Ray Sono successfully carried out a code sprint for a client again. Congratulations! What exactly is a Code Sprint and what are the advantages of this process?
Sebastian: A code sprint is a methodology used for projects with many unknown and often complex technological conditions. The aim of a code sprint is to demonstrate a project's feasibility, develop the prototype of a core technology, and finally, to define a basis for making the right decision. And without a huge investment!
In this specific case, Bosch gave us two days to find a technical basis that could enable expanding “Frizz” into a voice-supported chatbot. Our second challenge was to determine the optimal flow structure and logic required to generate the dynamics in Cognigy.AI (chatbot technology) and ensure there wasn’t any unnecessary repetition of previous content. Ultimately, our aim was to deliver initial tangible results: a technical prototype.
The Code Sprint was a good opportunity to test the chatbot’s feasibility without first investing a lot of time and resources.
A big challenge ...
Nataliia: That’s why Rays from many different areas were involved in the two-day hackathon.
Which brings us to skills. What skills does such a project require?
Sebastian: To deal with this kind of complex project, you need backend, frontend, as well as UX specialists – in other words, a mixture of programmers, developers, and web designers. A successful and effective code sprint requires competence and teamwork.
Ok, the team is in the starting blocks. What happens then? Can you briefly outline the process? And did the pandemic-related circumstances influence the teamwork?
Nataliia: A hackathon always starts with a briefing from the product owner. The whole team learns about framework conditions like dependencies, exclusions, assumptions, and questions. This is followed by a smaller round in which the technical requirements are discussed and the general direction of the code sprint is defined. Together with UX specialists and our service desk, the environment is then technically prepared. That's when the sprint starts, and from then on, the teams continue to work autonomously.
Sebastian: The project ran remotely, but we're all used to that now – we work as well together virtually as in person! Naturally, of course, it's much more fun when you meet your teammates in “reality”.
Regardless of the difficult situation – were there any situations in which you felt you couldn't move forward?
Nataliia: Pain points are part of it. Of course, you may get stuck during an implementation test or even have to take a step back, i.e. try another approach or change your perspective. That's what a code sprint is all about: reflecting, testing, and adapting. In short: Inspect & Adapt – a principle for permanent improvement. In the end, it's the result that counts, and in this specific case, we even went beyond the task.
So you went a stage further.
Nataliia: Exactly. Bosch had initially expected that we would only work on speech recognition questions in the code sprint, but we ended up also dealing with “dynamic dialogue flows” as well.
What does that mean in concrete terms?
Nataliia: That the chatbot can react to the user's previous actions and does not play the same content back to them again. We built this dynamic into the chatbot system during the Code Sprint.
How do you concentrate for such a long period of time? Are there any special tricks? Were you able to switch off in between?
Nataliia: Coffee always helps! Of course, regular breaks are important, and we also held breakout sessions – where team members could discuss interim results or reflect among themselves. Basically, a code sprint like this is very popular at Ray Sono because you can work as freely, creatively, and open-endedly as you want – the necessary motivation and concentration arise on their own. Such projects are where our company DNA particularly shines.
What results did you communicate to Bosch?
Sebastian: That all requirements can be implemented. We defined the technical setup, identified possible stumbling blocks and open points, and pointed out possible solutions. “Frizz” will be able to talk to us in the future – which means that voice integration works.
What personal conclusions do you draw from the project? What kind of feedback did you get from Bosch?
Sebastian: Overall, we are very satisfied with the result. In addition, this project has once again shown what a valuable tool a Code Sprint is. The cherry on top was the very positive feedback and appreciation from Bosch.
We have worked with Ray Sono at a very high level for some time. I trust the expertise of the team.Kevin Ulmer, Project Manager, Bosch
Finally, a general question about AI: What everyday tasks will AI handle for us in five years?
Nataliia: If we can combine current AI applications with specialised knowledge to form generalist, linked AI systems, we will be able to develop many new possibilities – from new medicines to individualised learning setups in the educational system. AI will support us in our everyday lives much more than can now.
Sebastian: I think so too. However, I personally believe we will always need human competence and supervision in the future if we want efficient, high-quality results.