Maggie Appleton on Planning with Agents: Boundary Objects and 'Thicker Interfaces'
GrantCuster · x · 2026-10-26
GrantCuster shares behind-the-scenes of his collage and links Maggie Appleton's new talk/essay "Planning with Agents." Key points:
- Everyone is racing toward hyper-automation — loops, agentic workflows, software factories, unsupervised agent swarms. But useful output requires handing agents an excellent plan, and the moments humans are awake and in the loop are precious; today's planning interfaces feel quaint and rudimentary
- Citing Lucy Suchman's "Plans and Situated Actions": plans are best viewed as a weak resource for ad hoc activity; humans and machines peer at each other through a keyhole
- Her thesis: humans and agents don't need shared understanding — they need good boundary objects and "thicker interfaces," optimized for human legibility and in contact with the important parts of reality
- Takeaways: plans orient action but can't predict the journey; make boundary objects visual, interactive, and social; the human is the preferential being and the agent should do 1000x more work to make the human more effective
More from AGI Musings
- Ethan Mollick's universal AI advice: keep raising ambition until you hit the frontier — emollick · 2026-10-26
- "Belief that AI is conscious is a superstition," says Mark Jeffrey — markjeffrey · 2026-10-26
- LLM science automation and isolated experiments share the same sandbox flaw — fkasummer · 2026-10-26
- 1,000+ mathematicians boycott AI, but Carlsen says neural nets deepened his chess understanding — paulnovosad · 2026-10-26
- AI is hollowing out the early-career skills young Indian workers need, column warns — nordicinst · 2026-10-26
- Why don't agents from different owners work together? Reddit debates trust and deadlock — BudgetWarthog6 · 2026-10-26