Volt is a decision model trained to operate a business. It joins Slack and lives where stakeholders work. It reads and integrates into any company. Goal: growth.
Volt is a decision model trained to operate a business. It joins Slack and lives where stakeholders work. It reads and integrates into any company. Goal: growth.
Voltaire Labs is the lab that trains Volt, a decision model built to operate a business.
Volt is a decision model. It is a policy over the state of a company: it selects the action with the highest expected effect on a stated objective, executes it through the people and agents already in the company, and is updated on the measured outcome. Its objective is growth.
It reads the company's systems of record, its communication and its history of decisions. It selects what matters most now, says why, and assigns the work to the right person or the right agent. You approve. Then it measures what happened and is updated on the result.
In Slack, where your stakeholders already work. There is no new dashboard to open and nothing for the team to adopt.
The systems a company already runs: revenue and billing, the CRM, support, product, finance. It starts with read-only access.
Yes. The film on this page runs just over two minutes. The demo shows a simulated company, Larkspur, from its first day. Ask for early access and you get both.
A large language model maps a prompt to a completion. It is reactive and stateless between calls, optimised for the likelihood of text, with no objective of its own. A decision model is triggered by a change in the state of the company, its output is an executed action, it is trained on decisions and their measured outcomes, and it has an objective. For Volt the objective is growth.
Because capability is deployed only where a person has already specified a task. AI tools raise the productivity of a task; the allocation of work stays manual, so the output of the firm is roughly unchanged. Volt allocates work across the whole company against one objective.
An agent executes the task it is handed, inside one function. Volt sees the whole company at once, chooses the work that matters, and decides which person or agent does it.
No. It directs them. The agents and tools you have are what Volt puts to work.
Orchestration routes tasks someone already chose. Volt chooses them, routes them, and is scored on what happened.
No. Volt proposes and explains. Nothing runs until a named person says yes. Writing to your systems comes last, one system at a time, after the read-only value is proven.
You can, and the tools it already lives in. Volt reads your systems where they are, with read-only access you can switch off, and keeps no copy of your records. What it keeps is its own work: the decisions, your approvals, and the results.
Every call is frozen before the work runs and graded when the outcome lands. The record is yours to read, one tap from the decision.
Volt is Voltaire Labs' own model, trained on decisions and their measured outcomes. Language models read and draft. They do not hold your records.
Companies of 5 to 500 people, founder-run or not, and enterprise teams that want the same thing at their scale.
Voltaire Labs is onboarding its first design partners. A company ready to start gets Volt set up with the founder. Ask for early access, or book a call.
There is no public price. Terms are set with each design partner once Volt has proven its value inside the company.
Connect Slack and the first systems, and Volt gives its first read on Day 0. No long implementation.
Fabrice Le Parc, founder and engineer. He built the first version of Volt himself and has been building with machine learning for 8 years. His first company, Smartdate, reached $15M in revenue in its first year. HEC Paris and École Polytechnique.
Yes. Four founding seats are open: Research Scientist for training, Research Scientist for evaluation, Research Engineer and Co-founder. They are on the Builders page.
Yes. Investors can write to founders@govoltaire.com.