How messaging agents get their second user

The most consequential fact about a messaging agent is usually not what it can do. It is what a new person has to do before it does anything.

This is the field we record most carefully in this directory, because it is the one most obscured by marketing copy and the one that decides whether a product can spread through a group.

What 33 shipping agents actually require

Read it as a ladder. Four agents ask a new person to do nothing at all: they join a thread that person is already in. Ten ask for a single action. From there the cost climbs, and eight cannot be placed on the ladder at all because their public materials do not say what you do first, which is itself a finding for a category whose entire pitch is that there is nothing to install.

Why the group-thread pattern is different

Four agents in this directory work by being added to an existing group thread. That number understates their significance, because those four have a property none of the others do: the second, third and fourth people arrive with no onboarding whatsoever. They were already in the thread. There is no invite to send, nothing to install, no account to create, and no moment where someone has to be persuaded.

One product in this category prices accordingly, it charges per household rather than per seat and meters the agent's outbound messages instead. That follows directly from the onboarding model, because a per-seat price would tax the exact thing the product is good at.

Compare that with an agent requiring an app download. Every additional participant is a full acquisition: convince, install, sign up, return to the thread. In a group of four, that is three separate conversions before the product works as designed.

The cheapest onboarding in the directory

Two agents need a single tap on a link, no app, no signup, nothing typed. That is the lowest-friction arrival we have catalogued.

One of them is instructive for what people actually do with it. Its own pitch is broad companionship, but its most common real use is much narrower: someone screenshots a confusing message from a date or a friend and asks what to say back. Extremely low friction gets people through the door, and what they do once inside is often not what the pitch promised. That is worth knowing before you optimise onboarding at the expense of understanding the job.

The eight that do not say

Eight agents give no clear public answer to "what do I do first." Some require an account discovered only after clicking through. Some are invite-only without saying so. Some are simply unclear.

For a category whose entire premise is just text it, that opacity is self-defeating. If your product's advantage is that there is nothing to install, the sentence "text this number, nothing to install" belongs above the fold rather than in an FAQ.

What to take from this

• Count the actions before value, for the second user, not the first. Founders instinctively optimise their own onboarding. The person who decides whether your agent spreads is the one who did not choose it.
• If you want to live in group threads, being addable is the whole product. Everything else is downstream.
• State the first action publicly. A quarter of this market fails to, and it costs them exactly the users who would have found the low friction appealing.
• Match pricing to the onboarding model. Per-seat pricing on a product that spreads through existing threads charges people for the thing that makes it work.