Plumerai
Plumerai develops Tiny AI software that runs advanced computer vision and Vision Language Model features directly on low-cost, low-power camera chips and microcontrollers, licensing its Video Intelligence platform and Inference Engine to semiconductor manufacturers and camera OEMs including Chamberlain Group, Bosch, and Signify.
- Company typePrivate
- Founded2018
- HeadquartersLondon, United Kingdom
- Headcount11–50
- GTM typeB2B
- OfferingSoftware
What Plumerai does
Plumerai is a London-based AI software company founded in 2017/2018 that builds Tiny AI capable of running advanced computer vision and multimodal understanding on low-cost, low-power edge devices, particularly smart cameras and video doorbells. Its vertically integrated stack combines custom neural network architectures, a proprietary Inference Engine validated as the world's fastest on Arm Cortex-M microcontrollers on three consecutive MLPerf Tiny benchmarks, an in-house training pipeline (the "AI Factory") operating on a proprietary dataset exceeding 100 million images and videos, and a hybrid edge-cloud architecture in which on-device Tiny AI selectively activates cloud Vision Language Models for deeper scene understanding. The product surface spans the Plumerai Video Intelligence suite (People Detection, Familiar Face Identification, Multi-Camera Re-Identification, Advanced Motion Detection, AI Video Search, AI Captions) and the standalone Plumerai Inference Engine licensed into chip vendor SDKs.
Plumerai monetizes primarily through software licensing and royalties, selling to two distinct buyer types: semiconductor manufacturers (Renesas, STMicroelectronics, Texas Instruments, Espressif, Realtek, Ambarella, Alif, Sigmastar) that embed its Inference Engine into their SDKs and reference designs, and camera OEMs and consumer brands (Chamberlain Group's myQ and LiftMaster, Bosch, Signify/Philips Hue, ASUS, Quividi) that license the Video Intelligence software for integration into shipped devices and activation via over-the-air updates. The company also operates a free public benchmarking service as a developer-acquisition funnel and publishes the open-source Larq ecosystem for binarized neural networks. Pricing is quote-based with multi-year enterprise contracts, and revenue mechanics scale with device volume rather than per-seat SaaS metrics.
Primary customer segments are smart home camera manufacturers, enterprise security integrators, retail media and digital signage operators, consumer electronics/IoT device makers, and the semiconductor firms themselves. The company is headquartered in London with engineering offices in Amsterdam and Warsaw, has raised over $17 million in venture funding including an $8.7M Series A in September 2025 led by Partech and OTB Ventures, and is backed by individual investors Tony Fadell (Nest founder) and Dr. Hermann Hauser KBE (Arm founder). Compliance posture covers GDPR, CCPA, BIPA, and the EU AI Act for biometric and surveillance-adjacent deployments.
Plumerai firmographics
Firmographics- Name
- Plumerai
- Legal name
- Plumerai Limited
- Website
- https://plumerai.com
- Company type
- Private
- Founded year
- 2018
- Operating status
- Operating
- Headcount range
- 11–50 employees
- Short description
- Plumerai develops Tiny AI software that runs advanced computer vision and Vision Language Model features directly on low-cost, low-power camera chips and microcontrollers, licensing its Video Intelligence platform and Inference Engine to semiconductor manufacturers and camera OEMs including Chamberlain Group, Bosch, and Signify.
- Ownership category
- akta.pro rank
Plumerai industry classification
Industry- Product category
- Edge AI Video Intelligence Software
- NAICS
- Software Publishers (5132), Search, Detection, Navigation, Guidance, Aeronautical, and Nautical System and Instrument Manufacturing (334511)
- akta.pro primary industry
- Model Deployment, Serving & Inference Platforms (HDAAABAF)
Keywords
Where Plumerai is headquartered
LocationHeadquarters
- HQ city
- London
- HQ country
- United Kingdom
- HQ region
- Europe
Offices3 records
Markets served
Plumerai business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Personnel, Technology or R&D, Marketing or Sales, Infrastructure, Operations
Revenue model
- Software Licensing - Video Intelligence: Plumerai licenses its complete Plumerai Video Intelligence software suite (People Detection, Familiar Face Identification, Stranger Detection, Multi-Camera ReID, Advanced Motion Detection, AI Video Search, AI Captions) to semiconductor manufacturers and camera OEMs. The software is deployed on cameras, SOC platforms, and cloud systems across millions of devices.
