Koolaidrain

Silk

Oct 9th, 2026
19,767
0
Never
2
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 3.74 KB | None | 0 0
  1. •⁠ ⁠intros
  2. •⁠ ⁠⁠sameen pitch
  3. •⁠ ⁠⁠how do you use AI? daily wf
  4. •⁠ ⁠⁠what do you want from AI?
  5. •⁠ ⁠⁠this is what weaver does
  6. •⁠ ⁠⁠quick product vision. no one else has this vision
  7. •⁠ ⁠⁠what we’ve been working on. burning bugs, rebrand, plugins, cool features lined up
  8. •⁠ ⁠⁠ritesh ad-libs
  9.  
  10. - core ai platform
  11. - tom: represent all workflows that people and humans do
  12. - gav: what tom just said. too many years in the industry
  13. - how to replace the PDLC, what does that look like?
  14.  
  15. onboard/ingest?
  16. vms?
  17. rbac?
  18.  
  19. I was hoping we could spend a few minutes introducing
  20. hey I'm sakib, this is ritesh, ritesh do you wanna introduce yourselves
  21. my background is in large scale systems at fb, I was a PE working on traffic infrastructure
  22. that's when I was in San Francisco - moved onto MongoDB at some point and worked on building and scaling an enterprise kube platform that supported the entire atlas product
  23. we both use a lot of agents all the time. we both have an issue with multiplayer AI
  24. we know it's a bit of a crowded space, but we're both a little frustrated that none of the products out there have the complete user experience that we were looking for
  25. Ritesh and I met through some projects we were working on
  26. since then we've been dev'ing weaver (formerly pix) for about 6 months
  27. we've been spinning wheels on the rebrand recently
  28. then we realized YC actually put out a request for startups to solve multiplayer AI!
  29. so that's how we got into this.
  30. I've personally seen such a massive amount of variance in how my peers make use of AI
  31. y'know, some people still don't use AI at all, most people use AI to varying effect.
  32. and very few people make effective use of AI. very few have established the rigorous setup needed to 10x themselves.
  33. we don't think it has to be like that. we're pretty focused on the details
  34. we want to build the BEST ai workflows possible and make those available to everyone, community and enterprise
  35.  
  36. tom: local claude, design something, dump artifact into platform, platform gets the job done
  37. what I picture is mirro or stitch where I can work with a bunch of people on my team to collaboratively build the artifact that can then be fed into the development platform
  38. gav: 200 jobs to be done by the business - "I want a calendar that can make things better"
  39. we want a platform to shape the intake of intent into the PDLC
  40. so imagine a hub that leads you, your fellow engineer, and your fellow PM into a shared view of 4-5 agents side by side. these agents can talk bidirectionally to each other. everyone can collaborate with RBAC, the PM can maybe propose changes to prompts to different agents, an engineer is the driver for one or more of the sessions, the sessions carefully manage context. shared artifacts
  41. tom: the piece missing from this is the artifact management. I wanna be able to use my local claude. it's very text based.
  42. tom: if we've knocked out 20 odd tickets. new joiners can pull on the context of the conversation so far
  43. dont care what everyone is using but i wanna make sure
  44. tom: when you're working it comes down to, I've got my local thing, that's also this remote deployed thing too, but with local coding harnesses. I need my own local thing, I like my agent hearing it in math notation, no one else does
  45. gav: I've spent my career translating excel to tech.
  46. sameen mentioned
  47. gav: models will get cheaper and cheaper. I've got
  48.  
  49.  
  50. recently we put up new branding and a new landing page - you can see that at https://silkstack.ai/
  51. here's what we have lined up: we'd love to get your thoughts
  52. (share major items on the product roadmap items -> talk about high-level specifics)
  53. - multiplayer
  54. - p2p
  55. - PM and engineer work on the same IDE together (agentic XP)
  56.  
Advertisement