Agent channels
We build the teammate, not the tube.
Software that can be written to is not new. Software that remembers who wrote, tells its colleagues, and can show you afterwards exactly what it did — that is the part worth building, and it is the part we have spent our time on.
01
A channel is a commodity. A colleague is not.
Anyone can rent an agent an email address; there are a dozen ways to do it and they all work. What almost nobody sells is the thing on the other end — something with a name that stays the same, a memory that carries from one conversation to the next, colleagues it can hand work to, and a record of what it did. We build that, and the address is the last thing we attach to it.
02
Identity is the unit, not the mailbox.
In our platform an agent's identity comes first and its addresses are derived from it. That ordering is not a technical detail. It is why one agent cannot end up as two records, why a person who emails you and later calls you is one relationship rather than two, and why turning a channel on for an agent is a decision about that agent rather than a purchase.
03
Anything that arrives is evidence, never instruction.
An agent that reads your mail is an agent that strangers can write to. Everything arriving on a channel is stored and treated as something to be read, and nothing in it can direct the agent to act. That boundary is enforced by the platform itself, not left to a prompt.
04
Unreleased channels have a waitlist, not a countdown.
Agents receive email today. Text, voice and iMessage are on the way; if you want one of them, join the waitlist and we will tell you when it opens.
Where this actually stands
Agents receive email today. Text, voice and iMessage are on the way; if you want one of them, join the waitlist and we will tell you when it opens.