skáld

We built skáld because we realised that most of our clients could not get realtime answers regarding the data in the apps that we had built for them.

Standard reporting dashboards are great, and we build those into the apps we build, but often we'd still get support requests from clients asking for answers from their data that did not exist in the analytics and metrics dashboards.

This would usually result in someone in our team writing and running custom SQL queries to get the answer to send to the client. Or in a billable work request to build a new dashboard view.

Then we started thinking:

What if our clients could just query the database themselves?

At first this seemed like a ridiculous question to ask because we know that most users can't write SQL queries and have no idea about database structure and relationships.

So we started investigating and doing proof-of-concept (POC) work on getting AI to write the queries for the client.

Once we managed to connect the LLM (via a secure, read-only connection) to a database and give it context of the database structure, we knew we were onto something!

Lots of testing, and tweaking later, we had a working POC that was giving us reliable, replicable results.

That's when we went all in and built skáld into a full product that can be embedded into a client's application, allowing them to chat with it and get super valuable data from their own system.

But what really changed things was not just the ability to ask questions, it was the shift in how quickly decisions could be made.

Instead of waiting hours or days for answers, teams could validate assumptions immediately. Operations teams could investigate issues while they were happening. Managers can now explore trends without needing to ask for new reports.

It also removed a major bottleneck on Lonely Viking's side. Our team no longer needs to sit between the client and their data for every question that falls outside of a predefined dashboard. That time can instead be spent building better systems, improving existing features and focusing on higher value work.

skáld does not replace dashboards. It complements them.

Dashboards are still the best way to monitor known metrics. skáld exists for everything else.

The unknown questions. The edge cases. The moments where someone needs to dig a little deeper.

And because it is embedded directly into the applications we build, the context is always relevant. Users are not switching tools or exporting data. They are asking questions exactly where the data lives.

That is where skáld becomes powerful.

It turns any application into something that can be interrogated, explored and understood without friction and without coding chops.