Use Cases
Every one of these is work a company is already doing — usually by hand, usually by someone who should be doing something else. Each page below is a complete build: the problem, the architecture, the configuration that matters, what happens at run time, and what you get back.
Pick the one closest to your own annoying Tuesday and follow it end to end.
The catalogue
| Use case | What it stops | Built from |
|---|---|---|
| Support ticket triage | Outages buried under password resets; SLA misses nobody noticed until the customer did | Task trigger, Smart Agent, Condition, Approval, Track Event |
| Invoice and expense processing | A finance analyst reconciling a spreadsheet for two days a month | Task trigger, Node.js, Each loop, Track Event, Publish File |
| Contract review and renewal tracking | Auto-renewals nobody caught, contract terms nobody can find | Sandbox, Smart Agent, Flow Memory, scheduled digest |
| Security questionnaires and RFPs | A week of an engineer's life per deal, answering the same 300 questions | Data sources, Smart Agent, Skills, Each loop |
| Employee onboarding assistant | The same forty questions in Slack, every single month | Peer, data sources, Slack/Teams channels, Evaluation |
| The weekly executive report | Three people copy-pasting from four systems every Monday | Schedule trigger, HTTP steps, Smart Agent, PDF output |
| Lead qualification and CRM hygiene | Sales working stale leads while good ones go cold | Peer with tools, agent approval, enrichment flow |
| Customer feedback analysis | 4,000 survey comments nobody has read | Classifier, Each loop, Track Event, Reports |
Customer-facing
Support ticket triage that escalates safely
Tickets arrive faster than anyone can read them. The genuine outage sits behind eleven password resets, and nobody wants an AI deciding on its own to tell a customer their data is gone.
This build classifies every ticket, routes it, and stops at a human before anything consequential leaves the building — then records what it decided so you can prove the SLA later.
Customer feedback that someone finally reads
Four thousand free-text survey answers, one product manager, zero chance. This build turns them into a categorised, quantified picture — with the actual quotes attached, so the number and the evidence arrive together.
Back office
Invoice and expense processing
The monthly export lands, and a finance analyst spends two days turning it into a reconciled spreadsheet. This build does the pass in one run, records one event per line item, and hands back a CSV — plus a spend breakdown by customer that nobody had to build.
Contract review and renewal tracking
Contracts live in a folder nobody opens until something auto-renews. This build reads them in a sandbox, extracts the terms that matter, remembers them, and tells you sixty days before a renewal instead of sixty days after.
Knowledge and internal
The onboarding assistant that knows your policies
Every new joiner asks the same forty questions, and every answer costs a senior person twenty minutes. This build answers them from your own documents, in Slack or Teams, citing the source — and says "I don't know" rather than inventing a policy.
Security questionnaires and RFPs
Every enterprise deal ships with 300 questions you have answered before, in slightly different words. This build drafts the answers from your existing documentation and flags the ones a human must actually decide.
Revenue and reporting
Lead qualification and CRM hygiene
Sales works whoever is at the top of the list, which is rarely whoever is most likely to buy. This build enriches and scores inbound leads, writes back to the CRM — and asks before it writes anything.
The weekly executive report
Monday morning, three people pull numbers from four systems and paste them into a deck. This build assembles the whole thing on a schedule and fails loudly when a source is down, instead of quietly producing a report with a hole in it.
How to read these
Each page follows the same shape:
- The problem — what it costs today, in hours and in misses.
- What you build — the architecture in one diagram.
- Implementation — step by step, with the configuration that actually decides whether it works.
- What happens at run time — the walkthrough, including the parts that pause.
- What you get — outputs, files and reporting.
- Extensions and pitfalls — where teams take it next, and where they get burned.
If you are new to Flows, read Projects & Flows first — every scenario assumes you know what a project, a trigger and a final step are.