- Inference Engine Licensing: The Plumerai Inference Engine is licensed as a standalone product to semiconductor manufacturers (for inclusion in their SDKs) and to AI software developers. It runs any AI model faster and with smaller memory footprint than alternatives, with no accuracy loss.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Subscription | Multi-year contract | Enterprise licensing for semiconductor manufacturers |
Go-to-market motion2 records
Distribution channels4 records
Marketing channels9 records
Plumerai product offering
Product offeringCore offering
Plumerai develops and licenses a vertically integrated Video Intelligence software platform that combines on-device Tiny AI models (running on $1 microcontrollers with 1MB memory) with selectively activated cloud-based Vision Language Models. The platform delivers people/vehicle/animal/package detection, familiar face identification, multi-camera re-identification, advanced motion detection, and natural-language video search to camera OEMs and semiconductor manufacturers. The company also sells a standalone Plumerai Inference Engine, the world's fastest deep learning engine for Arm Cortex-M microcontrollers.
Product overview
Plumerai offers a unified Video Intelligence platform for cameras and video doorbells, combining Tiny AI models that run on-device with Vision Language Models in the cloud. The core platform includes Object Detection (people, vehicles, animals, packages), Familiar Face Identification, Multi-Camera Re-Identification, Advanced Motion Detection, and VLM-powered features (Video Search, AI Captions, Custom AI Notifications). A separate standalone product, the Plumerai Inference Engine, provides the world's fastest deep learning inference for Arm Cortex-M microcontrollers. All products are deployable on-device, on-prem, in the cloud, or hybrid.
Differentiator
Problem solved
Functional benefit
Brands
- Larq: Open-source library for training binarized neural networks and ecosystem including Larq Compute Engine for BNN inference
Products and services
- Plumerai Video Intelligence Complete AI stack for consumer cameras, video doorbells, and commercial security cameras, deployed on millions of devices. Combines on-device Tiny AI with cloud-based Vision Language Models for features including detection, face identification, and natural-language video search. Sold to camera OEMs and brands under multi-year enterprise licenses.
- Plumerai People Detection (Object Detection) Tiny AI object detection model that detects people, vehicles, animals, and packages in camera video streams with a 1MB footprint. Runs on $1 chips with 1MB memory at 13.6 FPS on Arm Cortex-M85 and 3.3 FPS on ESP32-S3. Licensed to camera OEMs and semiconductor companies.
- Plumerai Familiar Face Identification (FFID) On-device face recognition with automatic and manual enrollment that identifies known individuals, distinguishes strangers, and supports up to 40 familiar faces. Runs entirely on-device for privacy protection; 4 FPS on Cortex-M85. Licensed to camera OEMs and security camera companies.
- Plumerai Multi-Camera Re-Identification On-device person re-identification that groups events from the same individual across multiple cameras using non-facial appearance, reducing notification fatigue by up to 90% (e.g., 34 gardener alerts collapsed to 1). Sold to camera OEMs and security system integrators.
- Plumerai Advanced Motion Detection On-device motion detection that filters environmental noise (rain, snow, wind, lighting, dust) and produces spatial motion maps to guide downstream AI. Uses only 243KB RAM at 640x480 and 693KB at 1280x720; runs at 10 FPS on Cortex-M3. Sold to camera OEMs.
- Plumerai Video Search (AI Video Search) Vision Language Model-powered natural-language search across camera video history. Lets users query video archives with sentences like 'person in blue jacket walking a dog' and get results in seconds. Cloud costs are up to 135x lower than Amazon Nova and Google Gemini per customer benchmarks. Licensed to camera OEMs and security providers.
- Plumerai AI Captions Cloud-based Vision Language Model feature that produces plain-English descriptions of video clips for richer notifications. Activated selectively by on-device Tiny AI only for moments needing deeper analysis. Sold as part of the Video Intelligence platform to camera OEMs.
