the posts
What made the winner different was evidence of pain, not elegance. The most common problem small businesses report with social media is producing enough content. And there was already proven spending on automation around social replies.
So the idea became Autopilot (part 13 of Building Clienzo): AI drafts next week’s posts from real sources, code checks the facts, the owner approves. It fits what Clienzo already has — accounts, results, comments — instead of being a separate product to sell.
The filter I use
- Is the pain real, and do people already pay to solve it?
- Can I make it reliable enough that I would trust the output myself?
- Do the unit costs work at the price the customer will pay?
- Does it build on something I already have?
Next in the idea log: the social networks I decided not to support.
related
Chat apps are channels too
Why Clienzo posts to Discord, Telegram, Slack and Viber next to the usual social networks — and why Microsoft Teams did not make the list.
Why I'm building a social media scheduler in a crowded market
There are plenty of schedulers already. Here is the narrow bet that made me build another one, and who it is for.