Live data from Hacker News

Launch HN: Almanac (YC S26) – AI that knows your company

usealmanac.com

51–56 of 56 posts

Re: Launch HN: Almanac (YC S26) – AI that knows your company

#52
Congrats on the launch, the approach of spending compute upfront is a compelling shift from what the current standard is.

I am curious as to how you handle state invalidation and unseen side-channels during those long-running tasks, because if the agent pauses a task for X days, say waiting for an email reply, but the projects params changed offline or in a channel that the agent has no eyes on, how does the system realize that the pre-compiled context has now gone stale? Does that also mean that proactive actions always require a HITL?

Re: Launch HN: Almanac (YC S26) – AI that knows your company

#53

> that knows everything about your company Given that I don't know everything about my 4-person company, and that I never knew everything when it was a 1-person company (guess which person is that), this sound like https://x.com/mitchellh/status/2044529496217555333 (apologies for twitter link)

[dead]

Re: Launch HN: Almanac (YC S26) – AI that knows your company

#54
The hard part for an always-on company agent is what survives a restart. My runtime snapshots the whole JS heap to bytes and restores it in a fresh process, so conversation and working state come back with no serialization code — but timers don't survive, so an agent re-arms them from declarative state after restore. How do you handle that? Is a long-lived agent's state checkpointed, or rebuilt by replaying context on each wake?

Re: Launch HN: Almanac (YC S26) – AI that knows your company

#55
post #52

Congrats on the launch, the approach of spending compute upfront is a compelling shift from what the current standard is. I am curious as to how you handle state invalidation and unseen side-channels during those long-running tasks, because if the agent pauses a task for X days, say waiting for an email reply, but the projects params changed offline or in a channel that the agent has no eyes on, how does the system r…

This is an interesting edge case. When something similar happened, Almanac texted me asking for an update if I had followed up on email, and then I mentioned that I had followed up. So in this sense there was a HITL. A good solution could be to have account provenance for tasks, so that disconnection on accounts could update stale context

Re: Launch HN: Almanac (YC S26) – AI that knows your company

#56

Earlier quoted context omitted.

[flagged]

What does this comment add to the conversation? Is it just flame bait? Seems you have a habit of making this pointless comment. https://news.ycombinator.com/item?id=49189670

But does it cause flaming?

Twitter uses (used) a different mechanism for comment moderation and posting so it’s to reduce confusion

Post reply on HN