Coconut offers powerful ways to gather information from a wide array of sources (such as internal systems, third-party APIs, and the web itself) and turn it into highly actionable insights.
Create a custom knowledge agent
It starts with the creation of a custom agent that’s configured with your intelligence gathering goal. Common examples include (although are certainly not limited to) the following:
- Market analysis
- Competitor research
- Idea validation
- Customer discovery and enrichment
Coconut has a large registry of reusable agents
Please use the Firecrawl MCP server to crawl each of the following links in: .nut/context/knowledge/sphere-links.md and put a summary of your findings in a new document called: .nut/context/knowledge/sphere-gtm.md
Note the .nut/context/knowledge/filename.md path is a way of referencing knowledge base files (new or existing) within a Coconut’s filesystem
Set the schedule
The agent can then be run on-demand or scheduled to run on a pre-determined cadence (such as once a day or every Friday at noon).
Use your research
Everything that lives within a Coconut instance can be referenced by everything else (such as coding or image generation agents). Now that you have research, use it to create a visual representation of your findings with Nano Banana Pro.
