Automating Customer Onboarding with Open Source Tools

Closing a deal feels like the finish line. For your customer, it’s the starting line.

The first week after a contract is signed is when a customer decides whether they made the right choice. If onboarding is smooth, they feel confident. If it’s a mess of forgotten emails, manual setup steps, and “let me get back to you on that,” they start wondering if they should have gone with someone else.

Most businesses handle onboarding the same way: a spreadsheet tracking where each new customer is in the process, a folder of templates someone copy-pastes from, and a lot of hoping nothing falls through the cracks. It works for the first ten customers. By customer fifty, it’s a liability.

The fix isn’t an expensive onboarding platform. It’s connecting the open source tools you’re already running into an automated pipeline that handles the repetitive work — so your team can focus on the parts that actually need a human touch.

Here’s how to build one.

What Customer Onboarding Actually Involves

Before we talk tools, let’s map out what a typical onboarding process looks like. Whether you’re a consulting firm, a SaaS company, a marketing agency, or a managed IT provider, the steps are roughly the same:

Step 1 — Contract and kickoff. The deal closes. You need to send a welcome packet, schedule a kickoff call, and collect any remaining paperwork.

Step 2 — Account setup. Create the customer’s account in your systems, provision access, set up their workspace, and configure initial settings.

Step 3 — Data and integrations. Import the customer’s existing data, connect any third-party tools, and verify everything is talking to everything else.

Step 4 — Training and documentation. Get the customer up to speed on how to use whatever you’ve set up for them. Share documentation, schedule training sessions, and answer questions.

Step 5 — First milestone check-in. After a week or two, follow up to make sure things are working, answer lingering questions, and confirm the customer is getting value.

Every one of these steps has repetitive, mechanical parts that can be automated — and parts that genuinely need a human. The goal is to automate the former so your team has time for the latter.

The Tools You’ll Need

If you’ve been following this blog, you probably already have most of these running. Here’s the stack:

All open source. All self-hosted. No per-seat SaaS fees. No sending customer data to third-party clouds.

If you don’t have all of these yet, that’s fine — you can start with just n8n and Odoo and add the others as you go.

Building the Onboarding Pipeline

Let’s walk through each stage of the pipeline and what happens automatically versus what requires human attention.

Stage 1: Deal Closed — Kickoff Trigger

The trigger is simple: when a deal moves to “Closed Won” in Odoo, n8n fires off the onboarding workflow.

Here’s what happens automatically:

1. n8n detects the deal stage change in Odoo via its webhook trigger.
2. A welcome email is drafted by Ollama using the customer’s name, company, and project details from the Odoo deal record. The email includes a Cal.com booking link for the kickoff call and a brief overview of what to expect over the next two weeks.
3. An internal notification is sent to your team channel (Mattermost, if you’re self-hosting that too) with a summary of the new customer and assigned team members.
4. A project is created in Odoo with pre-filled tasks for each onboarding step, assigned to the right team members with due dates.

What a human does: Reviews the AI-drafted welcome email before it goes out. This takes about 30 seconds and ensures the tone is right for this specific customer.

Here’s what the n8n workflow looks like in plain terms:

Trigger: Odoo deal stage → "Closed Won"
  → Fetch deal details + customer info from Odoo
  → Send to Ollama: "Draft a welcome email for {customer_name}
     at {company}. Project: {project_description}.
     Include booking link: {cal_com_link}"
  → Create Odoo project with onboarding task template
  → Send Mattermost notification to team
  → Hold email in draft for human review

The key design decision here is holding the email in draft rather than sending it automatically. AI is good at drafting, but you want a human to hit send on the first message a new customer receives. This is one of those tasks where full automation would feel impersonal — and first impressions matter.

Stage 2: Account Setup and Provisioning

Once the kickoff call is scheduled (the customer picks a time through Cal.com), the next set of automated steps kicks in:

1. n8n creates a Nextcloud folder for the customer, named with their company name, and generates a share link with appropriate permissions.
2. A BookStack space is created for the customer’s documentation, pre-populated with your standard onboarding guide template.
3. If paperwork is needed, n8n creates a Documenso envelope with the right template (NDA, service agreement, etc.) and sends it to the customer.
4. Access credentials and onboarding materials are organized in the Nextcloud folder, ready for the kickoff call.

What a human does: Reviews the Nextcloud folder structure and BookStack content to make sure everything is correct and customized for this customer’s specific needs.

The time savings here are significant. Manually creating folders, setting permissions, copying documentation templates, and preparing paperwork can take 30-60 minutes per customer. Automated, it takes about 30 seconds of processing time plus 5 minutes of human review.

Stage 3: The Kickoff Call and Data Collection

The kickoff call itself is a human activity — and it should stay that way. But everything around it can be automated:

1. Before the call, n8n sends the customer a reminder email 24 hours in advance with the meeting link and a brief agenda.
2. During the call, your team takes notes (or records the call if the customer consents).
3. After the call, if a recording or transcript is available, n8n sends it to Ollama to:
   • Extract action items and create tasks in Odoo
   • Identify any data or assets the customer needs to provide
   • Generate a call summary to share with the customer
4. A follow-up email is drafted with the summary, action items, and a request for any outstanding information.

