Small team · production engineering
We make your product AI-ready.
Your customers now work through AI agents. We give your product a CLI, agent skills and an MCP server so those agents can use it, and we build decision models that run on your own servers, not ours.
Book a consultation →Email io@deemwar.com. A person answers.
Three things we do
Each one has a fixed scope, and you own everything we build.
Make your product AI-ready
We turn your API into a CLI, agent skills and an MCP server in one week, with one of our engineers working inside your team. Fixed scope; price agreed after a first conversation.
See the week, day by day →A decision model you run yourself
Jev / openjevx is open source. The v0.4 model is an 8-bit ONNX file (598 MB, about 1.6 GB RAM loaded) that runs entirely on your own machine, on CPU. Measured on an Apple M5 Pro laptop: a yes/no answer in about 178 ms (p50), three questions in one call in about 268 ms. We never run it for you. If you would rather not do the data prep and training yourself, we do that: on our GPUs, or inside your own cloud account.
Recipes and source on GitHub →Fixed-price audit before a vendor deadline
When a vendor shuts a product down, the migration tool moves the easy parts and quietly drops the rest. We find what it leaves behind before cutover. First up: Opsgenie to Jira Service Management, with end of support on 2027-04-05.
What the migration does not move →Proof you can open
Public work, not logos. Everything below loads for anyone.
- cli.deemwar.comThe AI-ready-in-a-week offer, with the five-day plan.
- Runnable engineering examplesFree, zero-dependency, no credentials. Includes the Opsgenie migration check.
- openjevxThe open decision-model server. Source and releases.
- jevxThe client for it.
- hand-drawn-diagramsOpen-source diagram skill for coding agents.
Tell us what your product does.
One email is enough: what you build, who uses it, and what you want agents to do with it. We reply with what it would take.
io@deemwar.com →