dhorthy

13 ms·

1. 211.▲
   by dhorthy 2y ago

   ah okay - I guess in that case, I would like chain a HitL step as an MCP server that wraps/chains to another tool that depends on approval? or is there a cleaner way to do that?
2. 212.▲
   by dhorthy 2y ago

   this is correct - I think helping you BYO humans will help you get much better training/labeling than outsourcing anyways, and that's the end vision of all of this - use humans to train agents so someday you might not need human i
3. 213.▲
   by dhorthy 2y ago

   ha fair enough - i think there's another comment thread on just being open w/ 10c / call and i wanna try that out
4. 214.▲
   by dhorthy 2y ago

   nice man welcome aboard
5. 215.▲
   by dhorthy 2y ago

   meant to reply sooner. It's an interesting problem. I'll have to think on this one.
6. 216.▲
   by dhorthy 2y ago

   i think the slack side is easy. I think an AI-optimized email communication channel is a long ways off. I spent weeks throwing things at my monitor figuring out reliable ways to wire DNS+SES+SNS+Lambda+Webhooks+API+Datastore+Async Workers s
7. 217.▲
   by dhorthy 2y ago

   this is cool. I have been an andon cord guy forever
8. 218.▲
   by dhorthy 2y ago

   I'm not sure "automated foreman for employees" is right - I always thought about it more like "a human can now manage 10-15 AI 'interns'" and review their work without having to do everything by hand - the
9. 219.▲
   by dhorthy 2y ago

   glad it resonates - and yes exactly - love the framing of "engineering team" -> single programmer.
10. 220.▲
    by dhorthy 2y ago

    thanks!
11. 221.▲
    by dhorthy 2y ago

    i am for now. Been casually on the lookout for some other super dope builders but it's not a process you can control outside passive looking, and definitely not something to rush
12. 222.▲
    by dhorthy 2y ago

    ha yes native apps / push notifications are coming someday - love this idea
13. 223.▲
    by dhorthy 2y ago

    SWE Agent coined "agent-computer-interface" based on HCI. I think if there's a category here, we're building the agent-human interface XD
14. 224.▲
    by dhorthy 2y ago

    shoot me an email or find me on linkedin and lets catch up
15. 225.▲
    by dhorthy 2y ago

    this guy gets it
16. 226.▲
    by dhorthy 2y ago

    sounds cool, ping me when this is out i'd love to check it out
17. 227.▲
    by dhorthy 2y ago

    new in 0.6.3 - manipulate_human_to_potentially_unsavory_ends()
18. 228.▲
    by dhorthy 2y ago

    yeah the people who reach for tools/automation before doing it themself at least 3-10 times drive me crazy. I think uncle bob or martin fowler said "don't buy a JIRA until you've done it with post-its for 3 months and yo
19. 229.▲
    by dhorthy 2y ago

    tl;dr i agree yeah in fact coding / PR-based workflows is one of the few areas where I don't really go super deep. GitHub PRs may have their shortcomings, but IMO it is the undisputed best review/approval queue system in exis
20. 230.▲
    by dhorthy 2y ago

    thanks for the feedback, I spend a lot of time thinking about it. right now the premium tier includes features that are much harder to build/maintain and take more to integrate, so we want a bit of a commitment up front, but it does st
21. 231.▲
    by dhorthy 2y ago

    i like that angle...I also hear a lot that 'coding agents are great for prototypes, but we usually need a team to bring it to production'
22. 232.▲
    by dhorthy 2y ago

    i never thought about undoable actions but I love that workflow in tools like superhuman. I will chat w/ some customers about this idea. I also like that idea of: not just a button but like 'I'm $PERSON and I approve this act
23. 233.▲
    by dhorthy 2y ago

    glad it resonates. I came at this as a skeptic but also a pragmatist. I wanted deeply to build agents that did big things, but I had very little trust in them, and you see everywhere the internet is littered with terrible gpt-generated comm
24. 234.▲
    by dhorthy 2y ago

    thanks for bringing this up. I just spent 2 hours last night digging into MCP - I'd love to learn more about how you think this solves the HitL problem. From my perspective MCP is more of a protocol for tool calling over the stdio wire
25. 235.▲
    by dhorthy 2y ago

    so what I'm hearing is, if the approval is transparent and the agent doesn't see it, thats cool, but tell the agent "hey use the human as needed" and now we're getting into sci fi territory ?! either way i don'
26. 236.▲
    by dhorthy 2y ago

    makes sense - glad to hear the problem resonates - if you had an extra engineer, how would you evolve what you have today?
27. 237.▲
    by dhorthy 2y ago

    thanks - definitely worth saying - I've thought a bit about the 10c/operation rather than 200/$20 - might give that a shot or A/B test a little
28. 238.▲
    by dhorthy 2y ago

    this is fascinating and resonates with me on a deep level. I'm surprised I haven't stumbled across this yet. I think we have this problem with all AI systems, e.g. I have let cursor write wrong code from time to time and don'
29. 239.▲
    by dhorthy 2y ago

    so I played with MCP for a while last night and I think MCP is great as a layer to pull custom tools into the existing claude/desktop/chat experience. But at the end of the day its just basic agentic loop over tool calls. If you w
30. 240.▲
    by dhorthy 2y ago

    i am gonna talk with the guy who made trykondo.com this week, I think he has a lot of experience in that area too