Glossary
Strategy engine
A strategy1 engine is a repeatable pod of strategists, AI agents with guardrails and a red team that produces one family of strategy deliverables.
The AI-native strategy company · Starting in India
Glossary · strategy1 term
Definition
The Agent Mesh is strategy1's design for how AI agents work together inside its engines. Eight agent types, from the Orchestrator and Research agents to the Red team and Governance agents, each have one job and one guardrail, and every agent reads and writes only through the client's Strategy Twin, never directly to the others.
Explained in full on The Agent Mesh.
Context
Multi-agent AI systems can be hard to audit when agents pass work directly between themselves. Routing every agent through one shared model, with a rule a person can check, keeps the work traceable and stops any agent acting outside its job.
In practice
In every engine, the Orchestrator plans the run, Research finds sources, Data and ontology maps client data into the Twin, Modelling and finance builds the models in deterministic code, a Domain specialist applies sector knowledge, the Red team argues against the plan, Narrative drafts the memo, and Governance and audit records approvals.
FAQ
Because every agent reads and writes only through the Strategy Twin, each step leaves a record in one shared model. That keeps the work traceable, and the Orchestrator can act only through Twin actions.
No. The Modelling and finance agent's guardrail is that numbers come from deterministic code; the language model explains, it does not calculate.
Schema changes by the Data and ontology agent need human approval, the Domain specialist's work needs licensed sign-off where liability exists, and the Governance and audit agent ensures nothing is written to client systems without approval.
A 72-Hour Bootcamp takes one problem your team would scope in weeks and works it to a decision memo your leadership signs. Fixed scope, fixed fee.