Editing Openai/6939b23e-87cc-8001-bdd9-9546d1143ce9 (section)

==== Here’s how Python orchestrates it: ====
# User prompt → Python - Python sends prompt to AI model for table generation.
# AI starts generating - Token by token, using context + trained patterns. - Reaches unknown info → outputs [MISSING: …].
# Python reads the output - Detects [MISSING: …] tag. - Interprets it as a request for external data.
# Python queries source(s) - Can query: - A local table or database - Another AI model - API, knowledge base, or any function
# Python collects the information - For example: “efficiency = 15-20%”
# Python appends new info to AI context - Either as part of the next prompt or inserted inline:

<syntaxhighlight>tech_output: 15-20% efficiency
Continue generating table…

</syntaxhighlight>
# AI resumes generation - Uses the new info as if it had “paused and looked it up.”