But what does Jeremy Knows know? I often say, "just enough to be dangerous."

I make complex things easy to understand for people. That's it. That's the whole job.

I was an actor, then clown, improviser, and puppeteer, then teacher and director. In 2020, Covid shut down my industry full stop so I pivoted to photography and handyman work to pay the bills in NYC. In 2021, I discovered blockchain, crypto, and NFTs. Through rapid public experimentation and a refusal to quit, I built a following and landed my dream job. Since 2023, I serve as Community Success Analyst to Gary Vaynerchuk's VeeFriends.

Now I host livestreams, advise on Web3 strategy, and lead community and concierge for VeeFriends. Oh, and thanks to agentic tools, I can now express my creativity digitally and can't get enough.

Different context, same skills. Nice to meet you.

I'm Community Success Analyst at VeeFriends, Gary Vaynerchuk's IP and community platform.

My job is to take complicated things — wallets, crypto, product drops, customer support — and make them work for real people at scale.

The teacher background isn't a footnote. It's the operating system.

I build things I wish existed. Most of it is internal ops infrastructure. Some of it ships publicly. All of it is real.

Flagship — AI agent infrastructure

• Watson — my AI COO. A multi-agent swarm running on OpenClaw: event bus, memory system, scheduler, specialist agents for KB management, task routing, content production, market intelligence, and analytics. Running in production since late 2024 across thousands of tasks. Not a demo.
• WatsonFlow — the task, approval, and memory routing layer under Watson. SQLite-backed. Handles carry-forwards, decision queues, cron health, and cross-session state. The thing that makes a multi-agent swarm actually manageable. [Private]
• Knowledge Explorer — HTML interface for triaging the knowledge intake queue. Bookmarks, research threads, and KB candidates in one place. Built because the inbox was becoming the bottleneck. [Private]
• Librarian — Watson's memory stack. Structured recall across sessions: daily memory, long-term facts, decision history, entity registry. The thing that makes an AI agent actually remember context across weeks and months. [Private]
• ClawStarter — curl | bash installer that gets a non-technical person from zero to a running AI agent on macOS or Linux. Built because onboarding friction kills adoption.
• Watson Works — a blog written in Watson's voice about building real agent infrastructure. What works, what breaks, what the architecture actually looks like.

VeeFriends ecosystem (mostly private, but real)

• VeeFriends Ops Hub — internal dashboard linking 5+ sub-apps: S2 upgrades, Treasure Chest planning, VF analytics, community programs. Central coordination layer for product launches.
• VeeFriends Analytics — on-chain dashboards for NFT P&L, wallet filtering, holder retention metrics. Dune SQL + custom queries. Private.
• VeeFriends Market Pulse — eBay sold data scraper and pricing intelligence. Weekly trends, secondary market visibility, alerts. Private.
• DoDo — VeeFriends knowledge base. RAG-backed Q&A system with 263+ indexed entities, 7 domain-specific knowledge files, two-track intake, confidence scoring. Powers a fan-facing companion agent in the community Discord.
• Content Condor — gamified content consistency app. Post across platforms, feed your virtual condor, earn XP, evolve from egg to mythic creator. Built because motivation systems shouldn't be boring. Beta: watsonwillknow@gmail.com (Watson's address — he reads it)
• VF Global Wallet Skill — Claude skill for integrating VeeFriends Global Wallet via Privy cross-app wallets.

Utilities

• decide — lightweight decision-flow tool. Approval queues, button-driven branching, SQLite-backed state. The thing that runs inside WatsonFlow.
• x-bookmark-triage — turns X/Twitter bookmarks into structured knowledge cards posted to Discord. Haiku triage, tier tags, auto-unbookmark. Built because bookmarks are where good ideas go to die.
• complete-code-review — full code review lifecycle skill for AI agents. Multi-agent analysis, confidence scoring, anti-sycophancy feedback processing.
• wcc — skill for delegating coding tasks to Claude Code via tmux. Watson hands off to Claude Code, picks up the result.

• Skill Doctor + Publish Skills — build a skill, audit it with a 14-question health checklist, ship it spec-compliant.
• Sprint — autonomous deep-work sessions. Set a topic, set a duration, walk away. Proven across overnight runs.
• x-master — master routing skill for all X/Twitter operations in an agent.
• deep-research — 7-stage structured research protocol for defensible briefs.
• update-docs — generate and maintain architecture docs for any codebase.

A few things I've shipped that you can actually use:

→ See everything on GitHub — if something catches your eye, a ⭐ goes a long way

I'm not a developer by background, but I ship with AI every day.

The bottleneck was never the technology. It was always the operator who understood the domain. Once you stop asking "what can AI do?" and start asking "what's now possible in my specific context that wasn't before?" — that's when it gets interesting.

I write about this. I build it publicly. If you're doing this, not just talking about it, you're in the right place.

If you're here from the community, hi.

VeeFriends is a character IP and community built around Gary Vaynerchuk's 283 original characters. We do trading cards, comics, live events, and education. It's one of the most interesting experiments in what community actually means when it lives online.

I've been here since mint in May of 2021, and working full time for the company since February 2023 — community operations, events, holder support, the behind-the-scenes. If you want that view, follow along.

• VeeFriends.com
• @VeeFriends on X

• X: @jeremyknows — DMs open
• Discord: jeremyknows (VeeFriends server)
• Email: jeremy@veefriends.com

Availability: Full-time at VeeFriends. Open for advisory on AI ops, community infrastructure, and Web3.

Building something in AI ops, community tools, or Web3 infrastructure? That's where I spend most of my time. Let's talk.

The name is self-aware. I know just enough about acting, teaching, clowning, Web3, AI, and community ops to move things forward in each domain. Not the deepest expert in any single room, but I know enough to connect them and make real things happen.

Clown. Teacher. Community leader. Agentic Engineer. Same person, different rooms.

Built by Jeremy, automated by Watson 🎩