Elisenda Bou-Balust

Who they are

Elisenda Bou-Balust is CEO and co-founder of Cala — previously co-founded Vilynx, an AI video-understanding company acquired by Apple in 2020, where she then ran Apple Media Knowledge.

Person

Elisenda studied Electronic Engineering at the Universidad de Las Palmas de Gran Canaria, then Telecom Engineering at UPC in Barcelona, and completed her MSc thesis at MIT (2007–2010) before earning a PhD at UPC in 2016 on wireless power transfer and software architectures for distributed spacecraft applications. She went straight from academia into founding: in 2011 she launched Zeptocode (marketing-campaign tech for brands like Coca-Cola and Ray-Ban), then co-founded Vilynx in 2012 — pioneering self-learning multimodal AI architectures for video understanding and knowledge graphs — before Apple acquired the company in 2020. Post-acquisition she stayed on as Head of Apple Media Knowledge, giving her rare inside exposure to how a hyperscaler applies AI at scale. She also co-founded Estimtrack in 2016, applying ML to surgical scheduling and OR efficiency, which was subsequently acquired by dir/Active. In January 2025 she left Apple to found Cala, an infrastructure company building what she describes as the world's largest information graph — a verified data layer for AI agents. The through-line across all of it is the same: representing knowledge in machine-readable form, from video semantics to hospital workflows to agent-ready graphs. She posts on LinkedIn about knowledge graphs, agentic AI, and the shift from SaaS to agent-driven economies — practitioner-level, not pundit. She joined the Ferrovial board in early 2026, and remains an affiliated researcher and associate professor at UPC.

Company

Cala was founded by Elisenda Bou-Balust in Barcelona in 2025 and secured one of Spain's largest pre-seed funding rounds that year — a €7M round led by Lightspeed Venture Partners. Creandum included it in their 2025 #EuroSeed50 list of the most promising early-stage European startups. The product is infrastructure for AI agents: a verified, explainable information graph built for grounding agent queries in trustworthy data. Cala is already integrated with Anthropic's Claude, and has partnered with ElevenLabs to build a voice-first product on top of the graph.

Market

Cala operates in the emerging AI data-infrastructure layer — specifically the problem of giving autonomous agents access to verified, structured knowledge rather than hallucination-prone retrieval. The competitive field is nascent but contested: knowledge-graph and context-layer companies are racing to become the trusted data substrate for agent orchestration frameworks. Elisenda has been vocal at 4YFN and MWC 2026 that the shift from SaaS to agent-driven economies is the defining market transition, which frames Cala's positioning against both general RAG/retrieval vendors and proprietary enterprise knowledge bases.

Network

Elisenda's most visible recent connections are fellow Catalan AI founders and the global agentic-AI circuit. At 4YFN Barcelona 2026 she shared a panel with Alvaro Martinez (CEO, Luzia) and Daniel Hulme (Chief AI Officer, WPP); at Spain MWC she appeared alongside Ariadna Font (CEO, Alinia AI), Sergi Bastardas (CEO, Orbio AI), and Daniel Carmona Serrat (CEO, Supersonik). She is frequently co-featured with Ariadna Font Llitjós in events spotlighting Catalan women in international AI.

Alvaro Martinez· CEO, Luzia
Daniel Hulme· Chief AI Officer, WPP
Ariadna Font· CEO, Alinia AI
Sergi Bastardas· CEO, Orbio AI
Daniel Carmona Serrat· CEO, Supersonik

How they likely show up

Serial founder (Zeptocode → Vilynx → Estimtrack → Cala) with each company building on the same knowledge-representation thesis → she thinks in compounding bets, not pivots.
PhD research on distributed spacecraft software architectures, then AI at Apple scale → comfortable operating at both deep-technical and systems levels; won't be impressed by surface-level architecture talk.
Active LinkedIn presence on knowledge graphs and agentic AI with practitioner framing → she engages publicly, values substance over hype, and will have noticed whether you've read her posts.
Mixed tenure shape — left Apple after the Vilynx acquisition to found again immediately → low tolerance for large-org inertia; moves when she sees the thesis clearly, not when it's safe.
Ferrovial board seat alongside running an early-stage startup → operates across multiple registers (governance, deep tech, early-stage execution) simultaneously; time is genuinely scarce.
Speaks at MWC, 4YFN, and regional Catalan AI events → embedded in the Barcelona ecosystem and uses it deliberately, not just for visibility.

Conversation tips

→ Reference her specific LinkedIn framing — 'for ten years the answer is a graph' — she'll know immediately whether you've engaged with her thinking or just skimmed her bio.
→ Ask about the Apple chapter specifically: what it was like running Media Knowledge at that scale after the Vilynx acquisition, and what it confirmed about the data-layer problem she's now solving.
→ Don't conflate Cala's information graph with generic RAG — she has a precise technical thesis about verified, explainable data vs. retrieval; show you understand the distinction.
→ The Ferrovial board seat is a useful opener: it signals she's thinking about enterprise trust and governance of AI systems, not just infrastructure plumbing — worth exploring that angle.
→ She presents the SaaS-to-agents shift as a structural economic transition, not a feature cycle — engage with the macro argument, not just the product.

Toolbox

Openers

Open on the LinkedIn post where she writes 'for ten years the answer is a graph' — it's the clearest public statement of why she left Apple to build Cala, and referencing it shows you've done more than read her title.
Mention the ElevenLabs partnership — Cala building a voice-first product on top of a verified knowledge graph is a specific architectural bet about how agents will consume data, worth unpacking.
Bring up the Estimtrack exit — ML for surgical scheduling is an unexpected detour between Vilynx and Cala, and it reveals something about how she thinks about applying knowledge graphs in high-stakes, real-world environments.

Discovery questions

When you were running Apple Media Knowledge post-acquisition, what did you see about how hyperscalers handle knowledge at scale that convinced you the infrastructure layer had to be rebuilt from scratch for agents?
Cala is already in Claude — how do you think about being embedded inside a model provider's ecosystem versus staying infrastructure-layer independent as the agent orchestration market consolidates?
You've now had two companies acquired and founded a third from the same underlying thesis about knowledge representation — at what point in building Vilynx did you realise the problem was bigger than video, and that agents would eventually need this layer?

Avoid

Don't frame Cala as a 'data startup' in the generic sense — she has a specific, technical position on verified knowledge graphs for agentic AI, and flattening it into a data-management pitch will signal you haven't engaged with what she's actually building.

Make it yours