Zartis

Meet Victor Palomares, one of our Engineering Directors at Zartis! With 20+ years in software engineering and 7+ years at Zartis, Victor is a key part of what makes our culture so strong: supportive, approachable, and committed to helping people get better. Thanks for helping make Zartis a place where people can genuinely grow, Victor! 🤝 #TeamZartis #MeetTheTeam #SoftwareTeam #Engineering #Zartis

📍 Claude Code Lab - Amsterdam (11 Feb) & Berlin (12 Feb) You’ve got a working AI prototype. It looks great in a demo. Then you try to ship it. Suddenly the real questions appear:  – How do we integrate this into an existing system?  – What breaks under real user behaviour?  – Where do prompts stop and engineering start?  – How do we make this reliable, secure, and maintainable? These are exactly the problems teams worked through at our Claude Code Lab in Dublin: hands on, with code open, and real constraints on the table. 🧑💻 Next week, we’re bringing the same workshop format to Amsterdam and Berlin, delivered in collaboration with Anthropic! The focus stays the same: practical implementation, real trade-offs, and learning that only comes from building. 📸 Moments from our last workshop in Dublin Killian O'Connor, Ricky Hill, Jose V., Aleix Canet Vidal, Angel G., Albena Krasteva, and Paddy Clifford

Your AI demo impressed the board. Six months later, you're still running it manually. Why? The gap between an AI MVP and a production-ready system is where most initiatives quietly fail. Not because the idea was wrong. Because production demands something entirely different from a prototype. Production-ready AI is not a model milestone. It's an engineering, operational, and organisational threshold. In our latest piece, we break down what that actually looks like: the reliability engineering, the observability requirements, the cost controls that get designed in (not bolted on), and why human-in-the-loop is a feature, not a failure. 🔗 https://lnkd.in/d2UW5V8D

📆 Amsterdam (11 Feb) and Berlin (12 Feb) 📣 This February, we’re bringing the Claude Code Lab to two new cities! Last year in Dublin, we filled a room with engineers and spent the day doing what matters most: building. There’s a different feel in a room when people are actually building. That’s what we saw in Dublin; engineers testing ideas live, challenging assumptions, and learning from each other in real time. 🛡️ Same principle as before: Practical, hands-on, and focused on real implementation rather than theory. Delivered in collaboration with Anthropic, the Claude Code Labs are designed around practical work: integrating, experimenting, breaking things, and figuring out what really holds up in production. 📽️ Moments from the Dublin edition. Reach out to secure a spot! #AI #Claude #ClaudeCode #Anthropic #Zartis #Workshop #Hackathon Killian O'Connor Ricky Hill Adrián Sánchez de la Sierra Aleix Canet Vidal Jose V.

🎙️ Season 5 kicks off with a topic that actually matters: real AI transformation! For the first episode of the new season of Story of Software, we’re joined by our own Adrián Sánchez de la Sierra, Head of AI & Innovation at Zartis. Adrian works where AI ambition usually breaks down: the gap between demos and production, hype and sustained value. In this episode, we unpack why most organisations are still using AI to replace existing processes, and why the real opportunity lies in reimagining how systems, teams, and organisations work altogether. Drawing parallels with electricity and other general-purpose technologies, Adrian makes a clear case for why AI transformation is: ➡️ slower than people expect ➡️ harder than most teams plan for ➡️ and far more powerful when done properly If you’re serious about moving beyond proofs of concept and building AI that actually holds up in production, this episode is for you. 🎧 Listen to the full episode here: https://lnkd.in/dB3MCvMm #StoryOfSoftware #AITransformation #ArtificialIntelligence #Leadership #TechPodcast #Season5 #Zartis

We just wrapped up a 2-day hackathon with Ovation Incentives focused on agentic coding and #ClaudeCode! 🎯 The goal was simple: Enable the entire team - product and engineering - to use Claude Code in a way that genuinely improves performance and speed of delivery. How we approached it: 🔹 Day 1 started with the core building blocks of Claude Code — what it can (and can’t) do, and how to get real value from it. 🔹 Then we went hands-on, splitting into product and development groups to work on their actual use cases. The product team took an idea for a new app and used Claude Code to turn it into clear epics and user stories, ready for implementation. The development team picked those up and built a working MVP. 👉 What stood out: The work produced in those two days had originally been estimated by the team to take weeks, if not months. Yes, there’s an upfront investment in setting things up properly, but they’re now in a far stronger position to keep compounding productivity gains with Claude Code going forward. 🤝 Big thanks to the Ovation Incentives team for the trust, engagement, and openness to experiment! #AI #CodingAgents #Claude #ClaudeCode #AITransformation #Anthropic #Zartis #OvationIncentives

Only 5% of companies report ROI from GenAI because true control lies in managing the main challenges: hallucination and non-determinism. This cycle of inflated expectations and technical reality leads to skepticism and shelved projects - a challenge reflected in industry reports suggesting that a significant portion of AI proof-of-concepts never reach production. How can you avoid this trap? 🧾 Download our #whitepaper to find out: https://lnkd.in/dvsxUcFQ #AI #Hallucinations #Determinism #LLMs #Zartis #Resources

Deals are won or lost long before term sheets. Most of the time, they’re decided in the code. In financial services M&A, technology isn’t a support function; it underpins compliance, scalability and valuation. Are you really looking under the hood before you sign? Find out the sector-specific risks buyers need to assess: https://lnkd.in/d9ywj9Fv

Most AI initiatives fail for a boring reason. We keep treating AI like a tool. In reality, production AI behaves like a system: 🔹 Probabilistic, not deterministic 🔹 Sensitive to context, data quality, and infrastructure 🔹 Expensive in subtle ways (tokens, latency, human review loops) That’s why so many “successful” pilots collapse in production. Not because the model was bad - but because nothing around it was engineered. The teams we see succeed stop asking: “Which model should we use?” And start asking: “How do we design reliability, cost control, and human oversight into the system?” AI isn’t plug-and-play. It’s architecture, governance, and discipline, or it’s just noise. 🧾 Read more about Engineering Determinism and Reliabity for LLMs: https://lnkd.in/dNtAazy2