AI strategy

AI use cases for business: sales, marketing, HR and finance

Concrete AI use cases from sales, marketing, HR, finance and customer service, grouped by department, with the kind of tool that fits each one. It includes two questions for choosing the tool and a simple way to decide where to start.

AI use cases for business: assistant, automation or agent

"Can every department send me their ideas for using AI?" Managing directors ask it often, and what comes back is usually a long, jumbled list. Sales wants a chatbot, accounts wants to stop typing in invoices, and nobody knows which tool each idea needs. Sorting AI use cases for business by department, and by the kind of tool each one calls for, turns that list into something you can decide on.

A use case is one specific task where AI is applied. The ones that work best in a small business tend to be small, come round every week and produce something you can check in seconds. They are handled by three kinds of tool.

An assistant such as ChatGPT, Claude, Gemini or Copilot helps one person while they work. They write the request and check the answer. An automation built in n8n chains steps across your email, CRM and spreadsheets so nobody copies data by hand. An AI agent is an automation that also decides the next step based on what it receives, such as answering a question or passing it to a person.

Two questions tell you which fits. First, what starts the task? If a person sits down to write or analyse something, an assistant is usually enough. If something arriving sets it off, such as an email, a form or an invoice, an automation fits. Second, what happens if the AI gets it wrong? If the mistake is visible and fixed in a moment, light review will do. If it affects a customer, money or a person, someone should approve before anything goes out.

AI use cases in sales and marketing

These departments produce the most text and are usually first to adopt assistants. Automated workflows help most with the enquiries coming in.

  • Preparing client meetings

    The salesperson asks the assistant for a short brief on the client based on their notes, plus questions to ask at the visit. Tool: assistant, with facts checked before use.

  • Drafting proposals and follow-ups

    From the meeting notes, the assistant writes a first draft in your company's tone. The salesperson adjusts prices and terms. Tool: assistant.

  • Logging contacts in the CRM

    A workflow picks up incoming emails and web forms, extracts contact details and creates or updates the record. Tool: n8n automation.

  • Qualifying and routing leads

    Each website enquiry is sorted by criteria you set, such as area, budget or product, and reaches the right salesperson with an alert. Tool: n8n automation.

  • Product descriptions at scale

    An online shop with hundreds of items generates descriptions from a spreadsheet of product features. The workflow writes each text back to the sheet for review. Tool: n8n automation.

  • Adapting a message to each channel

    From an approved text, the assistant suggests versions for LinkedIn, email or the website, and a first translation for a fluent speaker to review. Tool: assistant.

AI in HR: use cases that keep personal data out

HR handles sensitive data. The safest use cases work with general documents and leave individual employee files out of the picture.

  • Job ads and role descriptions

    The hiring manager describes the role in a few lines and the assistant suggests a clear job ad. Check that the wording is inclusive and every requirement is really needed. Tool: assistant.

  • Answering staff questions

    Holidays, leave, travel expenses. An agent fed with your internal policies answers common questions and passes anything it cannot handle to HR. Tool: AI agent built in n8n.

  • Onboarding new starters

    An automation sends the welcome pack, creates the tasks for IT and reminds the manager of the first week's meetings. Tool: n8n automation.

  • Preparing interviews and training

    The assistant suggests interview questions tied to the skills the role needs, or an outline for internal training. Decisions about candidates stay with people. Tool: assistant.

AI in finance and administration

Admin work is full of repetitive handling of documents that arrive by email. Here, automation usually adds more than an assistant.

  • Extracting invoice data

    Incoming invoices are read, and the supplier, date, amount and VAT are recorded in a spreadsheet or your accounting software for someone to approve. Tool: n8n automation.

  • Payment reminders

    A workflow checks overdue invoices and prepares a reminder for each customer. A person approves it before it goes out, especially for key accounts. Tool: n8n automation.

  • Sorting expenses

    Expense claims are sorted by category and cost centre, and items that do not add up are flagged. The team only reviews the exceptions. Tool: n8n automation.

  • Commenting on month-end figures

    The assistant helps write the commentary for a monthly close from the figures you give it. The numbers come from your accounting system, never from the model. Tool: assistant.

AI in customer service

This is the area with the most repeated messages, and where a mistake shows straight away. Human review matters most here.

