mirror of
https://github.com/hanzo-apps/industries.git
synced 2026-11-02 21:21:46 +00:00
The favicon was hand-made. `6c17299` replaced the generated scaffold icon with a drawn "Hanzo H mark" whose blob is byte-identical to what production still serves: three layers, fully opaque, polarity inverted, and offset about half a cell so two of the four corner squares fall off the canvas. At its real 16px tab size it is a pale smear. Take the mark from `@hanzo/logo` instead — already a dependency here, already the source of the header logo — and ship the whole set it publishes: favicon.ico (6 layers, 16→256), favicon.svg, apple-touch-icon. `<head>` now declares all three; it previously declared one 696px opaque PNG and nothing else, which also suppressed `app/icon.svg` (deleted — the mark has one home now). The header mark was invisible, and the fix on main only moved which theme it was invisible in. `variant="mono"` does not inherit `currentColor` as Logo.tsx claimed — it is hardcoded #000000, so main rendered black on the black page. It is one ink on transparency, exactly like every mark in `partner-logos.ts`, so it is carried the same way: `hz-ink-black`, none on light and invert(1) on dark. The inversion lands on #ffffff/#DDDDDD, which is the `white` variant byte for byte. The chat launcher was covering the logo. `hz-fixed` sets `position: fixed` and nothing else, so a fixed element with no offsets stays where it was: top-left, 48px square, directly over the mark, on every page and both themes. Nothing could catch it — every class on the button had a rule, the missing one was never written. Added `hz-dock` as that one rule, and sized the panel against the viewport so it cannot grow past the screen it is docked to. The panel also took `hz-bg` from `hz-bg-surface`, whose dark value is rgba(38,38,38,.5): right for a card on the page, wrong for one floating over it. A fabricated trademark is gone. `lux-network.svg` was hand-coded rects and polygons spelling "LUX", nothing like the real mark, duplicated under three filenames. Lux and Zoo Labs are sibling orgs in this estate, not our clients — their marks and their entries in `clients` are both removed. A Hanzo surface carries Hanzo branding. Every share card 404'd: og:image pointed at `/opengraph-image`, a file-convention route this app does not have, while the real `/og-image.png` sat declared and unused. og:site_name was carrying the tagline and twitter:title disagreed with og:title; both now say what they are. The repo count contradicted itself in one viewport — "727+" in the hero, "2,500+" in a tile 200px below it, 3.4x apart, and both in `<head>`. Neither survives measurement: 535 non-archived public repos across the six orgs the site names. Stated once now, as a floor, with the query that produced it. "All systems operational" was a hardcoded string on every page with no health check behind it — it would have said that straight through an outage. It links to /status as "System status" instead. Verified in a real browser at 390 and 1280, light and dark: the mark matches an isolated render of the same SVG (mean abs diff 0.05/255) and dark is its exact inverse; `elementFromPoint` at the mark's centre returns its own path; the served .ico is 2b734c3b, six layers. Routes 60/60, no overflow, no console errors. Co-authored-by: Hanzo Dev <[email protected]>
