Agentic workflow examples for small businesses
An agentic workflow starts on its own — a WhatsApp message, a document arriving, a set time each evening — hands the part nobody can script to an AI agent, and ends in a step that happens the same way every time: a row written, an appointment booked, a person asked to approve. The eight below are real jobs at small businesses, the kind we keep seeing owners do by hand, in salons, clinics, driving schools, accounting firms, translation offices and loan offices. Each page says what the workflow does in neonloops today, and what it won't do.
- WhatsApp AI agent
When a customer messages your business number on WhatsApp, the agent answers from your own information, gathers what the request needs and asks you before anything changes — and once the chat ends, the request is written as a row in your table and your team is told on Slack.
- Document collection on WhatsApp
When a client starts an application on WhatsApp, the agent asks for each document your process needs, checks every file as it arrives and chases what's missing — and only once the set is complete and you approve it does your process start, with every file in place.
- Translation project management
When a customer sends a document on WhatsApp, the agent works out the language pair and the kind of document, drafts the translation and builds the file — then your translator approves it, and only then does the customer get a secure download link.
- Invoice processing
When a client sends invoices and receipts through your form or on WhatsApp, the agent reads each one, extracts the fields and codes every line against your chart of accounts — doubtful lines wait for your accountant, and the rest are written as rows and a CSV export.
- Loan application documents
When an applicant's document set is complete and approved, the agents sort and read every file, check the set against your checklist and write a one-page summary — then the summary and the complete file wait for your credit officer's approval.
- Salon booking on WhatsApp
When a client messages your salon on WhatsApp, the agent checks your booking system for a time that fits, offers it and asks for your OK — then the appointment is written into your booking system and the client gets a confirmation.
On a schedule
Every evening, the workflow reads tomorrow's appointments from your booking system and sends each client a WhatsApp reminder from your approved template.
- Driving school on WhatsApp
When a student messages your school on WhatsApp, the agent answers from your own price list, collects the registration documents and offers lesson times — and once you approve, the registration workflow starts with every document and the lesson goes into your calendar.
- Clinic appointments on WhatsApp
When a patient messages your practice on WhatsApp, the agent finds a time in your calendar, offers it and asks your staff to approve — then the appointment is written into your calendar, and any feedback or complaint is logged for your team.
On a schedule
Every evening, the workflow reads tomorrow's appointments from your practice system and sends each patient a reminder from your approved WhatsApp template.
What makes a workflow agentic?
Three parts, in this order. A trigger starts it, so it runs whether or not anyone remembered to ask. In the middle, an agent does the part you can't write down in advance: reading a payslip photographed at an angle, finding a time that suits a client, coding an invoice line it has never seen. At the end, a fixed step does the same thing every time with whatever the agent produced — a row written, a file sent as a secure link, a request held for your approval.
Take away the trigger and you have a request you must remember to make. Take away the agent and you have a fixed automation — fine when every step can be written down in advance, and then you don't need an agent at all.
Why these eight workflows?
Because they're real work, not demonstrations of what a prompt can do. Small businesses do these jobs by hand, over and over: a salon owner answering booking messages between clients, an accountant retyping receipts from photos, a translation office juggling files across chats, a driving school answering the same price question every evening.
Most of them start on WhatsApp, because that's where a small business's customers already are. Everything on these pages is something neonloops does today; each page says what the workflow depends on and what it won't do, and in every one of them a person keeps the last word.