The archive · AI & Models · Product decision · 2026
Cala's knowledge-layer bet: fix AI agents with verified facts, not another model
Barcelona's Cala emerged from stealth in May 2026 with €7M to build 'Matter of Facts,' a verified knowledge graph meant to stop AI agents hallucinating
Cala
What the business is
Cala builds Matter of Facts, a verified knowledge layer for AI systems: public web information structured into entity-based datasets on companies, people, pharmaceuticals, legal cases, financial instruments and products, served through a REST API, CLI, MCP server and Python and TypeScript SDKs.
Starting capital：A €7M pre-seed round led by Lightspeed Venture Partners with Kibo Ventures, Kfund and Masia, announced when Cala left stealth on 2026-05-11.
How it started
After selling her computer-vision startup Vilynx to Apple in 2020, Elisenda Bou-Balust teamed up with Issey Masuda Mora in Barcelona to found Cala on a simple argument: AI agents keep failing not because models cannot reason, but because they lack a trustworthy layer of verified knowledge about companies, people, medicines and products.
What happened
Cala built Matter of Facts by curating and structuring public web information into entity-based datasets and made it accessible through a REST API, CLI, MCP server and Python and TypeScript SDKs. The company stayed in stealth until May 2026, when it emerged with a €7M pre-seed led by Lightspeed Venture Partners with Kibo Ventures, Kfund and Masia; days earlier TechCrunch had already put it on its 21 European startups to watch list, calling the verified knowledge layer the missing piece for professional AI agents.
No ending yet — it is still running.
Background
Cala is a Barcelona startup founded by Elisenda Bou-Balust, who sold her previous company Vilynx to Apple in 2020, and Issey Masuda Mora. Its bet is that the missing piece in the AI-agent stack is not another model or framework but a verified knowledge layer: structured, factual data about companies, people, pharmaceuticals, legal cases, financial instruments and products.
Its product, Matter of Facts, aggregates public web information into entity-based datasets and exposes them through a REST API, CLI, MCP server and Python and TypeScript SDKs, so developers building agents can pull current, machine-readable facts instead of letting models reason from memory.
Cala emerged from stealth on 2026-05-11 with a €7M pre-seed led by Lightspeed Venture Partners, joined by Kibo Ventures, Kfund and Masia. TechCrunch had already named it one of 21 European startups to watch, framing it as an answer to the problem that professional agents lack a trustworthy knowledge layer.
What has to be true
- Agents are only as reliable as the facts they can reach, and no one had built the verified data layer beneath them as a product
- Structured entity datasets with APIs are easier for agents to consume than raw web pages, cutting hallucination risk at the source
- Founder credibility from Vilynx's Apple exit gave a stealth startup immediate attention from investors and media
- Selling to developers through REST, CLI, MCP and SDKs lets the knowledge layer plug into every agent framework rather than betting on one
What can be applied
Pick the layer everyone else assumes exists: Cala bet agents fail on facts, not reasoning, and built verified data infrastructure instead of one more model or agent app
Aftermath
As of 2026-05-11, Cala is out of stealth and building: Matter of Facts is positioned as the verified knowledge infrastructure for developers creating agentic systems, and the €7M pre-seed from Lightspeed, Kibo Ventures, Kfund and Masia funds the build-out of its entity datasets. TechCrunch's inclusion of Cala in its 21 European startups to watch list signaled early ecosystem attention, and the company's stated ambition is to become the knowledge layer that professional AI agents currently lack.
Sources
- Barcelona AI infrastructure startup Cala emerges from stealth with €7 million pre-seed round
- Beyond Lovable and Mistral: 21 European startups to watch
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card