260 lines
11 KiB
TypeScript
260 lines
11 KiB
TypeScript
"use client";
|
||
|
||
import { motion } from "framer-motion";
|
||
import { Brain, Shield, Network, Sparkles, Cpu, Lock, Zap } from "lucide-react";
|
||
import { cn } from '@hanzo/ui'
|
||
import site from "@/site.config";
|
||
export default function PageClient() {
|
||
const caseStudies = [
|
||
{
|
||
icon: Brain,
|
||
category: "AI Training",
|
||
title: "Training-Free GRPO: Efficient Model Adaptation",
|
||
description: "Developed Training-Free GRPO achieving dramatic training cost reduction versus traditional fine-tuning methods",
|
||
results: [
|
||
"Significant reduction in training costs",
|
||
"100x data efficiency (100 examples vs 10,000+)",
|
||
"Comparable performance to full fine-tuning",
|
||
"Open-sourced via Zoo Labs Foundation"
|
||
],
|
||
technologies: ["GRPO", "LoRA", "Zen MoDE", "Distributed Training"],
|
||
impact: "Democratizing AI training for researchers and organizations worldwide",
|
||
link: "https://github.com/zooai/gym"
|
||
},
|
||
{
|
||
icon: Sparkles,
|
||
category: "Research Publication",
|
||
title: "Active Semantic Optimization (ASO)",
|
||
description: "Published ASO framework achieving 18.2% SWE-bench resolution through training-free adaptation",
|
||
results: [
|
||
"18.2% SWE-bench resolution rate",
|
||
"BitDelta 1-bit compression for model adaptation",
|
||
"Zero additional training required",
|
||
"Published HIP-002 specification"
|
||
],
|
||
technologies: ["TF-GRPO", "BitDelta", "1-bit Compression", "Semantic Optimization"],
|
||
impact: "Advancing state-of-the-art in efficient model adaptation",
|
||
link: "https://github.com/hanzoai/papers"
|
||
},
|
||
{
|
||
icon: Shield,
|
||
category: "Post-Quantum Cryptography",
|
||
title: "Quasar: Quantum-Secure Consensus",
|
||
description: "Developed dual-certificate quantum-secure finality using BLS and Corona threshold signatures",
|
||
results: [
|
||
"Post-quantum secure finality guarantees",
|
||
"Dual-certificate architecture for safety",
|
||
"Integration with existing consensus protocols",
|
||
"29 research papers published"
|
||
],
|
||
technologies: ["FALCON", "Corona", "BLS Signatures", "Threshold Cryptography"],
|
||
impact: "Future-proofing blockchain infrastructure against quantum threats",
|
||
link: "https://github.com/luxfi/papers"
|
||
},
|
||
{
|
||
icon: Network,
|
||
category: "Consensus Protocols",
|
||
title: "Lux Multi-Consensus Architecture",
|
||
description: "Built multi-consensus blockchain with Wave, Focus, and Quasar protocols",
|
||
results: [
|
||
"Sub-second finality with Wave consensus",
|
||
"High-throughput with Focus protocol",
|
||
"Quantum-secure with Quasar consensus",
|
||
"TEE attestation integration"
|
||
],
|
||
technologies: ["Snow++", "DAG Consensus", "TEE", "Post-Quantum Crypto"],
|
||
impact: "Creating the most versatile consensus framework for diverse use cases",
|
||
link: "https://github.com/luxfi/papers"
|
||
},
|
||
{
|
||
icon: Cpu,
|
||
category: "AI Models",
|
||
title: "Zen Model Family (600M–1T+ parameters)",
|
||
description: "Released 100+ model weights spanning text, vision, video, audio, 3D, code, and agents",
|
||
results: [
|
||
"41 models across 8 families",
|
||
"1.04T MoE flagship (zen4-max, Zen MoDE architecture)",
|
||
