My name is Sebastián, and I just wrapped VentureWell's E-Team Pioneer Program with Capicú Technologies and my partner-in-crime, Luis Luna-Betancourt. I expected sharper language for our deck. I got a higher bar for what I'm allowed to believe without evidence.
Technical founders love shipping. You can always justify one more sprint. You can always hide behind progress. The Pioneer program kept tugging at the same thread:
If we're building something for healthcare, who is it for, exactly, and what work does it remove on a Tuesday afternoon when everything's on fire?
That question followed me through Miami like humidity.
What the program forced us to do
We touched the classics: market applications, value chains, customer discovery, value propositions. What mattered was how they were used. Nobody let us hide inside pretty words.
If we said "downstream impact," someone asked, downstream for who? If we said "clinical burden," someone asked, which role feels it first? If we said "efficiency," someone asked, what's the unit: minutes, touches, days, denials, after-hours charting?
I'm comfortable in systems. Inputs, outputs, clean diagrams. Healthcare doesn't behave like that. People improvise. Workarounds become policy. Documentation becomes archaeology. If you're not careful, you build for the diagram instead of the day.
Two lines that made the work feel real
I've heard a lot of "this is interesting" over the years. It's flattering and useless. The lines that hit me in Miami were simpler, from people who live the problem.
"If your technology works the way you say it can, it would help me focus more on patients."
"You can make my practice earn more and schedule more patients next month."
That second one surprised me. It's blunt. It's not romance. It's operations. In healthcare, impact isn't a vibe. It's a calendar. It's staffing. It's whether an exhausted clinician spends another hour after a shift finishing notes.
Those lines turned the work into a responsibility.
My stance after Miami
Motion doesn't count unless it changes a workflow. A workflow doesn't change unless a specific role says it changed. A role doesn't care unless the cost shows up in their day.
So when we say we're building for healthcare, we don't get to say "clinicians" like it's one blob. We have to name roles: the ED physician completing documentation retrospectively, the charge nurse keeping the floor moving, the case manager chasing missing context, the post-acute intake coordinator reviewing hospital docs to accept a patient, the utilization management reviewer on the payer side trying to interpret what happened.
If we can't name who suffers, we can't claim we're fixing anything.
What we're testing, not assuming
The core problem we're investigating: acute clinical events happen in cognitively chaotic environments. Documentation gets completed after the fact. Clinicians reconstruct reality. Then as information moves from acute care to post-acute settings and payer review, meaning degrades. People downstream spend time validating, clarifying, and reinterpreting the record.
We're not claiming this is always true. We're treating it as a hypothesis chain. We're also not framing it as "documentation doesn't exist." It exists. The pain is the cost of reconstructing a high-stakes event after it already happened, across multiple people who weren't in the room.
We're aiming to deliver on-device intelligence directly where biomedical data is collected. If you want to see what we're up to: capicu.ai
And if you're a clinician who charts after chaos, a case manager who loses time to missing context, a post-acute intake coordinator forced to interpret half-stories, or a payer-side reviewer who spends hours validating what should've been clear, I want to talk. Interviews first. Evidence next. Decisions after.
S.
#MakingAIFitAnywhere