Codacy

*blushes bullishly* James, may your 2026 predictions come true. We're ready 🫡 ☑️ Supply Chain Malware Detection ☑️ AI Vulnerability Remediation and Prioritization ☑️ AI Visibility, Guardrails, and Testing ☑️ AI-Based Detections

This year, our customers showed what responsible software delivery looks like in an AI-accelerated world. As velocity increased, they doubled down on application security, coding standards and tech debt guardrails. Codacy’s automation and intelligence helped carry the load, which we are immensely proud of. One last stat to wrap up the year? On average, we helped our customers fix 1 critical security vulnerability for every ~14 pull requests. 2026, here we come.

The software engineering industry spent a decade professionalizing itself. Almost overnight, AI coding sent us back to the Wild West. That was the uncomfortable but honest framing from Codacy's VP of Tech Kendrick Curtis in our latest product showcase. The question now is: how do we bring order back? To help, we released 2 new capabilities: the AI Risk Hub and the AI Reviewer. But the showcase was more than just a product launch. We also brought together two experts from opposite ends of the development lifecycle to discuss the big picture, Aruneesh Salhotra, CISO and lead author of OWASP AI BOM and 🌀 Luca Rossi, founder of Refactoring. Here are the key takeaways from the discussion: → AI amplifies existing behavior. Disciplined teams raise quality. Undisciplined teams accumulate technical debt at scale. → Technical debt is the real time bomb. If AI lets you write 3x more code, you also ship 3x more vulnerabilities unless guardrails operate at AI speed. → Accountability cannot be delegated to AI. Even if AI writes and reviews the code, a human still owns the outcome. → AI governance is not optional. With 100+ AI regulations emerging globally and the EU AI Act already live, compliance is urgent. → Prompt injection is still the #1 threat. And agentic AI will only expand the attack surface. If you missed it, here's the full breakdown and video on-demand. 👇 https://lnkd.in/eKCMekW9

Context-as-a-Service? Teams using Augment keep coming back with the same request: “Your context engine is the part we rely on most. Can we get it as an API?” Teams want Augment’s context engine as an API, and Vinay Perneti confirms they’re building it. It’ll be a way for developers to push and retrieve context without rebuilding the entire layer themselves. Watch the entire session on-demand: https://lnkd.in/dEdYbTue

“It's not change management anymore; it's change fitness.” That’s the line that lands hardest in this AI Giants clip with Augment VP of Engineering, Vinay Perneti.   Model capabilities are jumping every 3–4 months. Each leap forces teams to absorb new workflows, new building blocks, and new expectations. The challenge for eng leaders now is how to build teams that lean into the shifts instead of getting blindsided by them. Watch the entire session on-demand https://lnkd.in/dEdYbTue

Our latest AI Giants episode took a hard look at something most dev teams don’t want to admit: a lot of “AI productivity” is just a feeling. In this session we sat down with Dax Raad, creator of OpenCode, to challenge some of the most common assumptions about coding agents, benchmarks, and multi-agent workflows. Key takeaways from the conversation: → Feeling productive isn’t the same as shipping good products → One fast model, used well, beats juggling multiple agents in parallel → Clean, well-structured codebases matter more than which model you pick → Benchmarks push teams to optimize for the wrong things → Developers still need to stay in the loop → What you ship matters more than how ellaborate your setup looks If you’re using AI coding tools in real teams, this episode is a necessary reset. 👉 Watch the full session on-demand: https://lnkd.in/eYAKjVGx

When developers avoid an AI tool, they’re telling you something important. Vinay Perneti argues that value speaks louder than mandates. When developers ignore a tool, it’s usually because it’s not solving their problem. His view: go bottom-up, get curious about the blockers, and be willing to switch tools if the one you chose isn’t delivering. Watch the entire session on-demand: https://lnkd.in/dEdYbTue

Most coding assistants treat context more like a side note. For Augment, it’s the core. VP of Engineering Vinay Perneti breaks down Augment’s context engine: how it ingests the entire codebase, builds embeddings, and keeps context fresh within milliseconds. Watch the entire session on-demand: https://lnkd.in/dEdYbTue

Don't feed the worm 🪱 Codacy now detects over 25,000 malicious packages like Shai-Hulud 2.0. Here's how it works: https://lnkd.in/gF4Azr6X

AI code reviews are starting to catch issues that normally require a senior engineer’s full attention. Vinay Perneti talks about how this shifts code review toward engineering judgment, leaving humans to focus on direction, intent, and how the system is evolving. Watch the entire session on-demand: https://lnkd.in/dEdYbTue