Process automation: examples you'll recognize in your company
The best way to understand automation isn't theory: it's examples. What gets automated today in sales, customer service, administration and operations — and where to start.
"So where would this automation thing apply in my company?" is the most common question — and the healthiest one. Automation isn't understood in the abstract: it's understood by looking at the concrete processes where it already works. These are the most frequent examples by department, with the pattern they share and the criteria to decide where to start in your case.
Sales
- Lead capture: the request that comes in through a form, WhatsApp or email is logged in the CRM automatically, with its source and complete information
- Follow-up: reminders and follow-up messages triggered by the lead's stage, instead of depending on the salesperson's memory
- Proposals: building the proposal or pre-project from the catalog and the requirement, so the salesperson reviews and closes instead of drafting
- Sales reports: the weekly pipeline snapshot is generated automatically, with CRM data, not from a spreadsheet assembled by hand on Friday
The pattern: the salesperson sells; the workflow captures, reminds and documents.
Customer service
- First response with context: the customer gets a reply with their history at hand — order, status, previous interactions — on the channel where they wrote
- Automatic logging: every conversation is recorded in the ticketing system or CRM without anyone transcribing it
- Classification and routing: the request reaches the right team based on its type, urgency or customer
- Escalation: cases that require judgment go to a person with all the context already gathered
Administration and finance
- Invoices by email: extracting the data, validating it against the purchase order and recording it in the accounting system
- Reconciliation: matching bank transactions against records, leaving only the differences for human review
- Collections: scheduled payment reminders and follow-up on overdue invoices, with every contact recorded
- Expenses and travel: from the photographed receipt to the classified record, with the expense policy applied as a rule
Operations
- Orders: from the confirmed order to the internal work order, without re-keying between the sales system and the operations system
- Inventory alerts: automatic notices for minimum levels, expiration dates or stockouts, instead of manual checks
- Operational reports: daily or weekly KPIs are published automatically from the source systems
- Process documentation: logs and quality records filled in as part of the workflow, not as a separate task
Human resources
- Onboarding: account provisioning, welcome emails and onboarding checklists triggered from the hiring event
- Internal requests: vacation, employment letters and leave as workflows with approval, not as email chains
- Answers for the team: frequent internal policy questions answered from the company's documentation
Real WITS examples
Two published cases illustrate the pattern at different scales:
- ABSA (industrial distributor): an AI chat that understands the requirement in natural language, builds a pre-project with components from the catalog and connects to the ecommerce site — leads receive an automatic proposal and the sales team focuses on closing
- DaLi / Coca-Cola: campaigns of more than 500 million coupons, with the validation and redemption of each coupon processed by automated workflows on distributed infrastructure
The details of each case are in the case studies section of the site, with the problem, the solution and the result.
Which one to automate first?
Not the flashiest one: the one that best combines three factors.
- 1Frequency: how many times a day or a week it happens. A daily process pays back automation much sooner than a quarterly one.
- 2Clarity of rules: what percentage of cases are handled the same way. The more exceptions, the more design the workflow requires.
- 3Real pain: how much time it consumes today and how much each error costs. The process the team hates is usually a good candidate.
And one golden rule: measure the baseline before automating. Without that number, you'll never know whether it worked.
What you may also be wondering
Where do you start automating in a company?
With a single process: frequent, with clear rules and obvious manual work today. You map how it runs, design the workflow, connect it to the existing systems and measure before and after. With that learning, you decide the next one.
What happens with the cases that don't follow the rule?
They're designed in from the start: the workflow handles the repetitive work and escalates to a person whatever requires judgment, with the context already gathered. A good automated workflow includes people at the right points.
Do these examples need AI or just rules?
It depends on the step. Moving data and triggering notifications are rules. Reading emails, extracting data from invoices or handling conversations need AI. Most real workflows combine both.