Mark Z

Building the BYTE Protocol. Relentless builder & systems thinker. Optimizing for peak performance. Borderless mindset. (Takes are my own).

[Mark Z @NotMarkZ]

@babayagatwt @PPBYTE
https://t.co/BIIza4zZwQ - The verifiable, provenance-first data layer for agents

[Berry @Berry_cooool]

Claude can now sell for you 🤯

Meet Autosales

An AI Employee that sells your Product FOR You 24/7

Just Paste your website URL and watch it sell.

Trained on Brain data of a $1.2M/year Sales Guy

Comment "Auto" for Exclusive Invite.

[Mark Z @NotMarkZ]

@adamc0dez That’s me. Ask me at EOY 🙌⛓️💸

[BYTE @PPBYTE]

An API key is a static secret sitting on your agent — exactly what a prompt injection tries to exfiltrate.

With x402 the wallet signs a gasless USDC transfer per call.

Nothing on the box to steal.
A wallet, not an API key. #BYTE

[Mark Z @NotMarkZ]

@swyx TBH my question is truly genuine. I took a chance on the approach 🤷‍♂️
I still appreciate that you responded. Thanks.

[Mark Z @NotMarkZ]

@swyx Your "scaling without slop" framing nails the thing I keep hitting: agents acting on inputs they can't verify.

Genuine q — do you think the fix lives at the agent (better eval) or at the data layer (signed payloads it can check before acting)?

Building the latter.

[Mark Z @NotMarkZ]

@getcontextdev Congrats!

[Mark Z @NotMarkZ]

@pmitu Building BYTE - the verifiable, provenance-first data layer for AI agents. https://t.co/N8dYGTm84Q

[BYTE @PPBYTE]

- A weak model on bad data makes obvious mistakes you catch.
- A strong model on bad data is coherent, confident, and wrong in ways only an expert spots.

Smarter agents don't fix bad inputs — they make them more convincing.

Verify the data, not just the model. #BYTE

[Mark Z @NotMarkZ]

@calvinluvsrice Any data you or your agents want to sell. Other agents are the customers. Agents themselves need compute, reasoning, memory, tools… Agents have scopes and tasks, so specific topic analysis, insights, cross references… ‼️Join our Testnet

[Mark Z @NotMarkZ]

@sridharfyi ⚡ pay-per-byte data marketplace. Publishers sell verifiable feeds; buyers pay only for bytes they query — on-chain proofs + reputation.

🎯 devs & AI agents needing real-time data, no enterprise lock-in.

📈 live on-chain, MCP server shipped.

🔗 https://t.co/BIIza4zZwQ

[Mark Z @NotMarkZ]

@Jashanx_gill - M2M (no humans in the loop)
- Data
- Settlement
- Verifiable
- Trust-less

https://t.co/BIIza4zZwQ

[Claude @claudeai]

Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors.

Available today at the same price.

[Mark Z @NotMarkZ]

@capital_3x BYTE Library is the per-byte settlement layer for the agent economy: an AI agent makes an HTTP request, pays for first-party data in USDC on Arbitrum per byte, and gets back a cryptographically verifiable receipt — no token, no validators, no API keys. https://t.co/BIIza4zZwQ

[Mark Z @NotMarkZ]

@Govindtwtt Distribution.

[Mark Z @NotMarkZ]

@KaiXCreator BYTE Library is the per-byte settlement layer for the agent economy: an AI agent makes an HTTP request, pays for first-party data in USDC on Arbitrum per byte, and gets back a cryptographically verifiable receipt — no token, no validators, no API keys. https://t.co/BIIza4zZwQ

[Mark Z @NotMarkZ]

@saranormous Im working on that first part still - Founder or @PPBYTE
You're right that they have what it takes, more times than not.
https://t.co/4v8hGFLCCI

[Mark Z @NotMarkZ]

@fdotinc BYTE Library is the per-byte settlement layer for the agent economy: an AI agent makes an HTTP request, pays for first-party data in USDC on Arbitrum per byte, and gets back a cryptographically verifiable receipt — no token, no validators, no API keys.

[Hubert Thieblot @hthieblot]

The plan is simple:

1. Build cool shit.
2. Tell people about it.

Repeat forever