Skip to content

Meet the agents

KOBI gives you six agents in one workspace. Each is a specialist with its own canvas, data model, and tools — but they all share the same project memory.

For ideation, brainstorming, and exploratory thinking.

A freeform React Flow canvas. Drop a thought; the agent helps you branch into chunks, concepts, and directions. Cards stay fresh — when an upstream node changes, downstream descendants are flagged for review.

Use it when you have:

  • A blank slate and need divergence.
  • A concept that needs to be broken down into sub-ideas.
  • A decision that deserves alternatives.

For SWOT, TOWS, Business Model Canvas, and Five Forces.

Pick a framework; the canvas pre-builds its quadrants, slots, and validators. You move thinking, not boxes. TOWS lets you cross SWOT outputs into action quadrants directly.

Use it when you have:

  • A market position to articulate.
  • A model to validate.
  • A team that needs a shared structure to argue inside.

For dashboards, KPIs, and data narratives.

Compose KPIs and charts as draggable widgets. Read-only at first glance, deep on demand. Export to a board template for reuse across projects.

Use it when you have:

  • Numbers that tell a story.
  • A board meeting to brief.
  • A trend to monitor across time.

For geospatial reasoning and place-based decisions.

A MapLibre GL canvas with anchors, datasets, and scope-aware hydration. Optimistic position writes keep the cursor honest while the team explores.

Use it when you have:

  • Locations that matter (stores, suppliers, demand).
  • A territory to plan.
  • A geographic dataset to layer.

For making documents into a queryable brain.

An Obsidian-style flat graph. Every uploaded document becomes a node, embedded and indexed. Other agents cite specific wiki pages when they answer.

Use it when you have:

  • Reports, contracts, meeting notes that live in folders.
  • A team that asks the same questions over and over.
  • A new hire that needs the institutional brain.

For when a canvas would be overkill.

A plain chat agent that shares the same project memory. Quick questions, fast answers, no geometry.

Use it when you have:

  • A one-off question.
  • A draft to refine.
  • A meeting recording to summarize.

Every agent reads from the same project memory. When the Knowledge agent indexes a new document, the Strategy agent can immediately cite it. When the Idea agent generates a direction, the Analytics agent can pick it up and look for evidence in your data.

You don’t need to “hand off” between agents — context follows the project, not the conversation.