"671B MoE (zen-max) with 71.3% SWE-bench",
|
||
"Zen5 (2T+) in training — on-chain via NVIDIA TEE"
|
||
],
|
||
technologies: ["Zen MoDE", "Flash Attention", "MoE", "Mixture of Diverse Experts"],
|
||
impact: "Providing accessible, efficient frontier AI to the research community",
|
||
link: "https://huggingface.co/zenlm"
|
||
},
|
||
{
|
||
icon: Lock,
|
||
category: "AI Economics",
|
||
title: "Hamiltonian Market Maker (HMM)",
|
||
description: "Novel invariant H(Psi,Theta) = kappa for decentralized AI compute markets",
|
||
results: [
|
||
"<200ms quote latency",
|
||
"Hamiltonian invariant pricing",
|
||
"No impermanent loss for providers",
|
||
"Published HIP-004 specification"
|
||
],
|
||
technologies: ["AMM", "Energy-based Pricing", "Compute Markets", "DeFi"],
|
||
impact: "Enabling efficient, fair markets for AI compute resources",
|
||
link: "https://github.com/hanzoai/papers"
|
||
},
|
||
{
|
||
icon: Zap,
|
||
category: "Agent Infrastructure",
|
||
title: "ZAP: Zero-copy Agent Protocol",
|
||
description: "The MCP Killer -- unified protocol achieving ~500x faster agent communication with zero-copy RPC",
|
||
results: [
|
||
"<1us local latency (vs 500us MCP)",
|
||
"1.2M/s throughput (vs 2.2k/s MCP)",
|
||
"~5% message overhead (vs 40% JSON)",
|
||
"40-50x infrastructure cost reduction"
|
||
],
|
||
technologies: ["ZAP Binary Format", "Zero-copy", "Metastable Consensus", "Post-Quantum"],
|
||
impact: "Enabling real-time agent swarms with native consensus and capability security",
|
||
link: "https://github.com/zap-proto/spec"
|
||
}
|
||
];
|
||
|
||
return (
|
||
<div className={cn("hz-min-h-screen hz-transition", "hz-bg hz-fg")}>
|
||
<main className="hz-pt-6 hz-pb-6 hz-px-4">
|
||
<div className="hz-container">
|
||
{/* Hero Section */}
|
||
<motion.div
|
||
initial={{ opacity: 0, y: 20 }}
|
||
animate={{ opacity: 1, y: 0 }}
|
||
transition={{ duration: 0.5 }}
|
||
className="hz-align-center hz-mb-7"
|
||
>
|
||
<h1 className="hz-t-5xl hz-w-bold hz-mb-5">
|
||
Research Impact
|
||
</h1>
|
||
<p className={cn("hz-container-narrow hz-t-xl", "hz-fg")}>
|
||
Delivering measurable breakthroughs in AI efficiency, cryptography, and distributed systems
|
||
</p>
|
||
</motion.div>
|
||
|
||
{/* Success Metrics */}
|
||
<motion.div
|
||
initial={{ opacity: 0, y: 20 }}
|
||
whileInView={{ opacity: 1, y: 0 }}
|
||
viewport={{ once: true, amount: 0.1 }}
|
||
transition={{ duration: 0.5 }}
|
||
className="hz-grid hz-grid-4 hz-gap-5 hz-mb-7"
|
||
>
|
||
<div className={cn("hz-p-5 hz-r-lg hz-align-center", "hz-bg-surface hz-bordered")}>
|
||
<div className="hz-t-3xl hz-w-bold hz-mb-2">130+</div>
|
||
<div className={cn("hz-t-sm", "hz-fg")}>Research Papers</div>
|
||
</div>
|
||
<div className={cn("hz-p-5 hz-r-lg hz-align-center", "hz-bg-surface hz-bordered")}>
|
||
<div className="hz-t-3xl hz-w-bold hz-mb-2">100+</div>
|
||
<div className={cn("hz-t-sm", "hz-fg")}>AI Model Weights</div>
|
||
</div>
|
||
<div className={cn("hz-p-5 hz-r-lg hz-align-center", "hz-bg-surface hz-bordered")}>
|
||
<div className="hz-t-3xl hz-w-bold hz-mb-2">{site.brand.ossRepos}</div>
|
||
<div className={cn("hz-t-sm", "hz-fg")}>OSS Projects</div>
|
||
</div>
|
||
<div className={cn("hz-p-5 hz-r-lg hz-align-center", "hz-bg-surface hz-bordered")}>
|
||
<div className="hz-t-3xl hz-w-bold hz-mb-2">4</div>
