WHEN SPEED BREAKS SOFTWARE - FAST TEAMS, FRAGILE SYSTEMS

Navigate the AI 'Velocity Paradox': balance rapid feature development with system complexity management

Dec 16, 2025 5m 29s

Zbigniew Sobiecki

Expert Insights

While the temptation is to fully exploit the unmatched speed and concurrency capabilities of AI, Zbigniew Sobiecki calls for a tempered approach. He warns of an overlooked "velocity paradox": while features are being built faster than ever, the resulting system complexity can stall progress and necessitate new, elaborate context-management solutions.

Zbigniew dares software teams to trade acceleration for agility: preserving a human grasp on system complexity and promoting software determinism to counter the unpredictable nature of AI toolsets.

Hear Zbigniew explain:

Why the rapid-fire pace of creating new features with AI can lead to system complexity that erodes progress speed.
The importance of structured communication and context management to cater to the AI-created, inherently fragmented software system.
The hidden cost of AI acceleration: managing multiple parallel conversations due to individual models only possessing partial context.
Why is keeping the software deterministic important even though it is created with non-deterministic AI, enabling comprehensive testing and documentation?
The unforeseen challenges and costs of working with AI that teams need to budget for, including managing multi-agent systems and cross-model communication.

The traditional software team operated sequentially. Plan, build, test, deploy. (...) I thought that we would need more structured communication with the user earlier on.

Monterail Team Analysis

Here’s how teams can manage the transition:

Prepare for the Velocity Paradox: Acknowledge that while the pace of feature building might accelerate with AI, the resulting system complexity can slow down overall progress. Forecast and manage these trade-offs to maintain a balanced development pace.
Develop Context Management Strategies: As software systems become more complex due to AI fragmentation, implementing effective context management solutions becomes vital. Promote clear communication and context sharing among team members to maintain grasp on the system's intricacies.
Preserve Software Determinism: Amid the unpredictable nature of AI, maintain an element of human control. Keep manual test suites extensive and regular to ensure code meets expectations, regardless of how it is written.
Manage Multi-agent Systems: Oversee multiple parallel conversations due to different models possessing only partial context. Have a strategy in place to handle context administration in this AI-powered, accelerated environment.
Budget for Hidden AI Costs: Be mindful of the unforeseen challenges and costs that may arise when using AI. These could include managing multi-agent systems, implementing cross-model communication, and creating automated, real-time document updates.
Facilitate Cross-model Communication: Make resources available to cater to the unique needs of AI-assisted development. This might involve maintaining an open line of communication across different models to allow for smooth functioning and context sharing, amongst other tasks.
Adapt to New Challenges: Be ready to navigate and solve new issues that didn’t exist before, like managing multiple agents and ensuring context consistency across different models, a reality of AI-accelerated development.