Functionary How it works
USE CASE Analyst Weekly Memo

The analyst's weekly memo, as a flow

A linear pipeline that turns a week of source material into a reviewed, sent memo — with a human sign-off in the middle.

Situation

An analyst writes a weekly memo from a recurring pile of source material: reports, feeds, meeting notes. Done by hand, it is slow and the output varies week to week — what gets cited, how findings are framed, which sources made the cut. A private chat may accelerate one memo, but it does not create a shared, repeatable process. The useful first run should become evidence for the next one.

The flow

The bundled starter contains six blocks in order. It does not currently include a scheduled trigger; add one when the flow is ready to recur:

  • Regularize — takes the incoming material and normalizes it into a consistent shape, so the downstream blocks see one format instead of five.
  • Extract Insights — reads the normalized material and emits structured findings that can carry model-supplied supporting passages.
  • Gate — pauses the run. The analyst reviews the extracted findings before any prose exists. Nothing past this point runs until they approve.
  • Summarize — condenses the approved findings.
  • Draft — writes the memo from the summary and findings.
  • Send — delivers the finished memo.

Key decisions

The Gate sits before Summarize and Draft, not after. The analyst reviews the findings — the claims and their sources — while they are still cheap to correct, before the model spends effort turning them into prose. Catching a bad finding here is one edit; catching it after the memo is written is a rewrite.

Extract Insights can attach source passages to the structured findings reviewed at the Gate. Downstream Summarize and Draft outputs retain execution lineage — their inputs, configuration, prompt, and model — but the finished prose is not currently guaranteed to retain a passage for every claim. Review source-grounded findings before synthesis when that distinction matters.

After the contracts are validated and a trigger is added, the same graph can run against each week’s inputs. Its declared structure stays legible; model-backed extraction and synthesis remain nondeterministic and should keep the evaluations and review appropriate to the memo’s stakes.

What it teaches

This is a useful sequential pattern: normalize, extract structured findings, review them, synthesize, and deliver. It can begin with an agent-backed Sketch and graduate toward these narrower blocks as repeated runs reveal the stable process. The same pattern fits a research digest, market brief, or internal status report.

Start from this template

Open the Analyst Weekly Memo starter and point it at your own sources. Resolve the currently unverified SummarizeDraft contract and Test the flow before adding a schedule or approving an external Send. Then refine it incrementally: tighten the extraction schema, add another human pause, or change the output action.