The Core Fleet is the AI agent layer that sits underneath every Compass fleet and every Catalyst product. It handles the work that cuts across all workflows — business planning, research synthesis, strategic analysis, and technical co-development. Most organizations start here, because the Core Fleet produces measurable value immediately across functions that every business has, before any industry-specific fleet is deployed.
The architecture is designed for permanence. Agents are modular independent units, which means you add capabilities without rebuilding what is already working. Context is retained across every session, so the fleet builds a richer model of your business over time — your vocabulary, your decisions, your team dynamics, your strategic priorities. The longer the Core Fleet runs, the more accurate and useful it becomes.
Business Intelligence and Planning
Every executive team makes recurring high-stakes decisions that are constrained by the same problem: the analysis takes longer than the window. By the time research is complete, the market has moved. The Core Fleet solves this by running business intelligence continuously — monitoring the signals that matter, synthesizing across sources, and surfacing the output when you need it, not when the analyst finishes.
Business planning agents co-build alongside your leaders. Working from your actual data — financial, customer, market — they produce AI business plans with credible RoAI projections, opportunity maps that show where AI creates the most value in your specific business, and decision playbooks your team can use independently. The output is not a template or a deck: it is a plan calibrated to your operations.
Technical Pair-Programming
For development teams, the Core Fleet acts as a persistent technical co-pilot. From the first engagement, Catalyst learns your codebase — your architecture decisions, your conventions, your patterns, your preferred libraries and naming approaches. It does not give generic AI answers; it gives answers calibrated to your stack.
Teams using the technical fleet report that work expected to take weeks takes days once the fleet has learned their architecture. The acceleration compounds: each project the fleet participates in adds to its model of how your team builds. New engineers come up to speed faster. Code reviews are faster. Debugging sessions are faster. The fleet effectively preserves institutional knowledge in a form that any team member can query.
Cross-Platform and Sovereign
The Core Fleet runs simultaneously on Microsoft, Salesforce, Glean, and your internal systems — without a rebuild for each platform. When you switch a platform, or run multiple in parallel, the agents and the context they have built come with you. Deployment is on your infrastructure via Orion, and your data stays on your systems throughout.
Every capability built on the Core Fleet is portable because portability is a design requirement, not a feature added later. The open foundation means your AI investment is yours: you can inspect, fork, and extend the underlying stack independently.
- Modular architecture — each capability is an independent unit; compose them without coupling your entire stack or rebuilding what already works
- Cross-platform deployment — runs on Microsoft, Salesforce, Glean, and your internal systems simultaneously with a single agent model
- Business planning agents — co-build AI business plans, RoAI projections, and opportunity maps calibrated to your actual data and market position
- Technical pair-programming — learns your codebase, conventions, and patterns to accelerate every development sprint and compound across projects
- Context retention — every interaction builds a richer model of your business; compounding usefulness is structural, not incidental
- Sovereign deployment — runs on your infrastructure via Orion; your data and your trained agents never leave your control