Software Insights · Newsletter
No opinion without evidence. No finding without a source.
Peer-reviewed software engineering research, translated into clear, actionable guidance for engineering leaders.
Subscribe on Steady →Weekly · Free
Archive
When the agent speaks first, whose judgment is it?AI in SE
What makes a coding agent worth keeping?AI in SE
The agents met in the package cache.AI in SE
The software we now have to ownAI in SE
What generative AI does to the engineers it does not replaceDeveloper Productivity
When oversight degrades the overseerAI in SE
The hires that did not happenAI in SE
When every agent is correct, and the system is notAI in SE
When engineers can direct AI, and when they cannotAI in SE
The capability you cannot fork: software’s break with universal accessAI in SE
Europe is about to make human-first software non-optionalAI in SE
Ericsson’s AI-native software programme: principles, deployment, and open questionsAI in SE
Disarm AI, Re-Arm Teams: Two Levers Against a Lost CohortAI in SE
Three years of AI: what the evidence settles, contests, and leaves unmeasuredAI in SE
A Bad Benchmark Does More Damage Than No Benchmark at AllResearch Methods
Cleaner Code, Foggier Teams: The Debt AI Moves UpstreamAI in SE
The Composition Trap: Why Agent-Dense Teams Break in Ways Human Teams Never DidAI in SE
Your Agents Pass Every Test. Your System Can Still Fail.AI in SE
Cutting the Junior Pipeline Before the Automation ArrivesAI in SE
AI Can Solve Your Benchmark. Can It Fix Your Codebase?AI in SE
56% of Organizations Get Nothing from AI Because Compatibility Was an AfterthoughtAI in SE
The New Engineering Judgment: Five Skills That Survive the Agentic TransitionDeveloper Productivity
When Code Becomes the Output: Rethinking What We Teach (and Hire For)AI in SE
Speed at the Cost of Quality: The Hidden Tax of Agentic AIAI in SE
The “Country of Geniuses” Requires a Constitution, Not Just a LoginAI in SE
Davos 2026: From “Chat” to “Work”Agile & Process
Europe’s Tech Renaissance Beyond the “Kill Switch” NarrativeAI in SE
Technological Sovereignty in a New Era: A Compass for Europe’s Digital FutureAI in SE
When the agent speaks first, whose judgment is it?AI in SEWhat makes a coding agent worth keeping?AI in SEThe agents met in the package cache.AI in SEThe software we now have to ownAI in SEWhat generative AI does to the engineers it does not replaceDeveloper ProductivityWhen oversight degrades the overseerAI in SEThe hires that did not happenAI in SEWhen every agent is correct, and the system is notAI in SEWhen engineers can direct AI, and when they cannotAI in SEThe capability you cannot fork: software’s break with universal accessAI in SEEurope is about to make human-first software non-optionalAI in SEEricsson’s AI-native software programme: principles, deployment, and open questionsAI in SEDisarm AI, Re-Arm Teams: Two Levers Against a Lost CohortAI in SEThree years of AI: what the evidence settles, contests, and leaves unmeasuredAI in SEA Bad Benchmark Does More Damage Than No Benchmark at AllResearch MethodsCleaner Code, Foggier Teams: The Debt AI Moves UpstreamAI in SEThe Composition Trap: Why Agent-Dense Teams Break in Ways Human Teams Never DidAI in SEYour Agents Pass Every Test. Your System Can Still Fail.AI in SECutting the Junior Pipeline Before the Automation ArrivesAI in SEAI Can Solve Your Benchmark. Can It Fix Your Codebase?AI in SE56% of Organizations Get Nothing from AI Because Compatibility Was an AfterthoughtAI in SEThe New Engineering Judgment: Five Skills That Survive the Agentic TransitionDeveloper ProductivityWhen Code Becomes the Output: Rethinking What We Teach (and Hire For)AI in SESpeed at the Cost of Quality: The Hidden Tax of Agentic AIAI in SEThe “Country of Geniuses” Requires a Constitution, Not Just a LoginAI in SEDavos 2026: From “Chat” to “Work”Agile & ProcessEurope’s Tech Renaissance Beyond the “Kill Switch” NarrativeAI in SETechnological Sovereignty in a New Era: A Compass for Europe’s Digital FutureAI in SEThe newsletter covers the evidence. When a question needs that evidence applied inside a specific organisation, that is what the advisory work is for.
by Daniel Russo, Ph.D.