- Plumerai Inference Engine Standalone deep learning inference engine for Arm Cortex-M and other microcontrollers, offering 40% lower latency and 49% less RAM than TensorFlow Lite for Microcontrollers. Runs any AI model without quantization, binarization, pruning, or compression, with no accuracy loss. Sold under multi-year licenses to semiconductor manufacturers for inclusion in SDKs and to AI software developers.
Quantifiable outcome
- Cloud costs up to 135x lower than Amazon Nova and Google Gemini for AI Video Search (per customer benchmarks)
- +7 more outcomes
Companies that use Plumerai
Customer profileNamed customers8 records
Segments6 records
Ideal customer profiles4 records
Plumerai technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
AI capability9 records
Feature10 records
Plumerai partnerships and signals
Strategic signalPartnerships
Eight partnerships are on record, tiered core, flagship / core customer, core / flagship, core / customer and strategic.
- Realtek SemiconductorcorePlumerai Video Intelligence demo on the Realtek Ameba Pro2 (RTL8735B) chip. People Detection and Familiar Face Identification run at 7-10 FPS on 1920x1080 video input within only 6MB of flash, leveraging the Ameba Pro2 NPU. The demo can be requested from Plumerai and targets the AMB82 MINI board.
- Chamberlain Groupflagship / core customerMajor partnership announced October 2024. Chamberlain Group (brands include myQ and LiftMaster) integrated Plumerai's Tiny AI into its smart camera lineup. Products are found in 50+ million homes and 11+ million people use the myQ app daily. Chamberlain Group's VP AI & Innovation stated the integration enables advanced AI features on affordable, low-power devices. The partnership represents Plumerai's flagship commercial deployment.
- ArmcoreArm is referenced as a technology partnership. Plumerai's inference engine is optimized for Arm Cortex-M, Cortex-A, and Helium vector extensions. Dr. Hermann Hauser, Arm's founder, is an investor in Plumerai. Plumerai's technology is validated on Arm reference platforms in MLPerf Tiny benchmarks.
- Renesas Electronicscore / flagshipRenesas and Plumerai jointly demonstrated AI on the new Renesas RA8D1 MCU (Arm Cortex-M85) at CES 2023. Plumerai People Detection runs at 13.6 FPS and Familiar Face Identification at 4 FPS on the RA8D1's Arm Cortex-M85. The RA8D1 features Helium vector extensions for AI acceleration. Renesas includes Plumerai's Inference Engine in their ecosystem.
- Espressif SystemscorePlumerai People Detection AI is available on Espressif's ESP32-S3 microcontroller. Running at 3.3 FPS on a single Xtensa LX7 core at 240MHz with 166KB peak RAM. Enables smart home, smart building, smart city, and smart health applications. The demo runs on the ESP32-S3-EYE board.
- Quividicore / customerQuividi adopted Plumerai's people detection for its audience measurement platform. Quividi is a world leader in measuring audiences for digital displays and retail media, with over 600 customers analyzing billions of shoppers monthly across tens of thousands of screens. The integration enables anonymous, privacy-compliant camera-based detection.
- Texas InstrumentscorePlumerai partnered with Texas Instruments to enable AI applications on TI's SimpleLink Wi-Fi CC3220SF MCU (Arm Cortex-M4). People detection runs at 80MHz with peak RAM usage of 170KB, program binary of 154KB, and demonstrated controlling lights for building automation. Product was shown at CES 2022.
- XMOSstrategicStrategic partnership announced April 2020 combining Plumerai's Larq software library for training BNNs with XMOS's xcore.ai crossover processor which provides native BNN inference support. The combined solution delivers 2-4x more efficiency than existing edge AI solutions, enabling a new generation of low-power, low-cost AI-enabled devices.
Scale indicators12 records
Recent moves6 records
Expansion highlights6 records
Plumerai competitors and assessment
Company assessmentDirect peers
- Syntiant: Syntiant builds ultra-low-power neural decision processors and TinyML models for always-on edge AI on microcontrollers and battery-powered devices. Directly comparable to Plumerai's Tiny AI for Cortex-M class hardware targeting smart home, IoT, and camera-adjacent applications.
- BrainChip Holdings: BrainChip offers the Akida neuromorphic IP for ultra-low-power on-device AI in vision and audio. Both companies sell neural-network IP optimized for sub-dollar, low-memory edge silicon aimed at similar smart sensor and camera workloads.
