AI in Product Development: Case Bittium
Artificial Intelligence is becoming a central tool in how we build products, but especially in regulated industries it cannot be adopted blindly. At the latest SW4E “AI as Spearhead” working group, Pertti Kontio from Bittium shared how his team approaches AI integration: with a clear strategy, strong governance, and concrete results from pilots.
From Connectivity to AI in Development
Bittium has decades of experience in wireless communications. Its work spans secure defense systems, medical monitoring, and advanced engineering solutions around 5G, 6G and hybrid networking. Across all of these, security and reliability are non-negotiable.
The same thinking now guides Bittium’s AI efforts. As Kontio emphasized, in this presentation we discuss using AI tools to improve product development value stream, and not about embedding AI into products.
Why a Strategy Matters
During the first pilots it became clear that Bittium needed a company-wide strategy for AI in product development. Without it, efforts risk becoming scattered, uncoordinated, or even risky. Kontio highlighted several reasons:
- AI use must be aligned with business goals to get the best return on limited resources.
- Management commitment is necessary for the significant investments in infrastructure, skills and governance.
- Risks such as copyright issues, compliance failures and data leaks must be actively managed.
- Goals and measurements are required to know if progress is real.
In his words, the point is simple: don’t waste resources on isolated local optimization. Use AI to holistically enable a leap in productivity, quality and developer satisfaction.
Pilot Results: Savings You Can Measure
The first AI pilot involved ten developers working with a self-hosted AI server. The team applied AI to 25 different problem-solving cases. Results were encouraging:
- Average success rate score: 3.8 out of 5.
- Effort without AI: 194 hours.
- Effort with AI: 25 hours.
One person-month of work saved in just two months.
Tackling the Risks
Kontio addressed the risks directly. The team identified three major ones and built clear mitigation approaches:
- Copyright infringement – addressed with documented AI processes, training, coding rules, peer review and automated license checks.
- Data leaks – risk reduced by using trustworthy providers, keeping sensitive workloads in-house, and strict server configurations.
- Compliance requirements – managed through continuous monitoring of standards, careful documentation, and integrated compliance checks.
These measures bring the probability of serious issues down to “very unlikely” or “extremely unlikely.”
Building the Infrastructure
Three models for AI infrastructure are under consideration:
- Cloud-based services for fast access to cutting-edge models.
- In-house AI platforms for sensitive and highly regulated work.
- Hybrid setups combining both.
Developer rules are equally clear: all AI-generated code must be marked, reviewed, and automatically checked reduce copyright risk. Modifications must take place where risky generated code appears.
Looking Ahead
For Bittium, the aim is not to replace developers but to empower them. AI is there to cut routine work, boost productivity and increase satisfaction, while keeping compliance and security intact. With strategy and discipline, AI can be the spearhead that sharpens product development for the years to come.
About the author
Pertti Kontio, Process Development Manager at Bittium. Passionate about applying AI in ways that improve developer productivity, satisfaction and product quality, while ensuring security and compliance in demanding environments.