Nebuly

Part 2 of Nebuly's blog about how businesses are deploying customer-facing AI is about how consciously quality standards are set and maintained. ŌURA prioritised what members actually experienced in support interactions, not how many cases were "contained" by the AI agent. Those metrics improved anyway along with the overall experience. If you liked part 1, and you're wondering what good looks like for AI both within and beyond the contact centre, learn from ŌURA! Next up, Part 3, on the strategic questions brands will have to take seriously, and soon. Agent-of-choice, choice-of-agents, and whether that's really a binary. https://lnkd.in/eEccZ8Bp

Since I joined Nebuly, I’ve been working on one problem: how do you actually measure AI proficiency? If that sounds like a small problem, it isn’t. Companies are spending millions on AI training, and most still can’t really say whether it’s working. Anthropic has been working on the same question with its AI Fluency Framework. I found their 4D model super interesting, and it became one of the starting points for the new product I’m building at Nebuly: AI Proficiency. Still very much in the works. Very complicated. Full of potential. I just wrote a blog about what we’re learning so far: https://lnkd.in/e6QZgnHR

We've spoken with hundreds of businesses since launching Nebuly in the UK in April. I've been thinking a lot about what's holding some back from really committing to AI. Beyond the obvious, governance, talent, and just the fact that we're still in the early phase of this new technology, I wanted to share a different perspective around readiness, focus and ambition, in customer-facing AI. Here's part 1 of 3, covering the premise along with some great examples like Revolut, Virgin Atlantic and ŌURA. Stay tuned for part 2, digging into what our client Oura do so well. Finally part 3, exploring some strategy questions. https://lnkd.in/eQPbWbir

We've spoken with hundreds of businesses since launching Nebuly in the UK in April. I've been thinking a lot about what's holding some back from really committing to AI. Beyond the obvious, governance, talent, and just the fact that we're still in the early phase of this new technology, I wanted to share a different perspective around readiness, focus and ambition, in customer-facing AI. Here's part 1 of 3, covering the premise along with some great examples like Revolut, Virgin Atlantic and ŌURA. Stay tuned for part 2, digging into what our client Oura do so well. Finally part 3, exploring some strategy questions. https://lnkd.in/eQPbWbir

We'll be joining the 5th Annual Conversational AI & Customer Experience Summit as a Startup Sponsor. Most AI investment in customer experience has gone into support: efficiency, cost, deflection. The next question is what happens when that same AI moves into the product itself, and personalization has to be built from how each user actually behaves, not configured once and left alone. We're looking forward to bringing that conversation to Munich, and hearing how other teams are tackling it. 📅When: 28-29 October 2026 📍Where: H4 Hotel München Messe, Germany Come say hi! Thank you Altrusia for the opportunity and partnership. #CACES #CACES2026

The AI literacy conversation is moving from personal advantage to institutional requirement. The UAE is the clearest proof so far. A national AI curriculum, mandatory across schools, with teachers trained to deliver it before the tools reach a single classroom. Governments are now formalizing what enterprises have been working through for a while: giving people access to AI is the straightforward part. Knowing whether they are actually getting more capable with it is the question that needs data to answer. That is what we built our AI Proficiency module for. It gives every team an AI Proficiency Score, so you can see how capability with AI develops over time and tie adoption investment back to real progress. It's also why the UAE is on our mind. We are expanding into the region soon, and we want to meet people who want to build this with us. If you know someone great thinking about AI adoption and measurement in the Emirates, send them our way!

AI coding agents are a tool. Probably the highest impact tool ever built for developers, but still a tool, with a learning curve like any other. So a developer’s output today tracks the skill set they built around that tool more than the tool itself. Giving the agent the right context is where everyone starts. Being able to properly test what gets built is the part nobody skips. That does not mean a human reads every line. With enough automation around it I am fine with code that gets written and reviewed without me in the loop, as long as the checks are strong enough to catch it when it is wrong. Which is where the real gap opens, and it is an engineering gap. Writing skills the agent can reuse instead of re-explaining the same thing every week. Shaping the harness it runs in: which tools it can call, what it is allowed to touch, where it gets its data. Setting up loops where it plans, runs, tests and corrects itself before anyone opens the diff. One example from our engineering team at Nebuly: for every bug that gets reported, a routine writes the end to end test that reproduces it. That test becomes the acceptance criteria for the fix, whether the ticket is picked up by an autonomous agent or by a person driving one. That work is not prompting. It looks a lot more like systems engineering applied to your own development process. How you make the software get built is the difference now.

Sono entusiasta di annunciare il nuovo passo della mia carriera andando a ricoprire la posizione di Senior Software Engineer presso Nebuly!! 🥳🥳 Ringrazio tutti per l'opportunità concessa! Non vedo l'ora di mettermi in gioco per contribuire al raggiungimento della mission di Nebuly e al suo continuo miglioramento🚀🚀💪

Welcome to the team, Francesco Lasagno! He joins our full-stack team as a Software Engineer, at a moment when demand for what we do is accelerating. As more enterprises run AI agents in production, understanding what those agents actually deliver becomes mission-critical, and we're building fast to give companies that clarity. That takes a strong engineering team, and we're still growing it. We're hiring for several roles right now (link in the comments). Great to have you with us, Francesco.

The Nebuly platform sits on top of enterprise AI deployments and turns those conversations into a read on whether the AI is actually working. Many of our customers are large enterprises with a hard rule: their data never leaves their environment. So the same product ships as our multi-tenant SaaS and as self-hosted installs running inside customers' own infrastructure. One codebase, two very different operational realities. We manage all of it as a single GitOps setup on Kubernetes. ArgoCD reconciles every cluster against the manifests in one repo, including ArgoCD itself. Terraform for provisioning, Helm charts for defining customer's K8S clusters, and a Grafana and OpenTelemetry stack so we can see what each deployment is really doing. A change that's easy on SaaS has also to work in an environment you don't fully control, and the reverse. Making that feel boring and predictable is the actual work. I'm hiring a DevOps / Infrastructure engineer in Turin to own this with me. Solid Kubernetes and IaC in production, and comfortable with the fact that some of our clusters live in someone else's cloud. Hybrid, working close to the whole engineering team. If that sounds like your kind of problem, the role is linked in the comments.