# New Cast AI Console Experience > Source: > Published: 2026-10-16 10:16:46+00:00 ## Console that works the way you do We rebuilt the Cast AI console around how you work, not how we ship features. Nothing’s gone. We moved things, grouped them better, and cut the guesswork out of where to start. That comes down to two big shifts: Cast AI is now fully automated, and the console’s structure finally matches how you actually think about your infrastructure. ## We heard you & What we did ### part 1: automation wizard **Cast AI is now truly automated** On call after call, our customer success team watched the same thing happen: a customer would get access, then stall. Nothing broken. They just didn’t know what to turn on, in what order, or what “correct” looked like for their infrastructure. And “correct” meant something different depending on who was asking. A brand-new user wanted Cast AI to just work. A power user wanted every dial available, and wanted to know none of them got hidden. For a while, every feature came with its own configuration and its own granularity. Auto-generated node templates were our first real step forward: Cast AI could read a cluster and generate templates on its own, instead of a user building them from scratch. It was real progress, but the user still had to read through what got generated, understand it, and decide which templates to turn on. That gap is what led to the wizard. If Cast AI could already read a cluster well enough to generate templates for it, it could read that cluster well enough to decide which fit, and just turn optimization on. The one thing Cast AI can’t read for you is what you actually want. Whether cost matters most, performance does, or something balanced between the two, is a judgment call, and only you can make it. So that’s the whole wizard: two questions. What matters most when Cast AI optimizes it, cost, balance, or performance? And what’s this cluster for, production, dev, a mix? Once you decide what is best for you, Cast AI reads your infrastructure in full and builds the node templates and configuration to match, tuned to your actual cluster, not a generic default. **Setup that used to take days or weeks now takes seconds**, and you don’t need to be an expert to get there. Power users lose nothing in this. Every setting the wizard set is still there, still editable, still as granular as it’s ever been. The wizard just means you don’t have to touch any of it unless you want to. ### part 2: Information architecture **The Cast AI console now matches how you actually run your infrastructure** We pulled this from years of tickets, customer calls, and meeting notes, then looked for the pattern underneath all of it. Why is there no single place to see which cluster needs attention right now? Why do savings numbers differ depending on which page you’re looking at? Why don’t workload counts match between two pages that are supposedly showing the same thing? We found the answer. The console grew the way most products do: feature by feature. Every new capability shipped with its own context, free to define its own logic and its own defaults. It was the fast way to build, and each page made sense in isolation. But step back a few years later, and the parts don’t add up to a whole. Getting the full picture means stitching it together yourself. **What we did at the organization level** Organization view is now a clusters cockpit. Reliability is a core pillar here, not an afterthought, with a cluster heatmap up front. You look at it, see which cluster needs attention, and land straight in that cluster’s view. **What we did at the cluster level** We reorganized the Cast AI console around what you actually run, not around what we happened to ship. Navigation now has one Inventory section, your assets in one place. It starts with Nodes and Workloads, and every asset type follows the same pattern. Inside inventory, you pick your lens: Autoscaling, Savings, Compute Cost, Reliability, Network, GPU. Each one is a set of widgets built around a specific way of looking at your inventory. Care about reliability first? Switch views and the page leads with [OOMKills](https://kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle/#container-states), error rates, and health signals. Care about spend? The cost view leads with what your [workloads](https://kubernetes.io/docs/concepts/workloads/) cost this month and where it’s going. It’s a principle running through the whole console now: every entity can be observed through different lenses, and the UI serves whichever perspective matters to you in that moment, instead of forcing you to hunt for it across separate pages. **And yes, we heard you on this too** Dark mode is here. It came up more than almost anything else in the feedback, and now it’s just… there. ## What’s next This is the start, not the finish line. A few things already in motion: - Inventory keeps growing. Pods, volumes, and more data are coming, with the same logic applied everywhere else. - We’re also building a cockpit for feature control: one place to manage every feature you’re using in the Cast AI console. - Workload autoscaler is getting the same treatment, simplifying how scaling policies get set instead of asking you to hand-tune every rule. - And most dashboards will eventually carry a customizable theme. You decide what your view looks like. That’s not just convenience: the views people actually build tell us what matters to them, and that shapes what we build next. **Built around what you need** That’s the thread running through everything above: a Cast AI console shaped by what you actually need. This is not a one-time thing. It’s how we build, and it’s how we’ll keep building. Want a more direct hand in what comes next? We run research sessions with real users, and we’d like you in the room. [Opt in here](https://docs.google.com/forms/d/e/1FAIpQLSeyaiaOnOfNmzJhmlSLipdKrHAs4VQ076p7ubcMl37P7Lrs1g/viewform).