Your Own AI Tools Instead of Subscriptions: How to Choose

I wish I could say I have saved SEK 200 000 a year. It looks more as though I am moving the money from subscriptions for marketing and sales tools to AI, data and production.
With AI it is now possible to build tools that only a year or two ago needed a development team of their own. So we have built our own versions of much of what we used to pay for every month. In this post I go through what we have built and what it costs to own the tools. After that you get a list to help you decide which of your own tools are worth building.
Why it became possible now
What has changed is that AI can write the code. Our meeting booking took a day to build. It does what we need and nothing more.
The largest item was the reporting tool. It charged per account and cost USD 699 a month for 26 accounts. Every new client therefore made the tool more expensive. The supplier required a new annual contract when we wanted to gather all of one client's offices in a single dashboard. That was when we built our own. Today there is no fee per account. I write about what belongs in a dashboard like that in the post about a dashboard that management reads.
What we have built ourselves
The list has grown over the year. These are the tools that have replaced subscriptions or that we would otherwise have had to buy.
- Reporting tools and dashboards
- Monitoring and budget control of Google Ads
- Measuring visibility in AI search
- Prospecting and lead lists
- Identifying which companies visit the website
- Scrapers and crawlers that collect information from web pages
- Keyword analysis and rank tracking
- Automation between systems
- Monitoring of servers and services
- Receipt handling
- E-signing of contracts
- Meeting booking
In my experience the easiest tools to build are those that gather data we already have or run a workflow we know well. In those cases we know exactly what the tool should do.

Owning the tools has a price too
The tools need maintenance and that is the price of owning them. When a bought tool stops working it is the supplier's responsibility to fix the fault. When one of our own tools stops working the responsibility is ours.
The biggest risk is faults that nobody notices, much like in the post about the landing page that stopped working. That is why we have built monitors that check dashboards, ad accounts and servers every day and let us know when something looks wrong.
Nor does it pay to build everything. We still pay for AI meeting notes and the cookie consent tool. My assessment is that a bought tool that does the job well at a reasonable price is rarely worth replacing.
Where the money goes instead
The money that used to go to licences now goes to what makes the tools and our client work possible.
- AI models that write code and analyse data
- API calls where the tools fetch answers directly from AI models and data sources
- Data from Google and AI services
- Video and images made with AI
- Servers and databases for our own tools
In my experience these costs largely follow how much we use them. The same AI model or data source can also be used by many tools at once.
How to review your own tools
My recommendation is that you list every tool you pay for and go through them in this order.
- Write down what each tool costs per year and when the contract ends.
- Mark the tools that charge per user, account or client. A price like that grows as you grow and is often the one most worth reviewing.
- Find out how much of each tool is actually used. Ask the people who work in the tool which features they use every week.
- Sort the tools into three groups: keep, build in-house and cancel. Good candidates for building in-house are tools that show data you already have, run a workflow you know well or are only used by a few people.
- Decide who owns each in-house tool and how you will notice when something goes wrong.
- Decide what the money should do instead before you cancel anything. For us it became AI, data and production. For you it may be something else entirely.
In my experience steps two and three pay off the most. That is where the cost can grow without the value growing with it.

The snowball: what we build for ourselves ends up with our clients
The most interesting part has been what happens afterwards. What we build for ourselves we often build for our clients too. And what we build for one client we can then bring to others.
Built for us and now for clients too
- Dashboards that clients log in to
- Google Ads checked automatically every day
- Measuring visibility in AI search by audience and market
- Prospecting and lead lists for sales teams
- Advertising films made with AI
Built for our clients and ready for more
- Cookie-free measurement of enquiries and bookings that Google Analytics misses
- E-learning courses produced through a fixed process from regulations to finished course
- Blog posts and social posts published every week
- Websites that we build from scratch
Everything is tested first in-house or with clients who want to push the limits of what is possible with AI. When it works there we can offer the same thing to more clients without starting from zero. That is how the snowball grows.
You can find examples of what this has achieved for our clients among our cases. All our services are gathered on our services page.
How are you shifting your resources in 2026 and 2027?

Daniel Masajada is the founder of DigitalMassa and has worked with search since 2010.
Want to know where you stand?
The audit shows what you are missing in Google and in AI search right now. Free for 74 more days. Valued at 15 500 SEK.
Read more
How to Improve Your Search Ads in Google AdsHow search advertising in Google Ads works and what to improve first, from keywords and ad copy to Quality Score, landing pages and tracking. With answers to common questions.
Negative keywords: a daily routine that stops costly clicksA short review of your search terms every morning stops clicks that never become business. Here is the routine and how to pick a match type for each negative.