NaNLABS ads

Follow %COMPANYNAME%, your tech sidekick, for useful content.

Companies with embedded AI governance are 3x more likely to reach ROI in a year. So why is almost no one doing it? We’ve all heard “move fast and break things.” But when it comes to AI, what breaks first is trust. Governance isn’t bureaucracy, it’s how you make AI scale responsibly. And the companies that understand that are alreadwpulling ahead. In our latest vlog, Daniela Buján shares what we’re learning from helping teams make AI innovation actually sustainable. Watch it now and let us know what you think!

Most companies don’t have a data problem. They have a data movement problem. Data that’s stuck in silos. Pipelines that break at scale. Streaming architectures that weren’t built for real-time decisions. And in industries like EV charging, that’s not a bottleneck, it’s a blocker. In 2025, we are seeing shift: more CPOs started asking for cloud-native, real-time solutions that are AI-ready by design. But that only works if the data stack can handle: - Multimodal ingestion from chargers, vehicles, and sensors - Stream/batch hybrid pipelines - Dynamic routing, retries, and latency thresholds - Lineage and observability from source to model In this clip, I break down what it actually takes to go from fragmented telemetry to actionable insights, and why most teams underestimate the complexity of production-grade streaming.

I’ve seen it happen too many times: AI projects that look great in a demo… fall apart in the real world. Why? Usually for one of these three reasons: - Dropping AI into legacy systems – fragile prototypes that break under load. - Buying black-box tools – new silos, no telemetry, no visibility. - Skipping data ops – small format changes can silently break models. One real example: a chatbot launched without full support history doubled escalations instead of reducing them. We broke down these traps, and how to avoid them, in the latest #NaNTalk 👇

Still think AI agents are just prompts? In real world systems, it’s never just one model making decisions. You need data pipelines. Context orchestration. Real-time memory. And infrastructure that can learn, adapt, and react, fast. In this episode of our #NaNTalks, I break down: – Why reasoning is just the tip of the iceberg – How we built a modular AI agent system for real-time pricing – And the architectural trade-offs that come with scale Not a demo. This is how we make AI work in production.

Most charging networks still rely on static dashboards to set prices. That works, until it doesn’t. Flat rates and fixed pricing rules can’t adapt to what’s happening in real time. When demand surges, or chargers sit unused, revenue takes a hit and the grid gets stressed. We explored what would happen if an AI agent could handle that complexity instead. Not just automating tasks, making real-time pricing decisions based on usage patterns, time of day, and location. A system that learns and adapts on the fly. The results? - Up to 25% increase in revenue. - Better load balancing across the grid. - A smoother experience for drivers. In this video I explain how we approached the architecture, what trade-offs we faced, and why reasoning in the moment matters more than ever.