How do AI agents keep working while teammates are offline?
By Soumen Sarkar · October 2026
Short answer: AI agents keep working while teammates are offline when the answers those teammates would give are already written down, and when the agent can state an assumption and carry on wherever they aren't. The teammate settles the assumption when they're back, and the correction reaches the agent at its next turn.
Why does an offline teammate stop an agent?
Because the agent reaches a point where it needs something only that teammate knows. The teammate is asleep, on a plane or in a different time zone, and the session sits there until they reply. Multiply that by every agent on the team and a lot of work spends the night waiting.
The fix isn't faster replies. It's making the reply unnecessary for the work to continue.
What lets an agent keep going?
- It looks first. Most of what an offline teammate would say is already written somewhere: a decision with its reason, a rule, a handoff from their last session. An agent that searches the team's record before anything else often finds the answer without anyone being awake.
- It states an assumption when the record is silent. It makes the most reasonable call, writes it as one line with its reason, and carries on.
- It marks the stakes. If the assumption touches data, money, access, a public interface or deleting something, it says so, and it chooses the option that is easiest to reverse.
- It leaves a handoff. When the session stops, what is done, what is left and what was assumed are written down for whoever comes next.
What does this look like across time zones?
A developer in Lisbon starts an agent at 7pm on the data pipeline, Ingest · pipelines/ingest. The task is to handle events that arrive late. The person who owns the pipeline is in Singapore, and it is 2am there.
The agent searches the team's record and finds one decision: events are deduplicated by event ID, because the mobile app retries sends. Useful, but it doesn't say what to do with events more than 48 hours old. So the agent states:
Assuming events older than 48 hours are moved to a quarantine table, not dropped,
because dropping data can't be undone. High stakes: data.
It keeps going, finishes the change, and stops for the night with a handoff that lists that assumption at the top.
What happens when the teammate comes back?
At 9am in Singapore, the pipeline owner sees one assumption waiting in their area. It is mostly right, but the business keeps late events for seven days, not two. They correct it in a sentence. The correction reaches the agent at its next turn, or the next session picks it up from the handoff, and the fix is a one-line change.
Nobody waited overnight. The quarantine choice meant nothing was lost while the call was open. This is what keep moving means across time zones.
How does Campfyrr route the assumption?
With Campfyrr, an agent states an assumption with one tool call. If the team's record already answers it, the answer comes straight back. Otherwise it goes to whoever made the closest decision or last worked in that area, and it shows in the project's Activity feed with an "Answer it" link to its own page, where it's answered in one click: right, keep going; one of the likely answers; correct it; or not mine to check. High-stakes assumptions are marked and shown first, and the agent keeps going either way.
Words used here
- Assumption: An assumption is the best available answer to an unknown, stated and acted on, kept visible so anyone can correct it.
- High-stakes assumption: A high-stakes assumption is an assumption that touches data, money, access, public interfaces or deleting things, so being wrong costs more and it should be checked first.
- Correction: A correction is the fix to a wrong assumption or call, made by someone who knows better and sent back to the work that relied on it, without stopping anyone.
- Handoff: A handoff is what unfinished work carries for whoever continues it: what's done, what's left, the next step, and what to watch out for.
The principle behind this: Keep moving: The work never waits, and neither do we.
Move as one at agent speed. Nobody waits. Add Campfyrr to Claude: it briefs every agent before it touches the code, keeps assumptions and handoffs where the team can see them, and never takes your code (what it keeps).