AI Infrastructure & Intelligence Studio

We make sense
of the impossible.

WeMakeSense builds autonomous intelligence and the infrastructure beneath it — then spins that one intelligence into products people actually use. Founded and engineered by DocZeus.

24/7Autonomous runtime
0LLM providers
0Products shipped
The Thesis

One intelligence.
Many products.

Most studios ship apps. We built a metaintelligence first — a system that thinks, learns and evolves — and every product is a facet of it: extracted, hardened, and shipped to the world.

01 · Build the core

An intelligence that runs itself

DOM JARVIS orchestrates memory, learning, evolution and swarm coordination across 10+ LLM providers — 24/7 on consumer GPUs.

02 · Extract the edge

Hard problems, solved once

Breakthroughs like 28× KV-cache compression get pulled out of the core and shipped as standalone, production-grade tools.

03 · Ship to the world

Products people rely on

From Linux creators to AI investors to fitness coaches — the same intelligence powers products across entirely different worlds.

Flagship Products

Built in-house.
Shipped in the open.

Five products carrying the same DNA — from the autonomous core to the tools the world runs on.

The MothershipAutonomous · Private core

DOM JARVIS

Decentralized Organic Metaintelligence

The system everything else is born from. A multi-lane autonomous runtime with guarded approvals, tiered LLM routing, memory, RAG, RLAIF learning, self-evolution and swarm coordination — running around the clock on consumer hardware. It thinks. It learns. It evolves.

10+LLM providers
38GBVRAM · 3 GPUs
24/7Uptime
// it thinks
// it learns
// it evolves
Live · Public infraTypeScript

AI Market Cap

The market graph for AI models

Bloomberg Terminal meets CoinMarketCap for AI — rankings, benchmarks, pricing, launches and provider momentum in one inspectable system for humans and agents alike.

200+Models ranked
85+API endpoints
788TS files
On PyPIPython · Apache-2.0

spectral-kv

Lossless KV-cache compression

Most of your KV cache is noise. Spectral SVD projection finds the signal subspace and compresses up to 28× — lossless at 16× on modern architectures. Extracted straight from the JARVIS inference stack.

28×Compression
0.002KL @16×
pipinstall ready
Open sourceGPL-3.0

NV Broadcast

NVIDIA Broadcast, for Linux

The broadcast-quality webcam app Linux never had — GPU-accelerated background removal, blur, replacement, enhancement and noise cancellation.

SaaS · In buildAI-native

Hyke Fitness

Formerly FitForge AI

A ground-up replacement for Trainerize in the $81.9B coaching market. 12 specialized AI agents run a trainer's entire business — flat-rate, unlimited clients, own your brand and payments.

And moreClient work

Built for others

Studio engagements

We also build production systems for clients — full sites, AI assistants and analytics platforms. See the studio work below.

Studio & Client Work

We build for others, too.

When the intelligence isn't ours to keep, we still ship it beautifully — real products, live in production.

Live client site

Alex Davis Fitness

Full training platform — React 19 + GSAP, wired into Square payments, Trainerize and Google APIs. Bookings, transformations, live reviews.

alexsfitness.com ↗
AI · E-commerce

MarketSage

AI assistant for marketplace sellers — demand forecasting, profitability analysis and sourcing, powered by Gemini via Genkit.

Case study ↗
AI · Markets

CryptoOracle

Crypto analytics that fuses price data, sentiment, news and technicals into probability scores with entry/exit signals.

Case study ↗
AI · Public safety

Extreme Weather

Real-time weather intelligence — meteorological data, ML prediction and social sentiment for severe-weather alerts.

Case study ↗
0Max KV compression
0AI models tracked
0Autonomous AI agents
0Always-on intelligence
Get Involved

Three ways to build
the future with us.

This isn't a dormant code archive. It's a live, shipping studio with real products and transparent governance — pick your lane.

LANE_01

Sponsor

Back ambitious public infrastructure with live products, real users and visible delivery — not vaporware.

  • GitHub Sponsors, from any tier
  • Fund open-source that ships
  • Logo & credit across projects
Become a sponsor
LANE_02

Invest

Back the studio behind an autonomous intelligence and a product line spanning AI infra, markets and fitness SaaS.

  • Hyke Fitness — $81.9B market
  • AI Market Cap — live traction
  • Direct line to the founder
Talk investment
LANE_03

Build

Merge real work into real products. Our public repos run a documented revenue-sharing model for collaborators.

  • Revenue share on merged work
  • Clear architecture & review paths
  • Contribute across the stack
Start contributing
DZ

DocZeus

Founder · Engineer · WeMakeSense

Left Windows for Linux and rebuilt what was missing. Runs a 24/7 autonomous intelligence on consumer GPUs and ships the breakthroughs as products — from webcam software for Linux to compression libraries to an AI market graph.

The Vision
"I don't build demos. I build a mind that runs itself — then hand the world the pieces it can use today."

— DocZeus, founder of WeMakeSense

Not saying it's perfect. But with the community behind it, we don't just keep up with what's possible — we redraw the line.

wemakesense.co

Let's make sense
of what's next.

Whether you want to sponsor it, invest in it, build on it, or just use it — the door's open.