- Edge Impulse: Edge Impulse provides a TinyML development platform and EON inference compiler for microcontrollers. Plumerai explicitly benchmarks against Edge Impulse EON on MLPerf Tiny, making them direct inference-stack rivals for embedded AI.
- Deeplite: Deeplite provides model optimization (quantization, pruning, NAS) for deploying deep learning on resource-constrained edge devices. It competes with Plumerai's combined tiny-model + optimized inference engine approach for tiny AI workloads.
- OctoML: OctoML automates ML model deployment and acceleration across hardware targets via Apache TVM. Plumerai explicitly competes on MLPerf Tiny against OctoML for edge inference on microcontrollers.
- SensiML: SensiML offers an end-to-end TinyML toolchain for collecting data, training, and deploying AI on microcontrollers. Comparable to Plumerai's AI Factory plus inference engine for smart home, sensor, and industrial edge AI.
Broad incumbents
- Ambarella: Ambarella is an established AI vision SoC vendor used in security cameras and edge-AI devices. Plumerai runs ON Ambarella chips and competes indirectly through offering an alternative, chip-agnostic AI software layer.
- Qualcomm (AI Engine / IoT): Qualcomm's broader edge AI stack (Neural Processing SDK, AI Engine on Snapdragon) targets IoT and camera workloads. Comparable as a broad incumbent offering on-device AI, but lacks Plumerai's sub-$1 microcontroller focus.
- Himax Technologies: Himax designs AI-enabled imaging SoCs and WiseEye ultra-low-power edge AI for smart cameras, doorbells, and IoT sensors. Comparable in targeting AI on battery-powered edge devices with integrated vision pipelines.
- Google (Google Nest / Edge TPU): Google Nest cameras and Coral Edge TPU represent an incumbent smart-camera + edge-AI platform that Plumerai explicitly competes against and has reportedly won commercial shootouts against.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat7 records
Key risks6 records
Key highlights7 records
Customer concentration
Plumerai social profiles
Digital presencePlumerai compliance and trust
Trust signalCompliance4 records
Plumerai financial estimates
Financial estimateRevenue estimate
Valuation estimate
Plumerai leadership team
Management profileNumber of profiles
Profiles6 records
Plumerai funding detail
Funding detailFunding overview
Funding rounds2 records
Investors1 record
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Plumerai M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Plumerai
What does Plumerai do?
Plumerai develops and licenses a vertically integrated Video Intelligence software platform that combines on-device Tiny AI models (running on $1 microcontrollers with 1MB memory) with selectively activated cloud-based Vision Language Models. The platform delivers people/vehicle/animal/package detection, familiar face identification, multi-camera re-identification, advanced motion detection, and natural-language video search to camera OEMs and semiconductor manufacturers. The company also sells a standalone Plumerai Inference Engine, the world's fastest deep learning engine for Arm Cortex-M microcontrollers.
Is Plumerai a public or private company?
Plumerai is a private company. It is classified as venture growth investor backed and is currently operating.
When was Plumerai founded?
Plumerai was founded in 2018. It employs 11 to 50 people.
Where is Plumerai based?
Plumerai is headquartered in London, United Kingdom, in the Europe region.
How does Plumerai make money?
Two revenue lines are on record. Software Licensing - Video Intelligence is the primary driver. The others are inference Engine Licensing.
Who are Plumerai's main competitors?
Direct peers on record are Syntiant, BrainChip Holdings, Edge Impulse, Deeplite, OctoML and SensiML. Broad incumbents are Ambarella, Qualcomm (AI Engine / IoT), Himax Technologies and Google (Google Nest / Edge TPU).
Does Plumerai have an API?
Yes. Plumerai Video Intelligence API available in C++, Python, Rust, and Java for integrating AI features (Object Detection, Familiar Face Identification, VLM Video, VLM Text Search) into camera applications. Supports cross-platform deployment on-device, on-prem, in the cloud, or hybrid. Developer documentation is at docs.plumerai.com.
What industry is Plumerai in?
Plumerai's product category is Edge AI Video Intelligence Software. Its primary akta.pro industry code is HDAAABAF, Model Deployment, Serving & Inference Platforms. Its NAICS code is 5132.