|
||
<div className={cn("hz-t-sm", "hz-fg")}>Research Organizations</div>
|
||
</div>
|
||
</motion.div>
|
||
|
||
{/* Case Studies Grid */}
|
||
<div className="hz-stack-6" data-case-studies>
|
||
{caseStudies.map((study, index) => {
|
||
const Icon = study.icon;
|
||
return (
|
||
<motion.a
|
||
key={study.title}
|
||
href={study.link}
|
||
target="_blank"
|
||
rel="noopener noreferrer"
|
||
initial={{ opacity: 0, y: 20 }}
|
||
whileInView={{ opacity: 1, y: 0 }}
|
||
viewport={{ once: true, amount: 0.1 }}
|
||
transition={{ duration: 0.5, delay: Math.min(index * 0.1, 0.3) }}
|
||
className={cn("hz-r-lg hz-p-6 hz-transition", "hz-bg-surface hz-bordered hz-hoverable")}
|
||
>
|
||
<div className="hz-row hz-ai-start hz-inline-4 hz-mb-5">
|
||
<div className="hz-none">
|
||
<div className={cn("hz-sq-7 hz-r-lg hz-row hz-ai-center hz-jc-center hz-transition", "hz-bg-inverse hz-hoverable")}>
|
||
<Icon className={cn("hz-sq-4", "hz-fg")} />
|
||
</div>
|
||
</div>
|
||
<div className="hz-grow">
|
||
<div className={cn("hz-t-sm hz-mb-1", "hz-fg")}>{study.category}</div>
|
||
<h3 className={cn("hz-t-2xl hz-w-semibold hz-mb-2 hz-transition", "hz-hoverable")}>{study.title}</h3>
|
||
<p className={cn("hz-fg")}>{study.description}</p>
|
||
</div>
|
||
</div>
|
||
|
||
<div className="hz-grid hz-grid-2 hz-gap-6 hz-mb-5">
|
||
<div>
|
||
<h4 className="hz-t-lg hz-w-semibold hz-mb-3">Key Results</h4>
|
||
<ul className="hz-stack-2">
|
||
{study.results.map((result, idx) => (
|
||
<li key={idx} className="hz-row hz-ai-start">
|
||
<div className={cn("hz-sq-1 hz-r-full hz-mt-2 hz-mr-3 hz-none", "hz-bg-inverse")} />
|
||
<span className={cn("hz-t-sm", "hz-fg")}>{result}</span>
|
||
</li>
|
||
))}
|
||
</ul>
|
||
</div>
|
||
<div>
|
||
<h4 className="hz-t-lg hz-w-semibold hz-mb-3">Technologies Used</h4>
|
||
<div className="hz-row hz-wrap hz-gap-2 hz-mb-4">
|
||
{study.technologies.map((tech, idx) => (
|
||
<span
|
||
key={idx}
|
||
className={cn("hz-px-3 hz-py-1 hz-r-full hz-t-sm", "hz-bg-surface hz-fg")}
|
||
>
|
||
{tech}
|
||
</span>
|
||
))}
|
||
</div>
|
||
<div className={cn("hz-t-sm", "hz-fg")}>
|
||
<strong>Impact:</strong> {study.impact}
|
||
</div>
|
||
</div>
|
||
</div>
|
||
</motion.a>
|
||
);
|
||
})}
|
||
</div>
|
||
|
||
{/* CTA Section */}
|
||
<motion.div
|
||
initial={{ opacity: 0, y: 20 }}
|
||
whileInView={{ opacity: 1, y: 0 }}
|
||
viewport={{ once: true, amount: 0.1 }}
|
||
transition={{ duration: 0.5 }}
|
||
className="hz-mt-7 hz-align-center"
|
||
>
|
||
<h2 className="hz-t-3xl hz-w-bold hz-mb-5">
|
||
Explore Our Research
|
||
</h2>
|
||
<p className={cn("hz-container-narrow hz-mw-md hz-t-xl hz-mb-6", "hz-fg")}>
|
||
Dive deeper into our published papers and open source projects
|
||
</p>
|
||
<div className="hz-col-row hz-gap-4 hz-jc-center">
|
||
<a
|
||
href="/research#papers"
|
||
className={cn("hz-px-6 hz-py-4 hz-r-lg hz-w-semibold hz-transition", "hz-bg-inverse hz-hoverable")}
|
||
>
|
||
View All Papers
|
||
</a>
|
||
<a
|
||
href="https://github.com/hanzoai"
|
||
target="_blank"
|
||
rel="noopener noreferrer"
|
||
className={cn("hz-bg-none hz-px-6 hz-py-4 hz-r-lg hz-w-semibold hz-transition", "hz-bordered hz-fg hz-hoverable")}
|
||
>
|
||
Open Source
|
||
</a>
|
||
</div>
|
||
</motion.div>
|
||
</div>
|
||
</main>
|
||
</div>
|
||
);
|
||
}
|