Assemblio's EIC Accelerator Agent for Planning Factory Assembly

Before a machine can be built on an assembly line, a planner has to work out in which order its parts go together and from which direction, and write the instructions the workers follow. Fraunhofer IPA, the Stuttgart institute Assemblio came out of, writes that this planning can take up to 50% of personnel costs and about 20% of total manufacturing costs. Assemblio's software reads the CAD model of a product and proposes the assembly sequence itself.

The EIC Accelerator selected Assemblio's project AI-APOA at the 1 October 2025 cut-off and recommended it for blended finance, a grant plus an equity investment. CORDIS records an EU contribution of €2,499,999 for the grant, signed on 28 May 2026, for a project that runs from 1 June 2026 to 31 May 2028. Five months in, there is nothing yet to measure against the plan, so this article sets out what was promised and what can already be checked.

Planning by disassembly, done by software

The conventional method, as Fraunhofer IPA describes it, is "assembly by disassembly": a real product is taken apart step by step to record how it goes together. Assemblio's software does this on the 3D model instead. According to the Motek trade fair profile of May 2024, an AI analyses the CAD assembly and determines the sequence, the type and direction of each step and its restrictions; planners correct the result by drag and drop in the Composer and export instructions as PDF or video, or share them as interactive 3D steps in the Smart Viewer. Fraunhofer IPA says the analysis works on STEP files, the neutral export format of any CAD system.

The company's figures are large. The Motek profile quotes a product of 800 parts whose instructions took about 100 hours by hand and 3 hours with the software, "a factor of 32", and Fraunhofer IPA cites user studies showing 92% less time to create instructions. Neither source names the product or publishes the study, so these are the company's and its parent institute's claims, not independent results.

Named customers, unpublished revenue

Assemblio names its users in public. Its seed announcement of 28 February 2024 named Krones and Bosch and quoted Mathias Rausch, head of module assembly at Krones; Startbase reported the same two customers. The Motek profile lists the WEISS Group, Siemens, the Karlsruhe Institute of Technology and the Fraunhofer-Gesellschaft. The company's website, checked on 4 November 2026, shows logos of Siemens, Voith, Daimler Truck, Krones, RAUCH, Erne, arculus and WEISS, and its ISO 27001 release of February 2025 quotes a product manager at RAUCH Landmaschinenfabrik.

What a logo or a quote means in money is not public. The Implisense company profile records the company as active with a share capital of €36,230 and shows no public annual statements, and no revenue, licence count or staff number appears in the company's releases. Whether these are paying production deployments or pilots cannot be verified from the sources reviewed.

What AI-APOA promises

The CORDIS objective widens the product from instructions to the link between design and production. Assemblio says product data sits scattered across PDM, PLM and ERP systems, that this causes errors and delays, and that assembly experience rarely flows back into design. AI-APOA, which it calls an Assembly Planning and Optimisation Agent, is meant to be a digital twin of the design process: it would pull data from CAD, PDM, PLM and ERP into one model, combine the design, engineering and manufacturing bills of materials, allow changes in real time and generate interactive 3D instructions. The objective also promises tracking of data for recycling and resource recovery, shorter training for workers and automated documentation. It gives no numeric target for any of these.

The grant sits under the 2025 challenge GenAI4EU, "Creating European Champions in Generative AI", and CORDIS tags 40% of the funding to the digital agenda and 40% to artificial intelligence. Close to the project's theme is the AutoSequencer, which the company's SPS announcement of 2 November 2026 says derives assembly sequences automatically from CAD data and keeps them current when the design changes; it will be shown at SPS in Nuremberg from 24 to 26 November 2026, with two customer talks whose speakers are not yet named. The page does not tie the AutoSequencer to the EIC project.

1. Assembly sequences generated from CADProduct existsComposer and AutoSequencer, per the Motek profile and the company; the Composer was on the market before the project began.
2. One model across CAD, PDM, PLM and ERPToo earlyNo integration has been announced since the project began in June 2026.
3. Assembly feedback into designToo earlyDescribed in the CORDIS objective only.
4. Recycling and resource recovery dataToo earlyNo public product feature yet.

The money: a grant signed, equity not announced

The EIC reported 923 full applications for the October 2025 cut-off, asking for almost €7 billion. On 17 February 2026 it announced 61 selected companies out of 121 interviewed, with Germany among the countries with the most; the selection list has nine German entries, Assemblio one of them. Assemblio announced on 13 April 2026 that it had received the European STEP Seal and "will receive €2.5 million in funding from the European Union", without naming the programme.

The equity half of the blended finance is not visible. The EIC says companies selected for equity go to the EIC Fund for an investment decision, and the EIC Fund portfolio, which listed 319 organisations on 4 November 2026, does not include Assemblio. That is a fact about the list, not evidence that no investment will follow. The only private round on record is the €2.1 million seed of February 2024, led by LEA Partners, reported by the company and by Startbase.

Why the EU funds it

The EIC placed the project in its generative AI challenge, and the company says the STEP Seal recognises projects that support Europe's technological sovereignty and industrial value creation. The project says scattered product data costs up to 50% of production time and adds 20% to production costs; whether AI-APOA reduces either is for its results, due after May 2028, to show.

Assemblio was not asked for comment. The reporting is based on publicly available sources, listed below.