Decades of software management taught us one rule: developers will always have more work than time. AI velocity just broke that rule. Teams are now moving so fast they’re outpacing their pipelines - a…
資訊系統設計
無論是哪種資訊系統、不管是 architecture 還是 infrastructure 都歡迎討論
Follow from your instance
@information-system-design@baudrate.tw
Search for this handle on your own server to follow from there.
In the InfoQ AI Security & Privacy Engineering Program, you can bring a current AI security or privacy problem to map, threat model, red team, and test. Review the five-week program: https://certifica…
Five practitioners who build, secure, and run AI systems are joining a free live InfoQ panel on Wednesday, October 14 at 11:00 AM ET: Hien Luu, Katharine Jarmul, Zichuan Xiong, Premanand Chandrasekara…
In this article, Aditya Mulik explains how treating the LLM stack as platform infrastructure reduced hallucinations from 15% to 1.5% in a production inventory recommendation system. Key platform primi…
Akka tested a spec-driven AI porting workflow across 65 open-source projects: ➡️ 9.41B tokens used over 99.3 hours ➡️ 57 ports delivered fewer lines of code or better performance. See how the workflow w…
✈️ Airplanes can fly themselves, yet we still need pilots. Why? Because automation generates output. Humans decide whether it’s safe. #Spotify found the same problem with AI coding tools: generating co…
AI can accelerate software development - but it can also slow skill development. Engineers still need to understand how systems work. Sometimes, learning means slowing down. Senior developers can help…
#AWS open-sourced Pizza Bot - a self-hosted app for running AI agents in the background. Agents can schedule tasks, delegate work, respond to webhooks, and pause for human approval. 🔗 Details here: ht…
What happens to observability, incident management, and SRE when agents write & run code in production? This #InfoQ panel explores why traditional observability falls short - and what production engin…
What verifies an agent’s code before a human sees it or the agent runs unattended in CI? Review the AI-Assisted Engineering syllabus: https://certification.qconferences.com/ai-assisted-engineering?utm…
A Schema Registry isn’t just storage. It’s your event-driven architecture’s contract enforcer. Configured for compatibility, it blocks breaking schema changes before they ever hit production. ➡️ No bro…
Regeneration is changing the economics of UI reuse. Why maintain a massive canonical component library when standard UI can be regenerated on demand? The real shift: • Keep the design system, tokens, …
Uber rebuilt the #UberEats search pipeline and cut end-to-end latency by 50%! How did they do it❓ • Above-the-Fold rendering metrics • Less retrieval work • Parallel hydration & ad path redesign • AI …
Architecture decisions can take years to prove right or wrong. The next InfoQ Certified Architect Program starts October 9. Review the syllabus: https://certification.qconferences.com/architecture?utm…
#Atlassian migrated its metrics platform from gostatsd to OpenTelemetry - without disrupting production alerts. ⚡ ~100,000 hosts ⚡ 14 regions ⚡ 99.95% SLO 🔗 Learn more: https://www.infoq.com/news/2026…
Without tests, you’re just guessing. Downloading agent skills from public registries without testing them in your own environment? A roll of the dice. Registries don’t keep your codebase safe. Testing…
A routine TLS 1.3 upgrade silently broke Route 53 health checks. The result? A CDN stopped routing traffic to a healthy region - while internal dashboards showed everything was fine. In this #InfoQ ar…
The painful truth about #MultiCloud: You’re paying double the storage and egress fees just to be capped at the lowest common denominator of features. Is the illusion of portability worth the hidden co…
Grafana Labs published a practical way to monitor Cypress test suites by converting test results into Prometheus metrics for Grafana Cloud - turning transient pass/fails into long-term observability. …
Most AI teams aren’t engineering. They’re shipping on vibes. Change a prompt. Run one test. It feels better. Ship it. That’s not Product Development - that’s pure randomness! Without an automated eval…