Cortex

Cortex KPI cards: give the metric that's on fire the room it needs, without losing sight of the ones that aren't. For this Feature Friday, Principal Product Manager Christine Byun walks through KPI cards in Engineering Intelligence, turning a dashboard of same-size charts into one you can actually prioritize. What it does: -Pulls metrics like change failure rate and rollback frequency into compact cards, so they stay visible without eating up chart space -Frees full chart space for the metric you're actively investigating (incidents, in this demo), so it gets the depth it needs -Lets you customize each card's header, text size, sparkline preview, and trend display (relative, absolute, or none), so the card shows exactly what you need and nothing else -Runs natively in Data Explorer, so building a KPI card takes the same steps as building any other chart, no separate tool to learn Watch the full walkthrough below.

Agent adoption is easy to count, but maturity isn't. Most engineering leaders can point to how many coding agents are running in their stack without being able to say how far that's actually moved them toward a self-improving, autonomous SDLC. Anish Dhar (Co-Founder & CEO) and Ganesh Datta (Co-Founder & CTO) built a maturity curve to answer that question, and at EVOLVE they'll map where the market really stands today, then walk through how to place, and advance, your own org on that curve using DRIVE and the OpEx review. They're also opening up the AI software factory Cortex runs on itself, as a first look at what the far end of that curve looks like in practice. EVOLVE is Cortex's conference for engineering leaders building their AI software factory, happening September 24, 2026 at Convene 237 Park in NYC. Apply to attend: evolve.cortex.io

The biggest disparity I’m seeing right now with AI is the gap that exists between enterprise adoption and the rest of the broader market narrative. There is absolutely no doubt that engineers at very large enterprises have been exposed to coding agents or experimented with them in their personal lives. But the gap that exists between these larger enterprises truly adopting agents in their SDLC and also seeing real productivity gains from those investments is way larger than most people realize. This morning I had a conversation with a CIO of a large public enterprise with 10k+ engineers where they’ve only just now kicked off a POC with a few different coding agents. Another one of our larger customers actually rolled out Claude Code to their engineers and pulled it back after seeing a huge spike in incidents. The reality is that larger enterprises have enormous foundational requirements that they have to first get a handle on before exposing more of their SDLC to AI. Guardrails have become a must have and not an afterthought in a Confluence doc or spreadsheet. And coding agents are still part 1 of a longer transformation in building the software factory that I suspect will take a few years to truly materialize. This transformation that is happening in the enterprise is not unlike the on-prem to cloud migration that has taken better part of a decade and still happening across the industry! 

You can measure your teams across all five SPACE dimensions and still not know whether your organization is turning their work into reliable software. SPACE, from Nicole Forsgren and Margaret-Anne Storey, changed how the industry measures developer productivity. DRIVE measures something SPACE was not built to measure: whether the organization as a whole sustainably turns customer needs into reliable software while balancing speed, quality, and cost. Same mission, different altitude. SPACE looks at the developers and teams doing the work. DRIVE looks at the system around them, humans and agents, and whether it can absorb and operate what they produce. As AI amplifies output and agents take on more of the SDLC, that organizational altitude is where the risk now concentrates. The two work together. Once you know your teams are healthy, DRIVE tells you whether the organization can keep its promises to customers as all that output scales. Cortex CTO Ganesh Datta lays out DRIVE, its five pillars, and the operational review that turns them into action, here: https://lnkd.in/eFJP_Cmc #EngineeringLeadership #PlatformEngineering #DeveloperProductivity #AI

Last year, we handed out our first-ever Engineering Excellence Awards on stage at IDPCON. Past winners include engineering leaders from H&R Block, Shell, Xero and more. This September, we do it again. EVOLVE 2026 lands September 24 in NYC, and a new class of Engineering Excellence Award winners will be named live on stage. Who makes the list this year is still TBD but the bar last year's honorees set isn't going anywhere. Want to find out who takes it home and be in the room when they do? Apply to attend EVOLVE 2026: https://lnkd.in/edRz-Nrf Congrats again to the leaders who set that bar: Trushar Shah, Nick Raccioppi, Naresh Kumar Bulusu, James S., Simon Irwin, Hariprasad Babu, Fred Mare, Dan Willman, and Leslie Brown.

To gauge how factory-like your SDLC is, work backward from the fully autonomous end state you are building toward. That is the method behind DRIVE's forward-looking metrics, mostly in delivery and reliability: → Diffs per R&D team member (team level, never individual) → % of PRs merged with no human review → Change failure rate or revert rate as the factory matures → Whether your defect backlog shrinks as background agents attack it Some of these are gameable, diffs per member especially. That's why they belong in a recurring operational review where humans interrogate the data, not on a dashboard nobody questions. Cortex CTO Ganesh Datta lays out the full metric set, and the review mechanism that makes it work, in the DRIVE framework. Grab it here: https://lnkd.in/eFJP_Cmc #EngineeringLeadership #PlatformEngineering #AI #SoftwareDevelopment

Our AI code reviewer isn't allowed to approve a single PR. Every pull request at Cortex gets a full review from an AI agent before a human opens it. It comments as the Cortex review bot and can request changes all day. Approving stays with a person, always. The bot runs as an orchestrator that fans out to specialist reviewers: security, performance, query engine, each defined as a markdown file in the repo. When a repo has a sharp edge, we write a specialist for it. Every review ends with a call on how much human attention the PR actually needs: minimal, low, medium, or high. Easy changes stop clogging the queue, and reviewers put their focus on the question only a human can answer: is this the right change to make? Chelsea Hohmann, one of our engineering managers, walks through the whole loop. How much of your review cycle still happens after a human opens the PR?

Most engineering orgs know where AI is writing code. Fewer know who's reviewing it, what it costs, or whether it clears policy before shipping. That's the gap a lot of teams are stuck in right now. At EVOLVE, Karthik Jayaraman, Esq. (VP, Cybersecurity & AI Governance at Fiserv) shares a practical blueprint for closing it: embedding security, governance, evaluation, and cost management directly into developer workflows, with policy enforcement built in from the start. The result: teams move from AI experimentation to AI in production, with confidence, not just speed. EVOLVE is Cortex's conference for engineering leaders building their AI software factory. September 24, 2026, Convene 237 Park, NYC. Apply to attend: evolve.cortex.io

Measuring developer productivity made sense when developers wrote all the code. That's no longer the world most engineering orgs operate in. Leaders need to measure at a systemic level whether all that code is reliable, secure, and worth what it costs to run. In our latest webinar, Cortex CTO Ganesh Datta made the case for measuring the engineering organization as a system with the DRIVE Framework. Its five pillars measure a different aspect of engineering health, Delivery, Reliability, Initiatives, Vigilance, and Efficiency, all in the service of understanding whether you're sustainably operating and shipping high-quality software. He also broke down how to run an operational excellence review, the mechanism that turns those signals into decisions your leadership team can act on. The full session is now available on demand. Access the recording here: https://lnkd.in/eA2Ukn-m 

The E in the DRIVE Framework stands for Efficiency. It asks: are we allocating resources to the right problems? The Efficiency pillar puts three numbers in front of leadership: cloud spend against budget, AI/LLM token costs, and % of capacity spent on innovation. It's the easiest pillar to game, which is why it needs to be read in an OpEx review alongside the rest of the DRIVE metrics. A green Efficiency number sitting next to stalled Tier 1 initiatives and Sev0s up 40% quarter over quarter is a red flag. Read the final post in the DRIVE Deep Dive Series here: https://lnkd.in/dPb2bVGr #DRIVE #EngineeringOperations