The pitch sounded almost too ambitious for a three-month accelerator program: millions of autonomous AI agents, orchestrated under human supervision, automating the kind of enterprise workflows that still employ armies of contractors and offshore teams. But when Daniel Kornev and Dr. Mikhail Burtsev walked into Techstars Columbus in March 2024, they weren't exactly starting from scratch.
Both had spent years deep in the conversational AI trenches at DeepPavlov, one of the more credible open-source NLP projects to emerge from the academic world. Burtsev founded it. Kornev led product. They knew the technology. What they didn't have—and what most research-heavy founders struggle to acquire—was the infrastructure to turn that expertise into something enterprise buyers might actually pay for.
That's a harder problem than it sounds. By the time Sentius presented at Demo Day on May 29, 2024, the AI agent narrative was everywhere in Silicon Valley. Autonomous workflows. Agentic systems. The end of manual knowledge work. But walk into a CIO's office with that pitch and you'll often get polite skepticism. Maybe a pilot if you're lucky. Rarely a check.
Kornev mentioned securing their first enterprise design partner by Demo Day—an encouraging signal, though the partner remains unnamed and unverified publicly. Whether that partnership translates into revenue, retention, and reference customers remains to be seen. For now, Sentius is doing what most enterprise AI startups do: building credibility, one conversation at a time.
An Accelerator With Unusual Backing
Techstars Columbus was announced in October 2023, launching its first cohort in March 2024. The $110 million Timashev Family Foundation gift established the Center for Software Innovation at The Ohio State University, supporting a partnership with Techstars—the kind of capital infusion that signals serious long-term ambition. Ratmir Timashev, the billionaire behind the gift, has made no secret of his plan to turn Ohio into an AI hub capable of competing with the coasts. His Oh.io initiative and the Center for Software Innovation at The Ohio State University are pieces of that broader bet.
The accelerator's mandate was specific: emerging tech advancing foundational industries. Healthcare, energy, manufacturing, defense. The usual suspects for Midwest innovation. Managing Director Tim Grace and the Techstars network offered the standard deal—$20,000 for 5% equity, plus a $200,000 MFN SAFE—but the real draw was access to Ohio's growing network of corporate partners and university resources.
Twelve startups made the cut from what OSU described as a highly competitive applicant pool. Aspect Health went after hormonal imbalances with continuous glucose monitoring and coaching. Pedal Data focused on gait tracking and AI for neurodegenerative diseases. Blomso built live sensor technology for precision agriculture. And then there was Sentius—automating enterprise tasks with generative AI and autonomous agents, arguably the most technically ambitious project in the group.
Perhaps more than the founders initially expected.
The "Agentic Automation" Puzzle
Here's what Sentius is trying to do: take workflows that currently require significant manual labor—document extraction for compliance, billing and forecasting tasks, orchestration across multiple enterprise systems—and automate them using coordinated AI agents. One agent handles data extraction. Another validates the results. A third routes outputs to downstream tools. All under human supervision, at least in theory.
The company calls this "agentic automation," a term that's become fashionable in AI circles but still lacks a universally accepted definition. Their white paper outlines a "Teach & Repeat" architecture that supposedly allows enterprises to train agents on internal workflows without writing custom code. The platform integrates with OpenAPI-compatible services, which in theory makes it pluggable into existing tech stacks.
There's a demo environment on their website, and the company has been running an Early Access Program aimed at design partners. But public case studies? Sparse. Third-party benchmarks or customer retention data? Not yet disclosed. The first design partner mentioned at Demo Day remains unnamed.
For enterprise software buyers accustomed to multi-quarter pilots and extensive diligence processes, this isn't unusual. For investors and founders watching the AI agent space, it's a reminder that go-to-market remains a work in progress—even for teams with strong technical pedigrees.
Research Credibility Meets Enterprise Reality
Burtsev's background carries real weight in NLP circles. DeepPavlov, the open-source conversational AI framework he founded, powered academic research and early Amazon Alexa Prize efforts. He currently holds a fellowship at the London Institute for Mathematical Sciences, where he continues publishing on AI and natural language understanding. His co-founder Kornev led product at DeepPavlov and previously worked on what he's described as a personal AI assistant with substantial daily active users—a claim that remains unverified publicly but suggests some experience operating at consumer scale.
The advisors they've assembled suggest an intent to navigate enterprise sales cycles that require more than impressive research papers. Brett Brewer, former VP at Microsoft 365. Dr. Bill MacCartney, who held senior machine learning roles at Cohere and Apple's AI division. That's the kind of roster you build when you know technical validation and enterprise credibility are distinct challenges.
Kornev's LinkedIn posts trace a compressed founding timeline: an invitation to Techstars in January 2024, an early product demo at an AGI House hackathon in April, the first design partner by Demo Day in May. It's the kind of accelerated iteration that programs like Techstars are built for—validate fast, find early traction, iterate before the cohort ends and the real work begins.
What Else Emerged From Columbus
Sentius wasn't the only company to gain traction post-program. Doohickey AI, another cohort member building AI-native app integrations, announced a $500,000 investment from Drive Capital in September 2024. Besample, which connects researchers to human biological samples globally, closed a $1.1 million pre-seed round in May 2025 led by Gutter Capital. Those outcomes suggest the program is delivering on its promise to generate early-stage momentum.
Columbus Business First covered Demo Day, capturing what it described as palpable energy as founders pitched to local investors, corporate partners, and university stakeholders. The timing mattered, too. By late May, the AI agent narrative was heating up across the industry, but most enterprise buyers remained cautious. Sentius was betting it could bridge that gap with orchestration tools, OpenAPI integrations, and—crucially—human oversight baked into the architecture.
Whether they pull it off is still an open question.
The broader Ohio ecosystem is certainly betting on this kind of momentum. Timashev's gift and the infrastructure around OSU are designed to keep technical talent local and attract early-stage capital that might otherwise flow to San Francisco or New York. Techstars Columbus is part of that apparatus—a deliberate effort to build deal flow, mentor networks, and investor relationships in a region that's historically struggled to compete with the coasts.
The Unsettled Question
For Sentius, the Techstars stamp offers credibility as Kornev and Burtsev court enterprise buyers who want more than a GitHub repository and a polished slide deck. But the larger question—whether autonomous agents become the breakthrough automation layer enterprises adopt at scale, or remain a niche tool for technical teams willing to tolerate complexity—remains unanswered.
The go-to-market playbook for AI agents is still being written. Multi-agent systems sound compelling in demos but can be brittle in production. Human oversight requirements can slow down workflows to the point where the automation advantage erodes. And enterprise buyers, understandably, want proof that the technology can handle edge cases, maintain compliance, and scale without blowing up their operations.
Kornev and Burtsev are betting that research pedigree, enterprise-grade architecture, and accelerator momentum can get them in the door long enough to prove the platform works. They're not alone in making that bet—dozens of startups are chasing variations of the same vision. But with a named enterprise design partner already engaged, at least according to Kornev's public statements, they've cleared one early hurdle.
What happens next will likely depend on whether that first customer becomes a reference account, a case study, and—eventually—a renewal. In enterprise software, everything else is just setup.
