Product
One platform, from your phone to your rack.
Apps, harness, models and machines, built around a model you own. Every part says what runs today.
- 01Apps
- 02Harness
- 03Models
- 04Machines
- Less forecast error than its base model, in our first training run
- 17.8%
- To roll back to the last good release, by design
- 60 s
- Ways to run it: Local, Hybrid or Cloud, as planned
- 3
- Named owner for every model, on its Model DeedThe Model Deed is in development.
- 1
Our first training run cut forecast error 17.8% against its base model. It is not yet ahead of a simple last value baseline, and we publish that too. Read Benchmark 001
The Formula
Your work makes a model of its own.
The model is yours, with a signed Deed.
Four layers
One system. Four layers.
- 01AppsWhere people meet their model.
- Eigna Companion, planned
- Eigna Console, in development
- Connectors and API, planned
- 02HarnessWhat answers, routes and acts.
- A decision model that says “no call” when unsure
- Agents, tools and memory
- A prototype runs on test data
- 03ModelsWhat you own.
- Experts on one open base
- Your weights, on record
- The Model Deed, in development
- 04MachinesWhere it runs.
- An Eigna Box, planned
- Your own servers
- A dedicated GPU in your region
How it learns
Keeps learning. Released only when better.
Your people approve the work. A new version trains on it, is tested on a sealed set it has never seen, and goes live only when it beats the one you use today.
Release gate
Release 13 against release 12, sealed test set
Forecast error, indexed to the live version. Lower is better.
Illustrative data. The live version only changes when a release beats it.
The Model Deed
Ownership is a document.
The owner, the release, the date and the fingerprint of the weights, signed by both sides. Each release that passes the gate adds a line.
Where it runs
Local or cloud. On your hardware or ours.
Security and control
Control you can prove.
- Keys held by youYour data on your own volumes, with keys you hold.
- Signed installerInstalled from signed, versioned templates.
- Audit recordEvery release, its test result and who approved it.
- RollbackBack to the last good release, designed for 60 seconds.
What runs today
What runs, and what is next.
Runs today
- The first expert
- The scoring harness and sealed tests
- Training
- benchmarks.eigna.ai
- A prototype of the app flow
In development
- Eigna Console
- The Model Deed
Planned
- Learning in production
- Run modes as a product
- Eigna Companion
- Connectors
- The first Eigna Box
- Experts beyond forecasting
- A tuned decision model
- Audit log and refusal training
Target
- Decision model under 10 ms
- 80 to 90% of requests
- A new install in under two weeks
- Rollback in 60 seconds
Questions and answers
Do we need an ML team?
No. We deliver it, set it up and run it. Your team names the task, approves the work and decides each release.
Which models do you build on?
Open models, chosen per task and never tied to one lab: for example Qwen, Llama, Mistral, Gemma, Kimi and gpt-oss. Every base model passes a licence check before we use it.
What runs today?
The first expert, the scoring harness and sealed tests, training on local machines and H200 GPUs, and benchmarks.eigna.ai. The Console and the Model Deed are in development. The Companion app, the run modes and the first Eigna Box are planned.
Start
Bring one task. See the proof first.
Thirty minutes with a founder, or two minutes to see if it fits.