The organization
This case study comes from a capacity-building services program — a team that other animal advocacy organizations apply to for support. More requests arrive than the team can take on, so deciding which to prioritize is a constant, consequential call.
The problem
Applications arrive with uneven context. Some come from organizations the team knows well; others are entirely new. Someone had to research each new organization — what it does, how it operates, and whether its public work met the program’s stated requirements — and write it up. That manual pass slowed the review, and the amount of context depended on who picked up the application.
What we built
An n8n workflow that runs the research pass automatically on every new application and delivers the results where the team already works.
It’s an AI-with-tool-use build: the AI searches the web to research each applicant, going beyond what the application itself says. The output is a source-linked research brief in the same format every time, plus a summary in the team’s Slack. The team makes every acceptance and prioritization decision outside the workflow.
How it works
- A new application lands in Google Sheets and triggers the workflow.
- The AI researches the applicant through web search using a short, approved checklist.
- It compiles source-linked findings into the same template for every applicant and marks uncertainty explicitly.
- Results are appended to a shared review spreadsheet alongside every open application.
- The same findings are reformatted into a review message.
- The summary posts to Slack, where people make the actual decision — accept, decline, prioritize, follow up — outside the workflow.
The checklist stays narrow: confirm identity from reliable public sources; record work and experience directly relevant to the program; and note clearly documented public conduct that conflicts with the organization’s published participation requirements. It excludes protected traits, private information, rumors, and irrelevant personal beliefs or associations.
What it produces
- Source-linked spreadsheet entries, one per application, from the same template
- Research notes that separate sourced facts from uncertainty
- A Slack summary for each new application
Where humans stay in the loop
The workflow supports research and context-gathering. It does not make acceptance or prioritization decisions.
A search result may be outdated, mischaracterized, or about a different group entirely. The team still reviews every application, checks the cited sources, and makes every acceptance and prioritization decision.
Why it works
- The work it replaced was repetitive. Researching an organization and writing a comparable summary followed the same steps every time — legwork the team can now skip to spend its time judging.
- A standardized template makes applications comparable. When every applicant’s context arrives in the same shape, the team compares requests side by side, rather than one person’s thorough write-up against another’s quick note.
- Every decision still gets a human review. The research arrives sooner, but the team still checks it and discusses each application.
What to copy
- Limit the workflow to research. The workflow compiles source-linked context; people make the acceptance and prioritization decisions.
- Standardize the summary format before automating it. The shared template is what turned scattered research notes into a comparable queue.
- Deliver results where your team already talks. Posting to Slack meant no new dashboard to check — the context shows up where the decision happens anyway.
The conference speaker review assistant applies the same research-then-review pattern to event submissions.
