#software-engineering
Every summary, chronological. Filter by category, tag, or source from the rail.
Governing AI Skills: Scaling Agentic Workflows
AI-native organizations must treat 'skills' as first-class, governed assets—similar to microservices—to avoid technical debt, ensure deterministic outcomes, and maintain security at scale.
AI EngineerFormal Verification for AI-Generated Code with Lean4
As AI agents generate code at scale, traditional testing and human review fail to guarantee correctness. Formal verification using Lean4 allows developers to define specifications that machines prove mathematically, ensuring code is correct for every possible input.
Scaling AI Agents Safely: A Roadmap for Engineering Teams
Adopt AI agents by prioritizing verification over prompting, treating skeptic feedback as a safety roadmap, and maintaining human-centric communication standards to avoid 'slop'.
From AI-Assisted to AI-Native: Frontier Development Habits
Productivity gains from AI aren't about the tools, but about shifting from 'vibe coding' (babysitting) to 'frontier development' (feeding agents), which requires intentional changes to team habits and codebase hygiene.
Building uReview: Scaling AI Code Review at Uber
Uber built uReview, a multi-agent code review engine, to solve the bottleneck of increasing PR review times. By focusing on observability, team-specific customizations, and feedback-driven tuning, they achieved a 60% cost reduction and a 67% addressal rate for AI-generated comments.
AgentRoom: Enabling Concurrent Multi-Agent Coding via CRDTs
AgentRoom introduces a shared, CRDT-backed workspace that allows multiple AI agents to collaborate on code concurrently, solving consistency and conflict issues in multi-agent software engineering.
AI Agents: Why the Harness Matters More Than the Model
AI system performance is driven by the 'agentic harness'—the tools, memory, and loops surrounding the model—rather than just the model itself. Distinguishing between the 'brain' (model) and the 'jar' (harness) is essential for building effective AI agents.
IBM TechnologyTerminal Agents: The State of AI in Command-Line Environments
This survey provides a comprehensive overview of AI agents designed to operate within terminal environments, detailing the architectures, evaluation methodologies, and challenges of automating command-line tasks.
Spec-Driven Agentic Development (SDAD) for AI-Native SDLC
SDAD shifts software development from code-centric to specification-centric workflows, using AI agents to enforce rigorous, machine-readable requirements that drive the entire lifecycle from design to deployment.
Moving Beyond Fast Code: Building Context-Aware AI Agents
AI coding agents often create 'fast chaos' by ignoring architectural constraints. To be effective, agents must prioritize repository awareness, explicit planning, and systematic verification over simple code generation.
IBM TechnologyHuman Judgment in the Age of AI Software Factories
As AI agents scale development, human judgment shifts from writing code to defining intent, system design, and verification strategy. A 'software factory'—a repeatable, event-driven loop—is the framework for managing this shift, provided you balance verification budgets with human oversight.
Building Agentic Platforms: The Potter's Workshop Approach
Safia Abdalla argues that AI agent platforms should abstract infrastructure complexity, provide consistent multi-harness support, and act as 'potter's workshops'—structured, observable systems that empower humans to ship software rather than just automating code production.
AI EngineerBuilding an Agent Kernel: Why Frameworks Fall Short
Instead of using complex agent frameworks, build a simple 'kernel' that treats agents as isolated processes. Use content-addressed prompts, event-driven architecture, and strict type boundaries to ensure reliability and auditability.
Give AI Agents a Budget, Not a Token
Stop giving AI agents 'god tokens' with unbounded power. Instead, treat them like junior engineers by enforcing budgets through asymmetric verbs, rate limits, trip wires, and the 'undo test' to bound their blast radius.
Mitigating Silent AI Tool Failures with Outcome Monitors
Silent tool failures—where an AI tool returns a technically valid but semantically incorrect result—are a major bottleneck. Outcome Monitors provide a framework for detecting these failures and enabling automated recovery.
Human Judgment in the Age of AI Software Factories
AI agents accelerate code generation, but they don't replace the need for human taste. A 'software factory'—a repeatable, event-driven loop—is the best way to encode engineering culture and quality gates while focusing human attention on high-risk decisions.
Scaling Agentic SDLC at Uber
Uber has shifted 70% of pull requests to AI agents by building a standardized infrastructure layer that manages model security, context retrieval, and automated validation, effectively moving the engineering bottleneck from 'how to build' to 'what to build'.
Adversarial Review: Improving Agentic Code Quality via Disagreement
Adversarial Review improves agentic code quality by forcing AI agents to engage in structured disagreement, moving beyond simple consensus to uncover hidden bugs and architectural flaws.
Prioritizing Concurrency Control in Multi-Agent Systems
Multi-agent systems must move beyond simple orchestration to prioritize robust concurrency control, ensuring state consistency and conflict resolution as agent complexity scales.
Modernizing Legacy Codebases with AI Agents
Tackle legacy code by treating AI as a coworker: use a three-step 'plan, execute, verify' workflow, prioritize test-driven development, and enforce strict guardrails to prevent hallucinations and errors.
Google Cloud TechMoving from AI Code Generation to Artificial Wisdom
To automate code review, teams must shift from line-by-line human inspection to codifying tribal knowledge and architectural constraints into a machine-readable context engine.
Building an AI-Native Health Company: Lessons from Maven Clinic
To become AI-native, shift from long-term planning to 2-4 week sprints, replace delegation with AI-assisted individual execution, and implement tiered reliability standards for non-deterministic AI outputs.
AI EngineerAccelerating Legacy Migrations with AI Agents
Asana replaced an outdated testing framework in two weeks using AI agents, reducing a projected five-year, $6M manual effort to a $12K infrastructure cost.
Warp Factories: Infrastructure for AI Software Development
Warp Factories provides an out-of-the-box infrastructure layer for building and managing AI agent-based software development pipelines, automating tasks across the full lifecycle from triage to verification.
Applying RAD Methodology to AI-Driven Development
Rapid Application Development (RAD) provides a proven framework for AI coding: plan lightly, prototype iteratively, and use spec-driven development to bridge the gap between AI-generated prototypes and production-ready software.
IBM TechnologyKog Optimizes GPU Inference Through Low-Level Software Engineering
French startup Kog is challenging the notion that GPUs are poorly suited for agentic AI workloads by using low-level assembly and binary-level optimization to unlock massive inference speed gains on existing datacenter hardware.
Fixing Computer Use Benchmarks: Beyond Replay Exploits
Current computer use benchmarks are often gamed by 'replay agents' that blindly repeat successful trajectories. Robust evaluation requires stochastic, verified environments and honest statistical uncertainty to avoid costly deployment errors.
Raising the Floor: Practical AI Agent Evaluation
Stop chasing benchmark scores and start treating agent evaluations like production software tests. Focus on identifying when issues start and their impact on user volume to build reliable, trust-based AI products.
AI EngineerMetaSpace: Metamorphic Testing for Embodied AI Spatial Cognition
MetaSpace introduces a metamorphic testing framework to evaluate spatial reasoning in embodied agents by applying geometric transformations to environments and verifying if agent behavior remains consistent.
Scaling AI Agent Adoption Across Engineering Teams
Moving from individual AI leverage to team-wide productivity requires treating agent integration as a leadership-driven infrastructure challenge rather than an individual task, focusing on harness engineering, self-healing systems, and psychological buy-in.
AI EngineerShowing 30 of 222