Drop-in remote worker
An agent you onboard like a new colleague, which then does a remote job end to end.
stageSpeculative Open on the map →
Facts
- Lane
- Agents and organisations
- Wave
- 9 of 12 (build order, not a date)
- Needs
- 24 boxes all the way back through hard links
- Unlocks
- 1 boxes downstream through hard links
- Leverage
- 0.1 points · stands in the way of 1 of 53 stages · rank 53 of 57
- Cluster
- Agents at work
- When
- Researcher estimate: 2028 to 2034
Evidence as of October 2026
Aschenbrenner’s drop-in remote worker is onboarded “like a new human coworker”. Epoch AI’s Earthborne Rangers benchmark found frontier models showed no improvement across 30 playthroughs, so they still fail at learning on the job.
Watch
Whether models improve at a new job over repeated attempts. Epoch found no improvement across 30 playthroughs.
Timeline note
My estimate. Aschenbrenner argued in 2024 that drop-in remote workers were plausible by 2027; in 2026 models still do not learn on the job.
Around it
Loops and off-ramps
Openings researchers' judgement, October 2026 scouting
Startup room 31/100 · rank 110 of 124 Research room 45/100 · rank 76 of 124
- Depth
- light
- Players
- Frontier labs; Mercor ($20bn talks); Remote Labor Index (15.8% best)
- Wedge
- Long-horizon reliability and absorbing firm context; lab territory. none