Open-source Rust infrastructure for agentic AI. We build it, we run it, we ship it.
We write Rust. We build the servers, the execution engines, the coding agents, and the search layers that let AI agents actually do work — not just talk. Long-horizon workflows. Systems that run 24/7. Infrastructure that builds itself, heals itself, and improves itself.
The human stays in the loop. The agent does the heavy lifting. Tools, skills, memory, and the human — all of it works as one system. That's not a pitch; it's how we operate every day.
Our infrastructure is modular by design. Each layer is a standalone Rust crate or service with a clear responsibility. They compose vertically: agent runtime at the base, coding agent on top, config management across everything.
-
ares — Agentic AI server. Multi-provider LLM (NVIDIA NIM, Ollama, Groq, Anthropic), structured tool calling, RAG, MCP integration, multi-tenant metering. The runtime layer that manages agents, routes requests, and exposes an OpenAI-compatible API.
-
eruka — Context intelligence engine. Structured business knowledge, workspace isolation, 13 MCP tools, the biological memory layer for agents that need to understand what they're working with.
-
pawan — Self-healing CLI coding agent. 18 subcommands, 17 tools, streaming TUI with markdown + vim keybindings. AI commit, review, explain, test analysis, watch mode. Named after Power Star Pawan Kalyan. Replaces Claude Code with a self-hosted, zero-subscription alternative. docs
-
aegis — Declarative config management. TOML manifests that generate tool configs for the entire stack.
-
daedra — Web search MCP server. Rust-native, DuckDuckGo-powered. Gives agents real-time web search and page fetching.
-
nimakai — NVIDIA NIM model latency benchmarker. Written in Nim. Measures ping, tool-use, and agent task times across all NIM models.
-
thulp — Execution Context Engineering Platform. Skills, memory protocols, environment awareness, identity for agents.
-
dui — Component library for Leptos WASM frontends. 29 accessible components, published on crates.io.
DolTARES is our Rust server that ties it all together — powered by ares, thulp, and daedra. It handles chat, workflow orchestration, scheduling, channel delivery, self-healing, and long-horizon DAG execution. DolTARES is where the open-source pieces meet production.
Doltdot is the AI agent that runs on DolTARES. It's live — handles real tasks, research, development workflows, automated pipelines, communication. We use it internally to run and improve the very infrastructure it sits on.
We run on DTrain — a 6-phase circular lifecycle that takes any operation from manual to autonomous:
DSprint (discover) → DBuild (develop) → DLaunch (deploy) → DWatch (monitor) → DTune (improve) → DGrow (scale)
DIRMACS is its own first client. We're at Sprint 73 of a 90-sprint bootstrap, building toward a self-operating AI platform. Pawan executes the sprints. Ares runs the agents. Eruka holds the context.
- Rust-first. Memory safety, performance, correctness. Agentic systems need to be reliable at runtime, not just at demo time.
- Composability over monoliths. Each crate does one thing well. They compose through clean interfaces.
- Verification over speed. "Autonomous AI execution without verification gates produces confident fiction." Every deployment is proven with actual command output.
- NVIDIA downstream. We build on NVIDIA NIM as our primary inference layer. Downstream integrators with upstream compute.
- Dogfooding. We run our own agents on our own infra. Pawan improves pawan. Ares serves ares' agents. If it breaks, we feel it first.
AGI-approved infrastructure. We're not waiting for AGI to show up — we're building the stack it'll need when it does. Automating startup, software development, research. We're accelerating with a system that runs 24/7 and we're doing it in the open.