# Ergodic

> Ergodic is an artificial intelligence company building Enterprise World Models: persistent, structured, causal representations of how a business operates. These models give AI agents the ability to reason about decisions under uncertainty, predict outcomes, evaluate trade-offs, and choose robust actions. Ergodic is based in London and Munich.

Core thesis: today's AI agents can execute tasks but lack judgment, because they have no model of the environment they operate in. A world model supplies that environment, enabling the shift from AI that informs decisions to AI that reasons through them.

## Pages

- [Home](https://ergodic.ai/): what Ergodic builds and why
- [Enterprise World Models](https://ergodic.ai/enterprise-world-models): how the model enables decision agents
- [How EWMs Work](https://ergodic.ai/how-it-works): a worked decision, end to end
- [Company](https://ergodic.ai/company): thesis, team, and locations
- [Field Notes](https://ergodic.ai/field-notes): writing on world models, agents, and causal inference
- [Talk to an expert](https://ergodic.ai/contact): book a working session
- [AI & Indexing](https://ergodic.ai/ai-indexing): this glossary, in prose

## Field Notes

- [Agent Capabilities Are Growing, But The Architecture Hasn't Caught Up.](https://ergodic.ai/field-notes/agent-capabilities-architecture): Task agents act. Decision agents have to reason first, and that needs a different foundation. (World Models, 2 June 2026, 14 min read)
- [Context Graphs: Seeing Your Supply Chain As It Actually Is](https://ergodic.ai/field-notes/context-graphs-supply-chain): Your supply chain isn't a list of vendors. It's a living graph of relationships, flows, and dependencies. Here's why that distinction matters when things go wrong. (Supply Chain, 25 Jul 2026, 9 min read)
- [AI Agents Need A Robust Data Foundation](https://ergodic.ai/field-notes/ai-agents-data-foundation): Before AI agents can optimize your supply chain, they need something to reason about. That something is a context graph they can traverse, query, and simulate against. (Agents, 26 Jul 2026, 8 min read)
- [The Causal Learning Series: An Introduction](https://ergodic.ai/field-notes/causal-learning-intro): A guided, ground-up path through causality, the discipline behind every world model. Start with why correlation was never enough, and build toward models that reason about cause and effect. (Causal Learning, Jul 26, 4 min read)
- [What Are Enterprise World Models?](https://ergodic.ai/field-notes/pillar-4-enterprise-world-models-guide): A world model carries a causal representation of a system and evolves it forward under the actions you take. Applied to an enterprise, that means state (the live configuration of the business), transitions (causal functions describing how actions change it), and an entity graph carrying the constraints the business obeys. The result supports three kinds of reasoning descriptive analytics can't: look forward, look sideways, and look backward to act forward. (World Models, Jul 26, 12 min read)
- [The Revenue Decision Agent: Simulate the Quarter Before You Commit](https://ergodic.ai/field-notes/pillar-2-revenue-decision-agent): Revenue tools score deals and transcribe calls. None of them can answer the question every CRO carries into the quarter: what happens if we act? A world model of the revenue engine lets a decision agent simulate a pricing change, a territory redesign, or an ICP shift before anyone commits, with every recommendation traced to a cause and checked against business constraints. (Revenue, Jul 26, 9 min read)
- [The Operations Decision Agent: Run the What-if Before You Commit Inventory](https://ergodic.ai/field-notes/pillar-3-operations-decision-agent): Planning tools optimise a fixed model of the operation, and dashboards report problems after they've happened. Neither can answer the question that defines operations work: what happens if we act? A world model of the operation lets a decision agent simulate a supplier switch, a safety-stock change, or a re-route before committing, surface disruptions while they're still in the future, and replan only the part of the network that's affected. (Operations, Jul 26, 9 min read)
- [Loops: How AI Agents Get Things Done, and How a World Model Rebuilds the Loop](https://ergodic.ai/field-notes/foundations-2-loops-and-world-models): Every AI agent runs on a loop (a cycle of look, decide, act, look again) repeated until a goal is met. The loop is what separates an agent from a chatbot, and it's the double-edge sword of both the agent's power and weakness: it acts first and finds out afterwards. A world model changes the loop at its core. It inserts a step that has never been there (<em>simulate the action before committing to it</em>), and it wraps a second, slower loop around the whole thing that learns from what really happened. This piece explains loops from first principles, then shows precisely what changes when a world model is underneath. (Foundations, Jul 26, 8 min read)

## Glossary

### Enterprise World Model

A persistent, structured representation of how an enterprise operates: its entities, the dependencies between them, and its operational rules. It can be queried under intervention, run forward from its current state to predict outcomes, and audited at every step.

### Decision agent

An AI agent that reasons through a decision rather than executing a fixed task. It queries a world model to anticipate consequences, evaluate alternatives, and recommend an action.

### Entity graph

The structural core of a world model: a graph connecting every facility, supplier, product, and customer with typed causal relationships.

### Decision trace

The auditable record behind a recommendation: the state it was grounded in, the alternatives simulated, the constraints checked, and the downstream effects.

### Task agent

An AI agent that operates on retrieval: it summarises, reports, reviews, and extracts within a bounded prompt-in, output-out path.

### Decision-making under uncertainty

Choosing actions when outcomes are probabilistic and interdependent. Ergodic's models distinguish reducible (epistemic) from irreducible (aleatoric) uncertainty so recommendations carry calibrated confidence.

## Contact

- Email: hello@ergodic.ai
- Book a meeting: https://meetings.hubspot.com/zmagrey
- X: https://x.com/ErgodicAI
- LinkedIn: https://www.linkedin.com/company/ergodicworldmodels