An agent is just an assistant with a job
If your first assistant is the brilliant generalist, an agent is the specialist you bring on for one kind of work — an inbox-keeper who drafts your replies, a researcher who reads and summarises, an editor who sharpens your writing. Each one remembers its own brief and gets better at it over time.
You do not fill in a form to make one. You describe the role, the same way you would explain a new hire’s job, and 1Presence sets it up — then it appears in your list of agents, ready to talk to.
Creating one
Make me an inbox-keeper that drafts replies in my voice and never sends without me.
It builds an email specialist, wires it to your inbox, and leaves the send button firmly with you.
I’d like a research assistant that reads links I paste and gives me the gist.
It creates a researcher tuned for reading and summarising, ready for whatever you drop in.
Build a weekly-review agent that sums up what I worked on each Friday.
It builds the specialist, ready to run on its own — you choose exactly when by dropping it into a routine.
Shaping how it works
Give my editor a drier, more concise tone.
It adjusts that agent’s style; the change sticks for every future conversation with it.
My research assistant should always give me sources.
It updates the agent’s brief, and the habit holds from then on.
Rename this agent to Scout, and give it a compass icon.
It renames and restyles the agent so it is easy to spot in your list.
Putting it to work
Hand this thread to my inbox-keeper.
It passes the work to that specialist, which already knows your voice and your rules.
Ask my researcher to look into this and report back.
The agent does the reading and returns a summary you can act on.
Every agent draws on the same memory of you, so a new specialist is never starting from scratch — it already knows who you are.
Brief it like a colleague — messier is better
When you hand an agent a task, do not engineer the perfect prompt. Brief it the way you would brief a capable colleague — digressions and all. There is a dictate button in the composer; tap it and just say it out loud.
This is counterintuitive, so it is worth saying plainly: agents do their best work on loose, spoken-style input — not on a tightly compressed instruction. When you squeeze a request into one clean sentence, you throw away the context that says what you actually care about: the doubts, the asides, the “oh, and make sure you don’t…”. Leave it in. The agent reads the texture, asks a sharp question if it needs one, and fills the gaps from everything it already knows about you.
How creating an agent goes
- 1
Describe the job
Say what the agent is for and how you want it to behave — a sentence or two is plenty.
- 2
Answer a couple of questions
1Presence checks the details that matter — its name, its tone, what it can reach — right there in the chat.
- 3
Meet your new specialist
It builds the agent, which appears in your list ready to talk to and improve over time.
From a helper to a habit
Some agents you simply talk to. Others you want running on their own — the inbox-keeper that sweeps your mail before you wake, the reviewer that writes your Friday digest. The way there is to give the agent a job that repeats: a place in a routine or a workflow.
You create the agent here, by describing it. You decide when it runs later, by dropping it into a routine — that is where the day, the time and your timezone live, not on the agent itself. One specialist on a schedule is a routine; several working in order is a workflow.