  • Reply drafts

    For each incoming email, AI suggests a reply based on your templates and product information. The support person checks it and sends it. Tool: n8n automation.

  • Sorting and routing enquiries

    Requests are tagged by topic and urgency and reach the right person. A serious complaint triggers an immediate alert to the manager. Tool: n8n automation.

  • FAQs on your website or WhatsApp

    An agent connected to your information answers questions on opening hours, delivery times or order status, and hands over to a person when needed. Tool: AI agent built in n8n.

  • Spotting recurring complaints

    Once a month, the assistant groups the enquiries received by theme and highlights the problems that keep coming back. Tool: assistant, with anonymised data.

Example

A Kent garden shop chooses its first use case

Here is a typical case: an online garden supplies shop in Kent with six staff. The owner asks each person to note, for one week, the five tasks they repeat most. Fifteen different tasks end up on a single sheet.

They put the two questions to each task. Replies to "where is my order?" arrive by email, always follow the same pattern, and a mistake gets caught if someone checks the draft before sending. Product descriptions fit too, but they are only written when a new range comes in. Payment reconciliation involves money and bank data, so it waits.

They choose reply drafts. An n8n workflow reads each email, looks up the order status and leaves a draft in the support inbox. The customer service person checks it, corrects it if needed and sends it.

Before starting, they measure two things for two weeks: the average time to first reply and how many of these emails come in. After the pilot they add a third, how many drafts need correcting. Those figures tell them whether to extend the workflow to returns.

How to choose your first AI use case and protect your data

Start with an inventory. Ask each team lead for the five tasks their team repeats every week, with a rough idea of how often they happen and how long they take. In a small business, that list fits on one sheet and usually shows where most time goes.

Then compare the tasks through three filters: they happen often, the output is quick to check, and a mistake costs little. Tasks that pass all three are good places to start. Measure the task before and after the pilot so you know whether the change is worth keeping.

HR and finance data need extra care. Payslips, appraisals, health data, bank details and customer invoices should not end up in tools your company has not approved. Check how each provider handles data on your type of account, anonymise where you can and give each automation access only to what it needs. Hiring, and any other decision about a person, stays with someone on your team.

Obinova's AI audit does this work with your team. We map each department's processes with the people who run them, prioritise the use cases with you and set the data rules, with follow-ups several times a year. We can then build the automations in n8n and train your staff.

Frequently asked questions

Which department should be first to use AI in the business?

There is no single answer. It usually makes sense to pick the area with the most repetitive, low-risk tasks, such as customer service or admin. Ask each department for its most repeated tasks, compare them on the same criteria and choose a narrow case you can measure within a few weeks. A first case that works makes it easier for others to follow.

What is the difference between using ChatGPT and automating with n8n?

With ChatGPT, Claude, Gemini or Copilot, a person writes the request and checks the result, one task at a time. An n8n automation runs on its own when something happens, such as an email or an invoice arriving, and connects several tools together. Many companies use both, depending on the task.

What is an AI agent and when does it make sense?

It is an automation that, as well as following fixed steps, decides what to do based on what it receives, for example answering a common question or passing it to a person. It makes sense when many similar enquiries arrive and the information needed to answer them is well organised. Start with a narrow scope and a clear hand-off to a person.

Can AI be used with employee data?

Only with great care. Use tools your company has approved, check the provider's terms and avoid entering payslips, appraisals or health data. Writing a job ad or a general policy needs no personal data at all. Decisions about people should always be made by people, with AI as support at most.

How do you measure whether an AI use case works?

Measure the task before changing anything, for example time spent, number of errors or response time. After the pilot, measure again using the same method. Add a quality check by a person, because working faster is pointless if the result gets worse. Those figures tell you whether to extend the case or drop it.

How much does it cost to put AI to work in a small business?

It depends on how many departments and processes are involved, the tools you already use, how complex the automations are and how many people need training. A narrow case with an assistant costs far less than several workflows connected to your CRM and accounts. At Obinova, we give you a quote after a first meeting.

Find your first use cases

Book a free 30-minute video call. Bring the tasks your team repeats most and we will tell you which ones we would tackle first, and with which tool.

Free 30-minute diagnosis →
Free diagnosisWhatsApp

WhatsApp · Reply within 2 hours during business hours