About
Hi, I’m Cai Felip.
I build voice AI that lets businesses actually talk to their customers – at any hour, in any language, without a phone tree in sight.
These days most of my waking hours go into Vocals, the company I founded and run. We make voice agents that handle real customer conversations – inbound and outbound calls, support, sales follow-ups, surveys – across all kinds of industries. The idea I keep coming back to is treating voice as an API and people as the interface: the phone call is still how an enormous amount of business gets done, and it’s been left almost untouched by good software. We’re fixing that.
Rather than reinvent every piece, Vocals sits on top of the best speech and language models out there and orchestrates them into agents that feel natural to talk to. The hard, interesting part isn’t any single model – it’s everything around it: latency, handoffs, knowing when to stay quiet, knowing when to pass a caller to a human with full context. That’s the craft I’m obsessed with right now.
How I got here
Vocals isn’t my first company. Before this I founded and ran Union Avatars, a Barcelona-based startup I started in 2020 to build realistic, full-body 3D avatars from a single selfie: portable digital identities you could carry across virtual worlds, games, and meetings. We tackled the messy problem of identity fragmentation in the metaverse, and worked with partners ranging from global brands to platform builders. I’ve been around the avatar, 3D, and blockchain worlds since the early days, well before “metaverse” was a word people argued about.
I’m a multiple-time founder and a builder at heart, with a path that’s wandered through video games, music, identity, and now voice AI. I like the early, messy part of a company – talking to customers, shipping something rough, watching what breaks, and doing it again the next morning. A lot of what I’ve learned came from selling and supporting real products to real businesses long before they were polished, which is probably why I care so much about the unglamorous details that make a product trustworthy.
I work from Catalonia, and I move between Spanish, Catalan, and English all day – which, conveniently, is exactly the kind of multilingual reality the products I build have to handle.
How I think about work
I’m a high-trust, fast-moving kind of operator. I’d rather make a clear decision and correct it quickly than wait for perfect information. I protect long, uninterrupted blocks for deep work, keep a “second brain” so nothing important lives only in my head, and try to leave every system a little more organized than I found it. Bias toward action, written things down, fewer meetings.
Off the clock
When I’m not building, there’s a good chance I’m on a motorbike. Riding is my favourite way to switch off – the kind of full attention that leaves no room for a to-do list, just the road, the bike, and the next corner. The rest of the time you’ll find me on a padel court, training, or going down a rabbit hole on whatever new AI capability just dropped. I’m endlessly curious about tools, workflows, and the small productivity systems that compound over time – and I like sharing the ones that actually stick. That’s a big part of why this blog exists.
What I write about here
This is where I think out loud: building an AI startup from outside Silicon Valley, the realities of voice technology, lessons from selling and shipping, and the systems I use to stay sane while doing all of it. Some posts are practical, some are just me working through an idea. If any of that is your thing, you’re in the right place.
What I’m doing now
Building Vocals – growing the platform, working with our early customers, and figuring out which vertical to go all-in on. Writing more. Playing more padel than my schedule technically allows.
Thanks for stopping by.
– Cai