Replacing SaaS with AI: when building your own pays off – the Calendly example

Many companies pay for SaaS tools of which they use only a small part – and they pay per head. Every new hire raises the bill while the features used stay the same.

Coding agents such as Claude Code or OpenCode have significantly cut the effort for clearly specified business applications. That shifts the classic make-or-buy question: building a small, well-scoped application yourself and running it in your own Azure environment is now often a matter of days rather than months.

What matters is not the coding but three things around it: a clean specification, mandatory quality and security checks, and clearly named ownership for operations.

Not every tool is a good candidate. The table helps with a first assessment.

Calendly charges per host; people booking a meeting need no seat. List prices on annual billing (as of November 2026): Standard 10 US dollars, Teams 16 US dollars and Teams with the AI add-on 24 US dollars per seat per month. Enterprise starts at 15,000 US dollars a year from 50 seats – only there are SAML SSO, audit log and a data deletion API included; on Teams, SSO is an optional add-on.

For the in-house build we define the scope a typical sales or consulting team actually uses:

We assume a day rate of €1,100 and development with coding agents under mandatory review. The application runs in an existing Azure landing zone. For simplicity we set 1 US dollar = 1 euro.

Over three years the in-house build costs €12,100 + 3 × €3,780 = €23,440, almost regardless of the number of users. SaaS costs, by contrast, grow linearly with every seat.

A booking tool looks harmless, but it has access to every host's calendar and collects customers' names, email addresses and meeting topics. With an in-house build, both stay in your environment.

The second benefit is freedom: you don't wait for a vendor's product roadmap and you don't pay a tier jump because a feature only exists in the next package.

Examples that take only a few person-days in an in-house build: store bookings directly on the CRM record, pre-qualify requests with your own rules, book rooms or demo devices alongside, add internal approval steps, summarise the request before the meeting with your own language model – via LLMrouter.eu with models from Europe.

How to assess and replace a SaaS tool without putting day-to-day operations at risk.

MAKONIS builds the SaaS inventory and make-or-buy calculation with you, develops the replacement with AI software engineering and, if you wish, keeps running it in your Azure landing zone.

For companies that want to build many such internal applications there is Scalor.io: the self-hosted platform where business teams and developers build apps together with AI agents – in isolated workspaces, in your Git repositories and through your approval processes all the way into your own cloud. Start with a free 30-minute call with our engineers.