continuous workflows + browser history

Z / Runner 🏃🏻‍♂️

🏃🏻‍♂️ Assume we want a perfect blend of "infra <-> app"?

• manus = Manus replaces continuous human prompting with autonomous cloud execution.

• OpenClaw replaces the application interface with direct local automation.

• Are we trying to scale async reasoning + continuous multi-agent workflows or full-system level automation?

• 🏃🏻‍♂️ this is the correct read on the market condition, if the reward is long-horizon, then it is DOA for us. Hence, rewards must be immediate.

• 🏃🏻‍♂️ manus IMO had PMF b/c it's rev model was tied to generating cash via "token consumption"... it was extremely expensive to run and automate tasks via their services.

• 🏃🏻‍♂️ if we make openclaw as easy & safe to use as manus (that's a $1M arr in 6 months).

• 🏃🏻‍♂️ We should assume our user are power ai users; they're already paying $1k+/mo across AI tools — coding copilots, design tools, research assistants. They don't want another prompt surface. They don't want to manage OpenClaw locally. They don't care about tinkering with agents. Security, time, attention — all constraints… What they want is unfair advantage (e.g. cluley). We give customers a "24/7 compounding cognition".

• 🏃🏻‍♂️ i have to assume the bigger loot box is gmail and calendar… but it does require so much trust. Or even get access to maps history would be better than safari.

Stephen / Basketball 🏀

• 🏀 my (contrarian) bet: in 2026 consumers pay for transactional days-long tasks – rather than continuous in the background / cloud (as ghost).

🏀 *claw: paid tasks, information security, verified results.

• but manus's product design and revenue model?

• 🏀 how so? meaning, already 10x valuable if we optimize manus 10%?

• use safari browsing history instead via local file access

• 🏀 i thought about exactly this (existing footprints as intents) the whole day as you showed openai's pulse. nah – i write very little email and plan no calendar. maps will be mostly the same 10 places in a month.