Which flow pays off first
Start where frequency meets effort. Eight flows in three areas: HR, customer service, admin. An enquiry landing in three tools every day beats a monthly report.
- HR: Onboarding, Offboarding
- Customer service: enquiries, support with assignment, CRM
- Admin: documents, appointments and reminders, reporting
- Always in the background: monitoring for every flow
- Each scenario needs real sample data before we build
1. Onboarding: from yes to the first day
The offer is accepted. Checklist, access and the welcome mail run as one case. Send and activation wait for a person.
-
1. Offer accepted
Contract and start date open the case.
-
2. Checklist runs
Contract, devices, accounts and contacts stand as tasks.
-
3. Access prepared
Accounts are created, not yet activated.
-
4. Welcome as draft
AI writes the mail to the new person, with the open points.
-
5. Approval
A person approves send and activation.
- Outcome: the first day is prepared without matching lists by hand
- Verifiable: checklist, draft and approval sit in the case
- Pairs with: Offboarding, small apps, support with assignment
2. Offboarding: handover, then removal
The leaving date is set. Handover, devices and the access list are proposed. Removal waits for approval.
-
1. Leaving date set
Date, role and last tasks sit in the case.
-
2. Handover
Open tasks and knowledge go to named people.
-
3. Removal proposal
Which access and licences drop is listed.
-
4. Approval
A person confirms the removal.
-
5. Devices back
Return is tracked until it is on the case.
- Outcome: access does not linger, handover is named
- Verifiable: list, approval and return per person
- Pairs with: Onboarding, Monitoring
3. Enquiry to CRM: no lead gets lost
Forms, mail and chat become one case. The contact is created automatically; the reply waits for approval. Built for service teams whose enquiries still sit in inboxes.
-
1. Enquiry arrives
Form, mail or chat, everything lands in one inbox.
-
2. Contact created
Name, company and request are matched or created.
-
3. Draft ready
AI drafts a reply from request and context.
-
4. Approval
A person checks and approves the send.
-
5. Confirmation out
Acknowledgement goes out, the case stays open.
- Outcome: every lead as a contact with history and status
- Verifiable: arrival time, source and approval in the log
- Pairs with: intro call, quote, customer start
4. Support with assignment: ticket, draft, send after approval
Requests become a ticket, gain context and a proposed team. The reply is a draft and goes out only when someone approves.
-
1. Ticket arrives
Mail or chat becomes a case with channel and time.
-
2. Classification
AI assigns topic and urgency.
-
3. Assignment
AI proposes team and person, a person confirms.
-
4. Context attached
History, contracts and open items are added.
-
5. Reply as draft
Draft from context, with sources in the case.
-
6. Approval and send
Only after review does the reply go out.
- Outcome: faster first response with the same responsibility
- Verifiable: category, draft and approval per case
- Pairs with: enquiry to CRM, onboarding, monitoring
5. Documents into accounting: recognised, checked, then booked
Invoices from inbox or chat are read, checked against duplicates and prepared as drafts. Booking happens after human review. Reading, duplicate checks and draft preparation drop out of manual work.
-
1. Document arrives
Inbox or chat as the entry, every document counted.
-
2. Duplicate check
Checksum compared against captured documents.
-
3. Read
Amounts, dates, numbers and tax recognised.
-
4. Classify
Category and tax as a proposal, uncertain items flagged.
-
5. Approval and booking
Review the draft, approve, then book.
- Outcome: document as a draft with category, tax and check notes
- Verifiable: source, checksum, draft and approval per document
- Pairs with: reporting, monitoring, CRM from day-to-day work
6. Appointments and capacity: honest answers without list matching
Calendars and line-of-business systems say what is still free. Enquiries get a reply with real slots; a person or a clear rule confirms the booking. No more mail ping-pong. Deadlines for contracts and certificates come back as reminders, with a draft and approval before send.
-
1. Request with wish
Date, duration and request captured.
-
2. Check capacity
Calendar and systems report free slots.
-
3. Proposal
AI phrases two or three real options.
-
4. Confirmation
A person or a rule confirms the slot.
-
5. Appointment set
Invite out, case and calendar current.
- Outcome: confirmed appointment with history and owners
- Verifiable: request, slots, confirmation and invite per case
- Pairs with: enquiry to CRM, onboarding, support
7. Reporting: figures arrive on their own
The weekly report collects itself from CRM, accounting and calendar. Figures reach management and team on schedule, as a document and a message. Meetings start with facts instead of searching.
-
1. Rhythm starts
A schedule triggers the run, for example Fridays.
-
2. Read sources
CRM, accounting and calendar deliver figures.
-
3. Report drafted
AI summarises, tables stay from the system.
-
4. Quick check
Short review before the round receives it.
-
5. Report out
Document and message go out on schedule.
- Outcome: weekly report as a document plus a team message
- Verifiable: sources, timestamp and send time inside the report
- Pairs with: documents, CRM, appointments
8. Monitoring: the watcher above every flow
Every flow reports errors, retries and silent stalls to one place. The team sees at once where something sticks, with history attached. Automation stays reliable in production.
-
1. Watch runs
Success, error and retry counted per flow.
-
2. Detect trouble
Rules report errors and stalls within minutes.
-
3. Message with history
Team gets the error, history and last good state.
-
4. Retry with a limit
Defined retries, then escalation to people.
- Outcome: incident message with history instead of silent stall
- Verifiable: logs, counters and last good runs per flow
- Pairs with: every flow above
Operations and privacy in every scenario
All eight flows run on European infrastructure, usually as managed hosting. Credentials live in the instance, access is per person and revoked after the project. AI models run in the EU; no training on your data takes place.
-
Services and hosting
Managed hosting, self-hosting as an option, handover included.
-
Tools and interfaces
Which tools sit inside the scenarios: mail, CRM, accounting, calendar.
-
AI with approval
Models in the EU, drafts with review, no training on your data.
-
Intro call
Describe the bottleneck. We say honestly whether automation pays off.
Frequently asked questions
Which scenario should we start with?
The one with daily frequency and noticeable effort: usually enquiries, support or documents. Appointments and reporting follow. Monitoring runs along from the start.
How long does a first scenario take?
A tightly scoped flow can be live in a few weeks. Many interfaces or unclear ownership take longer. The estimate comes after the first conversation.
Do we need new software for this?
Usually not. n8n connects your tools; small gaps are closed by microservices or a slim app. New licences only where truly missing.
What does a scenario cost?
It depends on scope. The contact form uses budget bands from about €1,000. Binding figures come after the statement of work.
Do the scenarios run GDPR-compliant?
Instance and backups sit in Europe, access is per person, AI models run in the EU. A processing agreement is signed per project.
Can we change things ourselves later?
Yes. You receive the relevant workflows and a handover, including under managed hosting. Later changes by others are outside our responsibility unless maintenance is agreed.