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 Summarize → Draft 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.