In-house team or external provider to automate processes with AI?

Medium-sized and larger companies raise this most: they'd rather have their own team and doubt what an outsider adds. It's a fair concern. Nobody knows the business like the people in it, and depending on a third party for something that runs every day is daunting. The question is what that team costs and when it pays off.

Why do so many companies prefer an in-house team?

• Business knowledge. People inside know how the process really works, not just how it's written down.
• Control. Priorities and changes are decided without going through an external quote.
• No dependency. If the provider disappears or raises prices, the business doesn't stop.

All three are legitimate. The third weighs most, and it can be solved without building a team, as explained below.

What does an in-house automation team need?

Automating with AI isn't a single role. A team that works needs:

• someone who understands the processes and knows what's worth automating
• someone who connects systems (email, ERP, CRM, industry software) through their APIs
• someone who designs and supervises what the AI does: what it decides, what goes to a person and how it's checked
• time to maintain it all, because software changes and so do processes

According to Spain's statistics institute (INE), in the first quarter of 2026 only 15.4% of companies with 10 or more employees had ICT specialists on staff. Most would have to start that team from scratch. And each person costs €3,388 a month on average according to the Quarterly Labour Cost Survey, before anything has been automated.

In-house team, provider or both

How to work with a provider without depending on it

Require this from the start:

1. Everything in your accounts. Email, servers, automation tools and AI provider keys in the company's name, not the provider's.
2. Documentation for every process: what it does, with which data and what happens when something fails.
3. Full access to what's built, so another team can take it over.
4. No lock-in. If the service doesn't deliver, you leave and what was built stays.
5. Training for one in-house person to supervise and request changes.
6. Whoever assesses, builds. If the person who analyses the process isn't the one who builds it, what was understood gets lost along the way.

When does an in-house team make sense?

When automation stops being a project and becomes a way of working. For example: many processes that change often, in a business where technology is part of the product. At that point, an in-house person or team pays for itself, and the provider supports them on the most specialised parts.

The combined model

For many SMBs it's the most sensible starting point. The provider designs and builds the first processes inside the company's accounts, and an in-house person learns to supervise them: reviewing exceptions, approving what matters and spotting what to automate next. The company gains experience without paying the learning curve from scratch, and decides later whether it needs a team.

That's how Amud works: a free assessment, a fixed price with no lock-in, everything in the client's accounts and the same person from assessment to go-live. We explain it step by step in how we work and in how to implement a digital employee.