Agentic AI for early-stage founders
Cosmic Lift helps pre-seed and seed teams build production AI agents, make them reliable, and integrate LLM capabilities into existing products. We also help R&D teams adopt agent-first development: preparing the codebase, setting up the development environment, and training the team to work effectively with coding agents.
Who you'd be working with
Cosmic Lift is Dmitri Kochelorov, and a short list of founders he takes on at a time.
17+ years building production software. Currently founder and CEO of FrontSail AI (a virtual operations assistant for service businesses). Before that, founding software developer at Starboard (AI agents for freight forwarders). Cosmic Lift is where I take a handful of consulting engagements with founders who need senior agentic-AI engineering before they're ready to hire a full team.
Before going independent: ten years at Planitar, from first hired developer to Director of R&D. I grew the team from one engineer to twenty-plus, stood up SOC 2 Type II, and built an in-house AI group for automated drafting. Before that, embedded systems at Cisco on Vodafone's residential gateways.
It all started a little over twenty years ago at the Technion Computer Science faculty. Some of the details have gone foggy. The 3 a.m. labs, less so.
We believe technology should not feel like magic. It should be predictable and transparent.
The Cosmic Lift principle
What we do
Agents that understand requests, use tools, follow workflows, recover from failures, and improve after launch. We design the full system around the agent: responsibilities, safe tool use, workflow orchestration, memory and retrieval when needed, guardrails, evals, observability, and feedback loops.
Make your dev team faster with agentic DevEx. Coding agents can boost delivery, but only when the environment is designed for them. We help teams structure repos, docs, tickets, reviews, tests, CI/CD, and quality checks so agents do useful work safely instead of generating confident nonsense at industrial scale.
Architecture for systems that need to last. Boundaries, data flow, service design, and the technical decisions that are expensive to reverse.
Add agentic experiences to the product you already have. Evolve the existing one with LLM-powered workflows instead of betting the roadmap on a rewrite.
End-to-end product engineering when you need someone senior building, not just advising. Frontends, backends, CI tooling, integrations, and the deploy pipeline.
Senior technical leadership without a full-time hire. Roadmap, tech strategy, hiring, vendor choices, and the hard architectural decisions.
Workshops for developers and coaching for the people who lead them. Practical guidance drawn from a decade of building and running engineering teams.