This is where Ollama running locally becomes important. Customer call recordings and transcripts often contain sensitive business information. Sending that to a third-party AI API means your customer’s confidential data is now sitting on someone else’s servers. With Ollama, the processing happens on your infrastructure. The data never leaves your control.

Stage 4: Training and Documentation

This is where many onboarding processes fall apart. The customer needs to learn how to use whatever you’ve set up, but training is often ad hoc — a screen share here, a rushed email there, and then the customer is on their own.

Here’s how to systematize it:

1. Your standard documentation lives in BookStack, organized by topic. You build this once and reuse it for every customer.
2. When the project is created in Odoo, n8n automatically creates a “Review Documentation” task with links to the relevant BookStack pages based on the customer’s project type.
3. Ollama can generate customer-specific documentation — for example, if you’ve set up a custom Odoo configuration, Ollama can draft a user guide tailored to their specific workflows and modules.
4. Training sessions are scheduled through Cal.com, with automated reminders sent before each session.
5. After each training session, n8n creates a feedback form (using OpnForm, if you’re self-hosting that) and sends it to the customer.

The result: every customer gets the same quality of training materials, customized for their specific setup, without your team rewriting documentation from scratch each time.

Stage 5: The First Milestone Check-In

Two weeks after onboarding starts, n8n triggers a check-in workflow:

1. A status report is generated by pulling task completion data from Odoo and summarizing it with Ollama.
2. A check-in email is drafted for your team to review and send, asking how things are going and offering to schedule a follow-up call.
3. If any onboarding tasks are overdue, the assigned team member gets a reminder.
4. A customer satisfaction survey (OpnForm) is sent to gather feedback on the onboarding experience.

This is also valuable data for your business. Over time, you’ll see patterns — maybe customers consistently struggle with a particular setup step, or training sessions on a certain topic get low satisfaction scores. That’s your roadmap for improving the onboarding process itself.

What Not to Automate

Not everything in onboarding should be automated. Here’s where to draw the line:

The kickoff call. This is your first real conversation with the customer as a paying client. It needs to be human, attentive, and responsive. Don’t try to automate it.

Personal check-ins. If a customer is frustrated or confused, an automated email makes it worse. Train your team to recognize when to pick up the phone.

Custom configuration decisions. AI can suggest configurations based on what the customer described in their intake form, but the final call on how to set up their systems should involve a human who understands their specific context.

Relationship building. Onboarding isn’t just about getting technical setup done — it’s about building trust. Automated workflows handle the process. Humans build the relationship.

A Realistic Time and Cost Comparison

Let’s put some numbers on this. Here’s what onboarding typically looks like before and after automation, based on a consulting firm onboarding 10 customers per month:

At 10 customers per month, that’s 25 hours saved monthly — roughly $2,500-3,000 in labor costs at typical consulting rates, assuming $100-120/hour. The infrastructure cost is essentially zero if you’re already self-hosting these tools.

More importantly, the automated process is consistent. Every customer gets the same thorough onboarding. Nothing gets forgotten because someone was busy that day. No step gets skipped because a team member was out sick.

Getting Started

You don’t need to build the entire pipeline at once. Here’s a practical rollout plan:

Week 1: Set up the deal-closed trigger in n8n that creates the Odoo project and drafts the welcome email. This alone saves 20 minutes per customer and ensures nothing is missed in the critical first hours after a deal closes.

Week 2: Add the account provisioning steps — Nextcloud folder creation and BookStack documentation setup. This is where the biggest time savings are, since manual setup is tedious and error-prone.

Week 3: Add the post-call processing with Ollama — action item extraction and call summaries. This turns your team’s call notes into structured tasks without manual data entry.

Week 4: Add the milestone check-in workflow. This closes the loop and gives you data to improve the process over time.

Each week adds one layer of automation. By the end of the month, you have a complete onboarding pipeline that runs itself — with humans in the loop at the points where they add real value.

The Bottom Line

Customer onboarding is one of the highest-leverage processes in your business. It’s the moment a prospect becomes a customer, and how smoothly that transition goes shapes their entire relationship with you.

The tools to automate it are free, open source, and likely already running on your server. What’s missing is the workflow that connects them — and that’s a few days of setup, not a multi-month implementation project.

Start small. Automate the welcome email this week. Add the folder setup next week. Before you know it, your onboarding process runs itself, and your team is spending their time on the work that actually needs a human — building relationships, solving problems, and making customers successful.

Want help setting up an automated onboarding pipeline for your business? ARDOT Consulting builds custom automation workflows with open source tools — no SaaS subscriptions, no vendor lock-in, no data leaving your control. Get in touch and let’s map out your onboarding process.