🌻 Automating the standard questions – slide creation functions#

The describes what a skilled analyst asks of a causal map and how they answer it: which filters, in what order, counting by sources or citations, with which caveats. A slide creation function (SCF) is that recipe written down as code, so the app can apply it for you. The 1-click report in the Causal Map app runs a catalogue of SCFs to draft a first set of report views.

What an SCF is#

Each SCF answers one question from the catalogue and has three parts:

What the AI does and does not do#

The deterministic part, which views to offer and how to parameterise them, is plain code. AI is used in three narrow places, each of which can fail without breaking the report:

The AI never authors filters. It only chooses from things that already exist: slides, factor names, field names.

Why the commentary is worded the way it is#

The commentary and narratives follow the reasoning rules of this chapter. They report evidence rather than fact: "12 sources said X influenced Y", never "X drives Y". Counts are treated as salience of mention, never effect size. And the is respected in the tracing views: the drivers-to-outcomes slide uses , so any pathway it narrates was told end to end by at least one source; if nobody told a whole story, there is no slide, rather than a stitched chain nobody claimed.

Where it is going#

The report reads its aim from the project description. The next step is an intake conversation that asks "what do you want to find out?", asks follow-ups until the answer is usable, and saves it there, so both the coding and the report are steered by the same stated aim.