Celestis is a private knowledge platform that turns proprietary documents, manuals, SOPs, and controlled reference material into fully private, AI-accessible knowledge—available offline or online, under the customer’s control.
Core Idea
Enterprises and experts need accurate AI over their own material, but cannot risk putting that knowledge into a third-party cloud for training, retention, or always-on connectivity. Celestis solves this with private capsules of knowledge that stay yours: hosted, local, or fully offline on mobile and edge devices. Work continues in the field, on a plane, mid-ocean, or on locked-down/air-gapped networks. Cloud is used when it helps; offline is available when it must be.
Platform and Products
The core building blocks include:
-
Capsules: Reference or live knowledge (documents, models, or connected data) that remain under customer control—hosted, local, or mobile offline.
-
Vessels: Groups of capsules with smart routing so answers stay within trusted sources.
-
Clones (Marketplace): Shareable copies that let creators monetize private knowledge.
-
Reader Mode: Side-by-side documents and AI for research and review.
-
Super AI: Specialized AI fed by the customer’s own capsules and clones.
-
Frontier AI: Access to leading models (including Claude, OpenAI, and Grok) in one place, used individually or mixed.
-
Shortcuts & Classifiers, BYO MCP (bring-your-own Model Context Protocol servers for live data or artifacts), and BYO Model (run with the model of choice, private or public).
Enterprise customers can use private ingest and inference enclaves. A B2C marketplace allows creators to monetize IP (70/30 split) while enterprises pay a capability premium for privacy and offline access on top of usage.
Business Model
Usage is metered by credits (self-serve subscriptions and one-time packs remain available). Enterprise customers pay a premium for private enclaves, offline capability, and custom terms on top of credit consumption. Self-serve tiers include Free, Pro, and Business plans with shared credit pools; enterprises are handled custom. This keeps the product accessible while capturing value from the privacy and offline guarantees that regulated buyers require.
Customers and Ideal Customer Profile
Celestis targets regulated and safety-critical operators whose manuals and SOPs cannot leave their control, and whose people often work where the cloud does not. Primary examples include airlines and rail, ships mid-ocean, construction and field crews repairing equipment with spotty coverage, and energy/industrial sites on locked-down or air-gapped networks. The focus is required reference material—not entertainment—that teams must consult on the job, often offline and always under their control.
Team and Credentials
Three veteran co-founders bring 80+ combined years across architecture, GTM, and technology:
-
Dom Eger (CEO & Co-Founder, 20+ years): Revenue and GTM leadership with proven velocity; deep expertise in RAG + knowledge graphs, AI at the edge, confidential cloud, Kubernetes, Kafka, and related systems.
-
David La Motta (COO & Co-Founder, 30+ years): GTM and execution background spanning sales engineering leadership and customer success across the Americas; AI/LLM/RAG expertise and prior startup and Series experience.
-
Keith Tenzer (CTO & Co-Founder, 30+ years): Product and GTM mastery scaling high-growth startups; AI infrastructure expertise (including work guiding large AI companies on agentic systems); software patent holder with deep experience in RAG, inference, Golang, Python, and cloud platforms.
The platform is live with early enterprise customers, stable inference serving users, private capsules (hosted/local/mobile), credit billing, an Apple App Store app, and a web platform. Differentiating technical elements include adaptive RAG (patent sought), hybrid cloud (Apple/NVIDIA), high accuracy and low latency, SOC 2 Type II in progress, and support for both private enterprise enclaves and a creator marketplace. The result is private offline reference at the edge—an underserved category where frontier cloud chat and hobby local agents do not compete effectively.
Ready to Ask For Funding for your company?
Post a Funding Request