Tell HN: ChatGPT needs a persistent workspace layer
7 points
13 hours ago
| 1 comment
| HN
Many heavy ChatGPT users have shifted from “conversation” into “continuous work”: planning, research, writing, coordination, and decision-making.

But conversations evaporate. Tabs don’t scale. Threads don’t map to how humans manage multi-week commitments.

What’s missing is a persistent workspace layer:

— account-level, not chat-level

— structured but lightweight

— model-aware (so the AI can reference context)

— opt-in (never intrusive)

— supports memory, plans, tasks, notes, drafts, decisions, and coordination

This would transform ChatGPT from an assistant for “this moment” into a partner across time.

Curious if others feel this gap emerging — especially people using LLMs for long-term personal or professional workflows.

ofabioroma
11 hours ago
[-]
This resonates hard. I'm a heavy branching user too and kept losing track of context across sessions and threads.

That's actually why I built UltraContext (Show HN is live today). It's infrastructure for this exact problem: persistent context with version history, so you can fork, rewind, and track how conversations evolve over time.

Not a ChatGPT plugin (yet), but the API could power something like what you're describing. Would love to chat if you're interested in exploring this.

reply