Semogram Docs
Getting started

Next steps by role

Choose your next Semogram learning path: operate and review workflows, build plugin capabilities, or connect an AI assistant through MCP.

You have a first result. What comes next depends on your job.

Operators — you run the show

Review and validate changes before saving, execute separately, and inspect results and evidence. Go deeper in this order:

  1. Authoring and review — review proposals, save deliberate changes and verify execution separately.
  2. Pipeline operations — runs, schedules, monitoring and recovery.
  3. Ontology verification and prediction results — assess evidence and outcomes in the relevant feature.
  4. Skills reference — manage and evaluate reusable guidance.
  5. Troubleshooting and Getting help when you need to investigate a failure.

Plugin builders — you extend what operators can do

Build connectors, transforms, skills, and tools; operators install and run them; their operating and review evidence helps you plan the next version:

  1. Custom plugins — the builder overview and plugin lifecycle.
  2. Apply your first skill — see the install-and-run path your plugins will travel.
  3. Authoring and review — learn the loop your work plugs into, especially review and feedback.

Assisting AIs

If an AI assistant is helping, point it at Semogram for LLMs first.

Keep the first result as a baseline

Before expanding the workflow, preserve the run, saved asset versions and a few outputs you have checked. Add one source, question or instruction change at a time so you can explain how the result differs.

For predictions, define the event or numeric outcome and its evaluation window before running more cases. For external delivery, verify the target after each bounded change. A repeatable check is more useful than a larger unreviewed demo.