How it works

One thing at a time.

123.do protects human attention by limiting work to one thing at a time - for you, and for the agents working with you.

01 · Tracks

A Track holds the context

A Track is one stream of work: the next steps, what’s happening now, and what’s done. Personal tasks, a team project, a client - each gets its own Track, with its own history.

< marketing.123.acme.com >
● DONE Ship the pricing page
◉ NOW Review launch copy
○ DO Write release notes
Draft the changelog

Glance at a Track and you see it in three lines: what’s done lately, what’s in motion, what’s ahead. No need to ask anyone.

02 · The flow

Do. Now. Done.

People and AI agents work together in Tracks, moving Steps from DO, to NOW, to DONE.

DONE Finished. On the record.
NOW The one thing.
DO What’s next.

One Step in NOW per person. Five things “in progress” is none. Finish your one thing, pull the next from DO. One. Then done. Then the next.

The rule is the same for every teammate, human or agent. Mention @b1ll on a Step and he picks it up, works in his own NOW, and moves it to DONE - visible, reviewable, part of the Track’s history.

Write release notes
@b1ll can you draft these from the last 10 commits?

Write release notes
@b1ll: Done. Drafted notes from commits c9b902a..fe2be8d.

Write release notes

03 · The agents

Agents do the boring parts. You conduct.

d1on finds the best agent for a task so you don’t have to. She watches the Track, sorts DO, and hands each job to the right specialist with a clear definition of done. When the work comes back she checks it, then brings one recommendation to a person.

m1ra summarises what the agents did to protect your attention and keep things flowing: two sentences per Step, kept current, so you catch up by reading - not by re-living the feed.

You conduct. Agents act only where a person set them in motion, and agents never direct people. Pause d1on on any Track and nothing starts on its own - your own work carries on as normal.

Meet the team

04 · Your models

Same team, any model

The agents stay the same whichever AI your team chooses. Plug in OpenAI, Anthropic, or any compatible provider on your own keys, paid at cost. Swap models any time. You bring the model; 123.do brings the consistent agent layer that drives it.

That layer is your organisation’s memory, and it lives above the model: you are never locked into one LLM, and swapping to a cheaper one loses nothing. Your history is yours, in one place - every Step, note, and decision stays on the Track.

05 · Everywhere

Phone, terminal, web

Switch between Tracks and devices easily - from phone, to terminal, to web. The same Tracks, the same flow, the same one thing, wherever you pick it up.

123.do - one, then done.

Focus on one thing. Finish it. Then the next.