What the week actually looks like
A short engagement mixes document review, interviews, analysis, and a deliverable draft. Each produces something you will need to cite later, but they are rarely kept together.
Interview notes are the most fragile. They live in a notebook, a transcript, or a voice memo, and the quote that anchors a recommendation gets lost between formats.
Keep one evidence file per claim
For each recommendation, keep the sources, the data, and the quotes that support it in one place. Short engagements produce ten to twenty claims; a single evidence file per claim is enough.
Each file should name the source, the date, and the person who provided it. Anonymised quotes still need a source ID, or the client cannot verify them.
Draft from evidence, then shorten
Write the deliverable from the evidence files, section by section. It will be too long. Shortening is easier than inventing the missing support later.
Ask AI to flag claims with one source and claims that rely on a single interview. Those are the sentences the client will question first.
Hand off the evidence, not only the deck
The deliverable is what the client keeps, and so is the evidence behind it. Package the sources and quotes in an appendix or a shared workspace, and name what is still an assumption.
A handoff that shows its own limits is more credible than one that looks complete. It also makes the next engagement faster.
What to measure
Measure the number of claims that can be traced to a source, and the number of client challenges that were answered without new research. Both numbers improve when the evidence lives next to the draft.
Track the time to first draft as well, but do not optimise it at the expense of the evidence chain.
How to run it on Sigla
Keep sources, interview notes, and the draft in one workspace per engagement. Use the built-in browser for pages that need a client sign-in, and ask AI to organise the evidence before it drafts. Share the appendix, not a screenshot.
For the research mechanics, read the AI deep research tool page. If the engagement produces a report from a call, continue with earnings call notes to report.
FAQ
Is client data safe in a workspace?
That depends on where the workspace runs and what your contract allows. Decide the data boundary before the first document goes in.
Can AI write the deliverable?
It can draft from your evidence. The recommendation and the framing are still yours, and the client will test both.
How do I keep quotes accurate?
Store the transcript or the note next to the quote, and check the wording before the draft goes out. Paraphrase and quote are different, and clients notice.
Does this work for short engagements?
Yes. Keep the same evidence files, just fewer of them. Short engagements benefit more, because there is less time to rebuild context.