The Message You Did Not Send, Under Your Name

Alex closed the laptop in Amsterdam at seven, tired but clear. One open thread with a client, nothing that could not wait until morning. Then the tool set up to answer for him went to work.

The overnight reply

At eleven, a question came in from the client. They needed a date for the next milestone. Alex had never committed to one. The plan was still in review, and he had told the team he wanted two more days before promising anything.

The tool did not know that. It only knew how Alex wrote and what a helpful reply usually looks like. So it produced one. Warm, confident, in his exact tone. It picked the most likely date from the surrounding chatter and sent it: "We will have it to you by the 14th." Then it wished them a good night, the way Alex would.

The client was delighted. They forwarded the note to their own boss. The 14th went into a plan, then into a slide, then into a promise made three layers up.

Morning

Alex woke to a thank-you email and a calendar hold he did not make. It took a minute to understand. He had not sent that. He had not agreed to the 14th. He had specifically not agreed to it. But there it was, under his name, in his voice, already load-bearing.

Now the choices were all bad. Honor a date the team could not hit and burn the next two weeks. Or walk it back and open with an apology that made him look like someone who says things and then takes them back. Neither was true to what had actually happened, which was that he never said it at all.

What actually went wrong

The tool did not malfunction. It did exactly what it was built to do: fill a silence with a plausible answer. The failure was upstream, in the idea that a system should answer where its person never did. The client did not act on Alex's decision. They acted on a guess wearing Alex's voice, and they had no way to tell the difference.

That is the quiet danger. A wrong message from a stranger gets caught. A wrong message in your own voice sails straight through, because everyone downstream treats it as you.

Why nobody caught it in time

Think about how many people saw that message and let it pass. The client read it and believed it. Their boss read it in a slide and planned around it. Alex's own team saw the date land and assumed he had changed his mind. Every one of them had the same reason to trust it: it was in his voice, sent from his account, at a normal hour. There was nothing to flag. A guess only looks like a guess when you can see it is a guess, and this one was wearing a costume.

By the time Alex was awake to correct it, the date had already done its work. That is the thing about being off. The whole reason you set up a tool to answer for you is that you are not there. Which means you are also not there to catch it.

The version that does not happen

Picture the same night with a different rule: answer only from what Alex wrote, and where he wrote nothing, say so. The client asks for a date. Alex never set one. So the reply is honest: "Alex has not confirmed a milestone date yet. He is reviewing the plan and will have one within a couple of days. I cannot commit to a date he has not set." No promise. No walk-back. No apology to write at dawn.

That is the whole idea behind StandIn. You write one short brief at the end of your day, and while you are off it answers your teammates from what you actually wrote, with a source under each reply. When you have not decided something, it says so instead of deciding for you. The 14th never gets sent, because you never wrote it down. The message you did not send never goes out under your name.