AI agent examples: 12 jobs by department
These AI agent examples show twelve everyday jobs an agent can do in a small company, two for each of six departments. An AI agent is a software helper that reads something, thinks about it and hands back a result. Each example names what the agent reads and what you get. All twelve are jobs people do by hand today.
The 12 jobs at a glance
| Job | The agent reads | You get |
|---|---|---|
| 1. Sales: preparing a meeting | Notes and emails on one customer | A half-page summary before the call |
| 2. Sales: follow-up | The last contact with a customer | A draft email to edit and send |
| 3. Projects: task summary | One task and its history | Three sentences on where it stands |
| 4. Projects: brief | A client's answers in a form | A tidy project brief |
| 5. Service: sorting requests | The text of a new request | Topic, urgency and a two-line summary |
| 6. Service: reply | The question and your handbook | A draft answer that names its source |
| 7. Finance: reading invoices | A scanned supplier invoice | Supplier, date and amounts, filled in |
| 8. Finance: sorting expenses | The description of an expense | A suggested category |
| 9. Staff: applications | A CV and the job requirements | Notes for the person who hires |
| 10. Staff: job advert | A few bullet points | A complete draft advert |
| 11. Management: goal check | One quarterly goal | Whether it can be measured, with better wording |
| 12. Management: reading a chart | The figures behind one chart | What stands out and two possible next steps |
Who checks what
The jobs fall into three groups by how much checking they need.
- Read and use (1, 3, 12). These are summaries. If one is off, the reader notices and loses a minute.
- Confirm before it counts (4, 5, 7, 8, 9, 11). The agent fills in or suggests, and a person confirms. In job 9 the agent writes notes. The decision about the applicant stays with a person.
- Edit before it goes out (2, 6, 10). Anything a customer or applicant will read is a draft until a person sends it.
Two examples in more detail
Sorting requests (5). A customer of a bakery writes: "My standing order for Friday was two loaves short." The agent labels it as a delivery problem with normal urgency. It adds: "Friday order short by two loaves. Customer asks for a correction."
Goal check (11). A team writes "Improve customer service" as a goal for the quarter. The agent answers that the goal has no number, and suggests "Answer every request within one working day". The team decides whether that is the right target.
What these examples have in common
- Each agent has one job that fits in a sentence.
- Each works on one thing at a time: one customer, one task, one invoice.
- Each returns something short enough to check.
- None of them sends, pays or deletes by itself.
Choosing your first one
Pick the job from the table that comes up most often in your company and annoys the person doing it. Try it on last week's real cases. If the results are useful, add a second job in the same department.
How Worknano handles this
Several of these jobs exist in Worknano as ready-made agents. The CRM Assistant sums up what is known about a company and suggests changes to deals for you to confirm. The Task Summarizer writes two to four factual sentences about one task. The Key Result Validator checks whether a goal's key result can be measured and recommends only what should change. You can also set up agents for the other jobs, such as scoring an application, sorting an expense or reading a scanned invoice. They run on one record, on many at once or inside a workflow.
- CRM Assistant: the agent that keeps companies, contacts and deals in order
- Task Summarizer: a short, factual summary of one task
- Key Result Validator: checks that a key result is measurable
Create a free workspace and try the ready-made agents on your own records.