What do we want to achieve?
Infodations AI strategy focuses on the improvement of our primary and secondary processes. Meaning, we started targeting software development and internal operations for AI usage. AI features and products for customers come as a tertiary priority. These priorities emerged from the Infodation business case: We make bespoke software for our clients.
We don’t sell SaaS software, and we did not want to start doing so with an AI tool. We do however want to reap the efficiency benefits from AI, and we do want to stay current with development trends and knowledge. Ergo, we focus mainly on the question: How can AI help us to code and work better.
What we do:
Infodation biggest assets are their employees, and AI is a tool that can touch nearly all areas of our business. Therefore, we want to provide the chance for everyone to learn and improve their workflows. The main structural pillar we set up is “Experiments.” These experiments are timeboxed, small-scale research opportunities, that are supported by Infodation. The motto is “Keep scope small, learn fast, don’t aim for production-grade outcomes”. As an organisation, we want to quickly cycle though opportunities, and hold on to the ones that prove interesting. Once initial experiments have been finished, we can review and decide if we want to move further with the technology.
How do we set up our experiments
AI experiments are grass-root drive, meaning that the idea for each experiment is born out of the interest of employees. Setting up for an experiment is purposely simple.
If someone wants to start an AI experiment, they will create a ticket on our AI board. This is a simple Jira board, to which the whole organization has access to. They describe the experiments and add learning goals to the ticket. As the experiment moves on,experiment goals and a learning summary will follow.
In the ticket, we also check for potential required licenses and compliance issues.
Infodation does have a dedicated AI crew. The AI crew is supporting others in running experiments, not taking them over. This ensures that AI knowledge does not only pool around a silo of people, but is distributed further into the population of employees.
“Keep the scope small, learn fast, and only use what works in practice.”
Please find a selection of our experiments here.
What are we seeing, what are we learning?
The response to the experiment set-up is positive. At the time of writing, we had over 110 registered experiments on our board.
The past experiments can be roughly divided into roughly four groups:- AI for coding
- Facilitate Testing
- Improve Communication
- AI for customers
As to be expected with an experimental approach, results vary from very successful to failure. Some go on to become features that are being used in our daily processes, some were stopped after a few days into the experiment. A satisfying trend is the number of new experiments spawning from old, finished experiments, meaning that our secondary goal of distributing knowledge about AI within the company is being achieved.
Whats next?
As our experiments-cycle reaches maturity, more AI features make it into our software. For example, our in-house testing tool ITMS (Also available on Atlassian market place) now has an integration with the AI-tool Cypress. While our understanding of AI has grown, this growth towards maturity means we will shift our focus to work on AI guided viewpoints of topics such as data protection, ethical compliance, the prevention of exploitation amongst many.
AI has already greatly impacted the Infodation workflow. We are all using it-Designers, QCs, developers, the marketing department, the founders. For now Infodation’s initial vision stays strong: AI is a tool that can support us in our work. Just like any other tool, we are currently in the process of understanding how we can use it safely and effectively. And this is the line we will continue to hold in the future.
