The average big-box retailer has thousands of cameras. They are fixed assets, sunk costs, and largely dumb pipes for security footage that someone reviews after a theft. Dragonfruit AI starts its pitch there, with the infrastructure already bolted to the ceiling. Its bet is that the intelligence layer, not the camera itself, is where the real margin and risk reduction live for multi-site operations [dragonfruit.ai, 2024].
Founded in 2019 and based in Menlo Park, the company has built an enterprise SaaS platform that applies computer vision to existing video feeds. It promises to turn those feeds into a source of actionable intelligence for security, loss prevention, and operations, all without requiring a forklift upgrade of cameras or video management systems (VMS) [dragonfruit.ai, 2024]. With an estimated $10.7 million in annual revenue and 43 employees, the company is pitching a pragmatic, incremental path to AI for a notoriously fragmented and cost-conscious market [Growjo, 2026].
The Wedge: No New Hardware Required
Dragonfruit's core proposition is compatibility. The platform is designed to work with a retailer's or warehouse operator's existing IP and CCTV cameras, as well as their incumbent VMS, even in low-bandwidth environments [dragonfruit.ai, 2024]. This is a deliberate wedge into accounts where capital expenditure for new camera fleets is a non-starter. The company claims its total cost of ownership is the lowest in the industry precisely because it avoids expensive hardware investments [dragonfruit.ai, 2024].
The technical foundation for this is what Dragonfruit calls its patented 'Split AI' architecture. Real-time inference happens on-premise at the edge, while cloud services handle aggregation, search, and dashboard management [Qdrant, 2026]. This split aims to deliver low-latency alerts for immediate threats while still providing centralized oversight and historical analysis. For sites with no power or internet, the company even offers Dragonfruit Watchtower, a wireless, solar-powered LTE-connected surveillance system [dragonfruit.ai, 2026].
A La Carte AI for Operational Blind Spots
Instead of a monolithic suite, Dragonfruit sells its intelligence through purpose-built AI 'Agents,' available individually. The idea is to let customers deploy only what they need, targeting specific, high-cost operational blind spots [dragonfruit.ai, 2024].
- Retail Loss Prevention. Agents can monitor for self-checkout fraud, detect slip-and-fall incidents, or identify suspicious loitering.
- Warehouse Safety. Computer vision can flag unsafe behavior like improper lifting or zone violations.
- Operations & Compliance. The platform offers apps for shelf inventory management, customer traffic analytics, and ensuring compliance with safety protocols.
Each agent is designed to improve over time using real-world feedback, allowing security teams to tune alert sensitivity and adapt AI behavior to their specific risk profile [dragonfruit.ai, 2026]. The company's traction claim of being "in action at 15,000+ sites" suggests a land-and-expand motion, where a single enterprise contract can cover hundreds or thousands of individual locations [dragonfruit.ai, 2024].
The Team and Backing
The founding trio brings a blend of technical and go-to-market experience. Amit Kumar serves as CEO, Padma Duvvuri as VP of Business Development & Operations, and Shivang Agarwal as Director of Engineering [Wellfound, 2026]. While their prior exits aren't detailed in the public record, the investor syndicate provides a signal of early confidence. Backers include Unusual Ventures, Upside Partnership, Coelius Capital, and Foundation Capital, among others [Tracxn, 2026]. The company's only disclosed funding round was a seed stage in August 2019 [CB Insights, 2026].
| Role | Name | Notes |
|---|---|---|
| CEO | Amit Kumar | Founder profile listed on Crunchbase [Crunchbase, 2026]. |
| VP, Business Dev & Ops | Padma Duvvuri | Listed on company site [dragonfruit.ai, 2026]. |
| Director of Engineering | Shivang Agarwal | Listed on company site [dragonfruit.ai, 2026]. |
Where the Model Faces Pressure
The market for video analytics is crowded and competitive. Dragonfruit's stated rivals range from specialized AI firms like Scry AI and Myst AI to broader physical security platforms like Verkada, which famously bundles its own hardware [Private candid take]. The primary risk for Dragonfruit is that its wedge,working with any camera,could become a vulnerability if larger VMS or hardware vendors decide to bake similar AI capabilities directly into their own stacks, creating a bundled alternative that is simpler to procure.
The company's answer appears to be depth and flexibility. By offering a wide menu of specialized agents and a platform that claims to adapt to any customer environment, Dragonfruit is betting that best-of-breed, agnostic software will win over walled gardens [dragonfruit.ai, 2026]. Its growth to an estimated $10.7 million in revenue indicates this pitch is resonating with a segment of the market [Growjo, 2026]. The next proof point will be moving beyond the seed round from 2019 to secure growth capital that can fuel a more aggressive sales motion against entrenched competitors.
The Realistic Buyer and Competitive Set
Dragonfruit's ideal customer profile is clear: the security or operations director at a multi-location retail, logistics, or hospitality chain. This is a buyer sitting on a depreciated asset (the camera network) who is under pressure to reduce shrinkage, improve safety, and optimize operations without a massive new capex request. They have a hybrid IT/OT environment, likely multiple legacy VMS vendors across different sites, and a team that needs centralized visibility without added complexity.
The competitive landscape breaks into three tiers. First, the legacy VMS giants (like Milestone or Genetec) adding basic AI features. Second, the hardware-forward new entrants (like Verkada) selling an integrated stack. Third, the pure-play AI analytics startups (like the cited Scry AI or Daily) fighting for the same software budget. Dragonfruit's position is distinct in its agnosticism and its à la carte commercial model, but it must out-execute on integration ease and ROI clarity to own that middle ground.
For Dragonfruit, the next twelve months will be about proving it can scale its site count into enterprise-wide deployments and convert its early beachheads into seven-figure annual contracts. A Series A round would be the logical next step to professionalize sales and build out a partner channel. The bet remains pragmatic: in a world where everyone is selling new eyes, they are betting on making the old ones see.
Sources
- [dragonfruit.ai, 2024] Dragonfruit AI homepage | https://www.dragonfruit.ai/
- [Growjo, 2026] Dragonfruit AI company profile | https://growjo.com/company/Dragonfruit_AI
- [Qdrant, 2026] Dragonfruit AI partnership page | https://qdrant.tech/partners/dragonfruit-ai/
- [Wellfound, 2026] Dragonfruit AI team page | https://wellfound.com/company/dragonfruit-ai/people
- [Tracxn, 2026] Dragonfruit AI company profile | https://tracxn.com/d/companies/dragonfruit
- [CB Insights, 2026] Dragonfruit AI funding details | https://www.cbinsights.com/company/dragonfruit-ai
- [Crunchbase, 2026] Amit Kumar profile | https://www.crunchbase.com/person/amit-kumar
- [dragonfruit.ai, 2026] Shivang Agarwal profile | https://www.dragonfruit.ai/people/shivang-agarwal