microagi
About
About microagi
Atlas captures manual factory work through dedicated sensing hardware and converts it into task demonstrations for robot training. microagi combines those demonstrations with simulation and model adaptation, fine-tuning control policies around the customer’s actual objects, workstations and operating sequence.
shift expands the data layer through first-person recordings of physical work collected across multiple countries. Together the systems connect broad human-action data with site-specific industrial demonstrations, giving robotics developers a route from general physical understanding to a deployed policy for a particular task.
Articles
Funding
- 2026-07-16 - Seed - $55M - microagi raises $55M seed round led by Hummingbird.
microagi raised a $55 million seed round to fund robotics-model compute, expand shift's data collection network and grow its U.S. operations. - source: businessinsider.com
Investors
- Hummingbird - Lead - Unknown - 1 funding event - first seen 2026-07-16
- LocalGlobe - Participant - Corporate / CVC - 1 funding event - strategic/corporate - first seen 2026-07-16
- Northzone - Participant - Unknown - 1 funding event - first seen 2026-07-16
- redalpine - Participant - Unknown - 1 funding event - first seen 2026-07-16
- Village Global - Participant - VC - 1 funding event - first seen 2026-07-16
Team
- Zeno Hamers - Managing Director, MicroAGI CH - Leads the Zürich research headquarters and Swiss operating entity.
- Bercan Kilic - Co-founder and CEO - Leads microagi's industrial robot deployment and embodied-AI strategy.
- Yoan Iliev - Co-founder - Co-founded microagi and contributes high-performance engineering experience.
- Anton Poletaev - Co-founder - Co-founded microagi and contributes AI research experience.
- Nico Nussbaum - Co-founder - Co-founded microagi and contributes engineering experience.
- Artjem Weissbeck - Co-founder - Co-founded microagi and brings prior company-building experience as a serial entrepreneur.
Research
- microagi releases MicroPlanBench next-action benchmark
2026-10-14 - Research Publication - Each model receives eight seconds of 720p video at ten frames per second, with five attempts per clip. GPT-6 Astra reached a 65.6% mean pass rate under the company’s LLM-judged protocol; the judge checks actions, objects, destinations and hand assignment. The evaluation set totals just under 27 minutes; human reviewers selected cases with one reasonable immediate next action.
Source
Timeline
Events
- 2026-10-14 - Research Publication - microagi releases MicroPlanBench next-action benchmark
Each model receives eight seconds of 720p video at ten frames per second, with five attempts per clip. GPT-6 Astra reached a 65.6% mean pass rate under the company’s LLM-judged protocol; the judge checks actions, objects, destinations and hand assignment. The evaluation set totals just under 27 minutes; human reviewers selected cases with one reasonable immediate next action.
Source - 2026-07-16 - Funding - microagi raises $55M seed round led by Hummingbird.
Hummingbird led the round, with Northzone, LocalGlobe, Village Global and redalpine participating. microagi said the capital will support model-training compute, expansion of shift and growth from New York into the U.S. Business Insider reported that five companies were collecting factory data through microagi, with one preparing a robot deployment.
Source - 2026-06-03 - Market Signal - microagi launches physical-AI research fellowship
The fellowship offers each selected researcher up to $2 million in accredited compute and access to large-scale human-task datasets. Fellows receive access to microagi's robotics hardware, proprietary infrastructure and partner network. The six-month research stay is based at microagi's laboratories in Zürich or Munich and includes a $10,000 monthly stipend.
Source - 2026-05-28 - Product Launch - microagi launches shift home-cleaning data service in New York
Vetted cleaners wear head-mounted cameras while completing household tasks, and shift anonymises sensitive information before the data is used. The service launched publicly on May 28 and received thousands of booking requests after its initial announcement. shift uses the recordings as physical-work training data for AI labs and robotics companies.
Source - 2026-05-07 - Facility Opening - microagi opens Global Robotics Research HQ in Zürich
microagi announced that it is opening its Global Robotics Research Headquarters at Bahnhofstrasse 63 in Zürich. The company said Zürich will become the center of its global robotics research programs after growing from five people in Munich to operations across 15+ countries. microagi said the research center is focused on bridging the data gap between robotics research and real-world deployment.
Source - 2026-04-30 - Partnership - microagi partners with the European Student Robotics Association
ESRA described microagi as both a financial supporter and a hands-on partner. The partnership supports ESRA's work to build a pan-European student robotics network. ESRA said the collaboration is intended to advance robotics research and education across Europe.
Source - 2026-03-26 - Company Formation - MicroAGI CH GmbH enters the Zürich commercial register
MicroAGI CH GmbH was entered in the Zürich commercial register with legal headquarters in Zürich. The registered company purpose covers collecting, processing and selling video data and digital assets for artificial intelligence and robotics applications. The Swiss entity lists MicroAGI GmbH in Düsseldorf as shareholder and Zeno Hamers as managing director.
Source
Notes
Profile notes
- microagi uses human service work to build physical-AI training data
reviewed 2026-07-18 - microagi's shift service uses first-person capture equipment to collect household-task demonstrations through human workers. Its Atlas platform applies the same data-and-deployment model in industrial settings, where operational task data is used to prepare and refine robotic systems.
Source
Relations
Relationships
- European Student Robotics Association
Student Robotics Partner - microagi supports ESRA as a financial and hands-on partner building a pan-European student robotics network.
Context
Comparable profiles
- micro1
stack adjacency - Physical AI Data Infrastructure
- Xspark AI
stack adjacency - Physical AI Data Infrastructure
- Vision Lab
stack adjacency - Embodied Data Capture · Physical AI Data Infrastructure
- Maniformer
stack adjacency - Embodied Data Capture
- Jiiov / 极豪科技
stack adjacency - Embodied Data Capture
- 6thSense
stack adjacency - Physical AI Data Infrastructure
- Sureform
stack adjacency - Embodied Data Capture · Physical AI Data Infrastructure
- Physen AI
stack adjacency - Embodied Data Capture · Physical AI Data Infrastructure
- Mecka AI
stack adjacency - Embodied Data Capture · Physical AI Data Infrastructure
- MANUS
stack adjacency - Embodied Data Capture · Physical AI Data Infrastructure
- XDOF
stack adjacency - Embodied Data Capture · Physical AI Data Infrastructure
- Lightwheel
stack adjacency - Physical AI Data Infrastructure
Shared workflows
- AI2 Robotics
workflow adjacency - Factory Robot Deployment
- Leju Robotics
workflow adjacency - Factory Robot Deployment
- Hexagon
workflow adjacency - Factory Robot Deployment
- Renault Group
workflow adjacency - Factory Robot Deployment
- HARIBIT
workflow adjacency - Factory Robot Deployment
- Sichuan Tianlian Robotics Co., Ltd.
workflow adjacency - Factory Robot Deployment
- Tianyuanxing (天元兴)
workflow adjacency - Factory Robot Deployment
- Mind Robotics
workflow adjacency - Factory Robot Deployment
- Lenovo
workflow adjacency - Factory Robot Deployment
- Nucleus Robotics
workflow adjacency - Factory Robot Deployment
Context tags
- Embodied AI - Capability - microagi develops data and deployment infrastructure for embodied intelligence.
- Physical AI data infrastructure - Stack Layer - shift collects and prepares real-world human-task data for robotics training.
- Embodied data capture - Stack Layer - Atlas and shift capture physical work in factories and everyday environments.
- Factory robot deployment - Workflow - Atlas supports task-specific model training and factory robot deployment.

