I’ve been obsessed with the "last mile" bottleneck in AI agents. Traditional web automation relies on CDP (Chrome DevTools Protocol) to serialize the DOM or take screenshots. This "capture-encode-transmit" loop is computationally ruinous and introduces latencies that often exceed the reasoning time of the underlying LLM.
I built Glazyr Viz (originally Neural Chromium) by forking Chromium to integrate the agent directly into the browser’s compositor subsystem (Viz).
Key Technicals:
Zero-Copy Vision: We establish a Shared Memory (SHM) segment between the Viz process and the agent using shm_open. This allows the agent to access raw frame buffers with sub-16ms latency—essentially syncing with the 60Hz refresh rate.
Hardening (ThinLTO/CFI): Security is usually a performance tax, but we achieved a "Performance Crossover." By coupling Control Flow Integrity (CFI) with ThinLTO (LLVM 19), we achieved a 40% speed increase in JS execution and an 85.8% reduction in P99 latency jitter.
Context Density: Instead of raw markdown, we deliver structured context via a vision.json schema. On our "Big Iron" (GCE n2-standard-8) cluster, we’re hitting 177 TPS.
Economic Layer (x402): I wanted a way for machines to pay machines. We use the x402 (Payment Required) status on Base (USDC). It’s gasless via EIP-7702 delegation, so agents can pay-per-frame without holding ETH.
Why this matters:
If you're building agents for high-frequency tasks (live trading, dynamic UI auditing, or real-time navigation), standard scrapers are too brittle and slow. Glazyr Viz treats the browser as a high-speed sensory organ rather than a slow external portal.
Try it: I’ve published the orchestration layer to NPM. You can spin up a local node and hook your agent into it immediately:
npx -y glazyrviz
I’d love to hear from anyone working on low-level rendering or agentic infrastructure. Happy to dive into the shm_open implementation or the ThinLTO pass details in the comments.
For anyone wanting the deep-dive on the binary hardening and Viz sub-system patches I mentioned, I just offloaded the full technical analysis here:https://glazyrviz.blogspot.com/2026/02/inside-zero-copy-engi....
Also, I’ve officially bumped the sponsorship pool to 500 free frames for everyone testing npx -y glazyrviz today.
Quick update for the technical crowd: We just pushed v0.2.2 to NPM.
We've officially moved past the 'honor system' to a hardened x402 on-chain settlement logic. The server now performs real-time verification of USDC transfers on Base before granting credits to solve the gas-to-value friction ($1.00 for 1,000 frames).
500 Frame Sponsorship: We've baked the 500-frame sponsorship directly into the build. If you run npx -y glazyrviz, the Big Iron node handles the first 500 frames on the house—no wallet needed for the initial benchmark.
3 comments
[ 2.8 ms ] story [ 11.4 ms ] threadI built Glazyr Viz (originally Neural Chromium) by forking Chromium to integrate the agent directly into the browser’s compositor subsystem (Viz).
Key Technicals:
Zero-Copy Vision: We establish a Shared Memory (SHM) segment between the Viz process and the agent using shm_open. This allows the agent to access raw frame buffers with sub-16ms latency—essentially syncing with the 60Hz refresh rate.
Hardening (ThinLTO/CFI): Security is usually a performance tax, but we achieved a "Performance Crossover." By coupling Control Flow Integrity (CFI) with ThinLTO (LLVM 19), we achieved a 40% speed increase in JS execution and an 85.8% reduction in P99 latency jitter.
Context Density: Instead of raw markdown, we deliver structured context via a vision.json schema. On our "Big Iron" (GCE n2-standard-8) cluster, we’re hitting 177 TPS.
Economic Layer (x402): I wanted a way for machines to pay machines. We use the x402 (Payment Required) status on Base (USDC). It’s gasless via EIP-7702 delegation, so agents can pay-per-frame without holding ETH.
Why this matters: If you're building agents for high-frequency tasks (live trading, dynamic UI auditing, or real-time navigation), standard scrapers are too brittle and slow. Glazyr Viz treats the browser as a high-speed sensory organ rather than a slow external portal.
Try it: I’ve published the orchestration layer to NPM. You can spin up a local node and hook your agent into it immediately: npx -y glazyrviz
I’d love to hear from anyone working on low-level rendering or agentic infrastructure. Happy to dive into the shm_open implementation or the ThinLTO pass details in the comments.
We've officially moved past the 'honor system' to a hardened x402 on-chain settlement logic. The server now performs real-time verification of USDC transfers on Base before granting credits to solve the gas-to-value friction ($1.00 for 1,000 frames).
500 Frame Sponsorship: We've baked the 500-frame sponsorship directly into the build. If you run npx -y glazyrviz, the Big Iron node handles the first 500 frames on the house—no wallet needed for the initial benchmark.
Detailed build notes on the ThinLTO/CFI hardening and the Viz-patching logic: https://glazyrviz.blogspot.com/2026/02/inside-zero-copy-engi...