Founder & CEO, Semantiv
Building typed runtime infrastructure for agentic work: graph execution, contracts, traces, replay, and operation design.
About Semantiv
Typed operating models for AI-supported work.
Semantiv was founded to make AI-supported operations easier to model, test, trace, and improve. The company builds from a practical belief: agentic work needs explicit structure around meaning, evidence, failure, replay, and improvement.
01 Bio
Founder background behind the product.
Before starting Semantiv, Ariel led engineering teams and shipped production systems across enterprise analytics, consulting, media, visualization, recommendations, and applied language technology.
At Moody's Analytics, he directed three engineering teams and worked on cloud-native, event-driven modernization for analytics and reporting workloads. At ThoughtWorks, he led delivery for Spotify for Artists onboarding and artist-claim workflows, and for a McKinsey people-review platform. Earlier work included Nokia OZO Live, enterprise data visualization at Platora, recommendation systems at Outbrain, and NLP research software at Carnegie Mellon.
Education Math foundation
- MS Scientific Computing, Florida State University
- BA Mathematics, Florida State University
- Community Founder of Miami Effect TypeScript and Boston d3.js meetup groups
02 Background
The relevant through-line.
Engineering leadership at Moody's Analytics
Directed three engineering teams across core analytics and reporting products, including cloud-native modernization work that improved PDF generation speed and cloud cost.
Product delivery at ThoughtWorks
Led client-facing product engineering for Spotify and McKinsey, translating stakeholder needs into shipped workflow and platform products.
Visual, data, and AI systems
Built real-time media tools at Nokia, enterprise visualization systems at Platora, recommendation systems at Outbrain, and NLP research software at Carnegie Mellon.
03 Approach
Why Semantiv
AI systems need operational structure, not just more orchestration.
Semantiv reflects Ariel's current thesis: useful AI products will need typed models of the work they perform, so teams can reason about correctness, failure, evidence, replay, and improvement.
-
01Model the operation, not just the automation. -
02Make validity explicit through contracts, evidence, and traces. -
03Design AI systems that can be inspected, replayed, repaired, and improved.
Next step
Talk to Semantiv.
Reach out to discuss whether typed runtime models fit your team, product, or AI-supported operation.