labs@atheron:~/locations/united-states/ai$ train agent --grounded

AI for US companies

Assistants and agents with written limits.

We build AI assistants and agents for US companies that want them grounded in their own documents, limited in writing and priced honestly, with the build and the monthly running cost shown separately.

Start a project Estimate your project

// why it differs here

What matters for this work here.

Grounded, cited, tested

Assistants answer from your own documents, cite them, and are tested against real questions before launch. When they do not know, they say so.

The model that fits

Any frontier model, an open-source model on our servers, or a custom one. We recommend one and price each route on its own, build and running cost apart.

Limits you can read

Every agent has written limits: what it may do, what it must ask a person about, and a log of what it did. It always says it is an AI.

// example project

One we could build for you.

Hypothetical, labelled as such, and priced from our rate card when you open the page.

Example projectRetail and e-commerce

Orders and returns agent

An agent in chat and email that looks up the order, starts a return within your policy, answers sizing from your own size guides and hands anything unusual to a person with the history attached. It runs on a frontier model through its API, which handles high volume well and is the quickest to start; only what each answer needs is sent.

// Example project

Build, In Canadian dollars: CAD 133,500 to 204,100, delivered within 21 weeks
Build, In US dollars: USD 93,700 to 143,300

Converted at 1 CAD = 0.70 USD, the rate on our rate card. The price itself is in Canadian dollars.

Open in the estimator Full breakdown

// services

The services behind it.

~/ai

AI agents and models

Frontier, open-source or custom LLMs, put to work on your data: agentic systems, retrieval, fine-tuning, inference and evaluation, from a team that trains its own models.

~/bots

Discord and Telegram bots

Bots for moderation, verification, rewards, support and commerce, which your team runs from its own admin screens.

~/data

Data and analytics

Data warehouses, pipelines, dashboards and reports, and the quality checks that make the numbers worth trusting.

// questions

What buyers here ask us.

Can health data stay out of public models?

Yes. We can run an open-source model on servers in the United States, with the HIPAA controls built around it. HIPAA compliance is yours; we build to its controls.