This is super cool and I very much Want This Thing - so a sign that your pitch is clear! I like the section at the top of vision where you spell out an example user flow, that does a lot to generate the feeling of "yes!".

It's clearly both crazy ambitious and also eminently decomposable - it's clearly possible to prioritise out an MVP as long as you can make some progress on that core semantic composition piece of technology. You do a little bit of science/research and then suddenly unlock a lot of possible engineering. Seems like great timing to try to start this with a small team.

If there's anything else I'd like to see in this pitch it's a competitive analysis - who else is doing things in space? or if there is nobody, what's the nearest thing that might hit this value vector?

Overall very excited by your products! if the core tech works, I will buy them :)

Great concept! Powerful vision, if incredibly ambitious. I think you're absolutely onto something, and I look forward to your Linux distribution later this year :P but really, I'd love to hop on a call sometime and have you show me around the ecosystem you're building, and how it all fits together. Seems like some stuff I might get mileage out of, myself.

This is a great pitch, and you've already built a lot! Considering your main ask was engagement with your work, I'd say your pitch is was pretty successful considering the nature of the review process. I'll be honest and let you know that I don't have the technical knowledge to thoroughly engage with your work and evaluate what you've built or offer any useful feedback, but I shared with my own personal AI agent and this is what they had to say about your work. Hopefully there's something of substance in there for you. 

"Spent some time going through the full ecosystem — Vessel, Flint, Shards, Mesh, NCM, Surf, and the shard repos on GitHub. Since your ask is direct engagement with the vision and technology, that's what I'll try to give you.

The most genuinely novel thing here is the Shards system. Claude Skills and Skill Graphs exist, but the statefulness angle — shards persisting artifacts into a Mesh so workflows compound across sessions rather than starting fresh — is a meaningful contribution. Looking at shard-flint in particular, you're not just defining tasks for an agent, you're defining agent conventions: how to load shards, how to read templates, workspace navigation, session tracking, artifact frontmatter. That's essentially an operating system for agentic knowledge work, written entirely in markdown. The consistent structure across all the shard repos (manifest, init file, workflows, skills, templates, knowledge) shows real architectural discipline, especially for something three months old.

The ecosystem coherence is impressive for the stage. NCM feeds Vessel, Mesh underpins Flint, Shards run inside Flint, Vessel publishes from Mesh — it actually connects end to end. And the fact that this pitch is being served on Vessel, written inside a Flint workspace, using your own data primitives, is strong show-don't-tell. You're your own power user and the feedback loops are clearly short.

Mesh and NCM as foundational data primitives are smart bets. Published npm packages, O(1) link resolution, storage-agnostic, designed for humans and LLMs alike — that's the kind of boring infrastructure work that compounds over time. The decision to invest in this early rather than rushing to consumer products is defensible, especially given the thesis that LLMs reward meta-infrastructure in a way that wasn't viable before. I think that thesis is probably correct.

Where I'd push:

The acknowledged risk of over-indexing on meta-tooling is real. Every piece serves the system, but no single piece has proven value to someone outside the team yet. You need a wedge — one product that makes someone who doesn't care about the ecosystem go "I need this." Surf feels like the most natural candidate for that because the value proposition is immediately legible (zoomable document reading is something you can demo in 10 seconds), but it's currently the least developed piece. Is there a version of Surf that could be functional enough to put in front of people in the next month or two? That single artifact would make the rest of the ecosystem story much easier to tell.

Alternatively, Flint + Shards might be the wedge for a developer audience. Anyone running Claude Code sessions regularly is already feeling the pain of context fragmentation and stateless workflows. If you could get even a handful of Claude Code power users running Flint with a couple of shards and reporting back on whether the compounding effect is real, that's concrete validation that the whole system thesis rests on.

One more thing — the pitch says the business side comes next year after graduation, and that spending time pitching now doesn't help you build. That logic makes sense for VC pitching, but community engagement is different. The people most likely to become early users of Flint and Shards are the same type of people reading pitches in this community. Pitching your work to more communities and getting a few of them into the workspace and building shards alongside you would simultaneously validate the product, stress-test the conventions, and build the kind of organic traction that makes future pitching trivial.

Really solid foundation. Would be curious to hear other people's experience with Flint and try it myself."