Planet AI: IDP Vendor Profile

German deep learning specialist acquired by Bechtle AG in 2023, bringing patented PerceptionMatrix IDP technology to Europe's largest IT system house.

Overview

Planet AI is a Rostock-based IDP vendor founded in 2015 as a spin-off from Planet intelligent systems, which dates to 1992, a lineage the company uses to position itself as Europe's oldest deep learning specialist. Its core differentiator is data sovereignty: both products deploy on-premises, private cloud, or hybrid with no US cloud dependency, a concrete architectural choice targeting buyers in insurance, banking, public sector, and logistics operating under strict EU data residency requirements.

In October 2023, Bechtle AG acquired a 51% stake through a competitive international bidding process involving investors from Europe, North America, and Asia, with a call option for the remaining 49%. The deal brought Planet AI's €3.1 million revenue and 41-person R&D team into Bechtle's €6.03 billion enterprise ecosystem. CEO Jesper Kleinjohann framed the constraint plainly before the deal closed: "We knew that our developments are top notch but we were short a partner who could accelerate our growth with broad market access and is able to deploy our solutions at scale." Bechtle provided both. The team has since grown to approximately 60 people.

Two major product releases followed in 2025: IDA 5.3 in April, adding LLM-based inference to the core digitization platform, and JAIDE in November, a multi-agent knowledge management assistant built on top of IDA's structured output. Both releases extend the same on-premises, no-US-cloud architecture that defines Planet AI's market position.

The company holds ISO 9001 certification and markets two go-to-market tracks: OEM white-label licensing for software vendors who embed Planet AI technology under their own branding, and a co-creation model for IT consulting firms involving joint project teams and knowledge transfer. Both tracks position "Made in Germany" as a tender differentiator, particularly in public sector procurement.

How Planet AI processes documents

Planet AI's processing architecture centers on its patented PerceptionMatrix neural network, which the company traces to 1992. The defining characteristic is a "no-rules approach": the system requires minimal training data to begin classifying and extracting from new document types, contrasting with rule-based competitors that demand extensive template configuration before deployment.

The platform ships as two distinct products. IDA handles high-volume batch document digitization, including OCR and ICR for machine-written and handwritten text, automated document classification without predefined rules, and data extraction across structured and unstructured formats including receipts and legacy archives. JAIDE sits on top of IDA as an employee-facing interface, surfacing IDA-extracted knowledge through natural language queries for internal knowledge access rather than batch processing workflows.

IDA 5.3 and LLM-based inference

The April 2025 release of IDA 5.3 introduced IDA Understanding, an LLM-based module that executes recurring prompts in batch mode and writes answers directly to document metadata. This is a meaningful architectural distinction: rather than requiring a human to query documents through a chat interface, IDA Understanding runs inference at processing time and embeds results into the document record for downstream systems to consume. Kleinjohann described the use case directly: "IDA Understanding is especially useful when information is not explicitly stated in the document and needs to be inferred from context. Unlike a chat application, these insights are produced through mass processing and can be integrated into subsequent processes, such as enterprise content management."

IDA 5.3 also introduced a layout-agnostic invoice module, eliminating the need to train separate extraction models per invoice template per customer. A redesigned IDA Client added hot folder monitoring for automated batch processing workflows.

JAIDE and multi-agent orchestration

JAIDE, launched November 3, 2025, is an AI assistant for corporate knowledge management built on a specialized agent architecture. Three separate agents handle distinct responsibilities: one processes and understands incoming documents, a second manages business process logic, and a third connects external tools and domain-specific sources. The system automatically routes each task to the appropriate model rather than passing all queries through a single generic model.

JAIDE operates on IDA's structured, clean output rather than raw document scans, which Planet AI argues produces more reliable answers than retrieval-augmented generation (RAG) systems built on unprocessed text. Kleinjohann stated: "JAIDE is not just another chatbot. This orchestration delivers far more accurate results than a single generic model." Three primary use cases are documented: AI-powered knowledge management for technical documentation, tender management analysis, and administrative process support. The platform ships with pre-built, industry-specific workflow templates and supports model-agnostic LLM integration.

Both products run on-premises, private cloud, or hybrid. Planet AI explicitly advertises no US cloud dependency, meaning the inference pipeline does not route through AWS, Azure, or GCP infrastructure. The company also describes multi-model flexibility, allowing customers to switch between underlying AI providers without lock-in, though specific model integrations and version numbers are not disclosed publicly. Edge deployment is supported for environments with constrained hardware, extending the platform beyond data center deployments into field or branch office scenarios.

Use cases

Banking and financial services

IDA's no-rules classification and high-accuracy extraction address the precision requirements of institutions with fiduciary obligations. The on-premises deployment model directly answers data residency concerns that prevent many European banks from adopting cloud-native IDP platforms. Planet AI competes here against ABBYY and Tungsten Automation, both of which offer broader feature sets but depend on cloud infrastructure that some regulated buyers cannot use.

Insurance and public sector

The combination of ISO 9001 certification, EU data residency architecture, and "Made in Germany" positioning targets insurance carriers and government agencies where procurement criteria explicitly favor sovereign infrastructure. Kleinjohann has stated: "This reliability is crucial, especially in regulated industries such as public administration." The OEM and co-creation partner tracks mean Planet AI often reaches these buyers through system integrators rather than direct sales, a distribution model suited to public sector procurement cycles.

Healthcare document management

IDA's handwriting recognition handles medical records, insurance claims, and patient documentation where legacy paper workflows remain common. The private cloud deployment option addresses GDPR obligations for patient data without requiring buyers to negotiate data processing agreements with US hyperscalers.

Historical archive digitization

Minimal training data requirements make IDA practical for organizations digitizing heterogeneous legacy archives: mixed document types, variable quality, inconsistent formatting. Traditional document classification systems require prohibitive setup effort for such collections. Libraries, notarial archives, and public registries represent natural buyers given the sovereign infrastructure positioning.

Knowledge management and tender analysis

JAIDE's three documented use cases, technical documentation management, tender analysis, and administrative process support, target knowledge workers in regulated industries who need answers grounded in organizational documents rather than general LLM knowledge. The tender management use case is particularly relevant for public sector buyers who already use IDA for document digitization and want to extend that investment into procurement workflows.

Accuracy claims and independent verification

GetApp's March 2026 review cites 99% machine-print and 95% handwriting accuracy for IDA. These figures originate from Planet AI's own reporting rather than independent benchmarking. Planet AI has published an OCR benchmark white paper comparing IDA against open-source engines (easyOCR, MMOCR, PaddleOCR, Tesseract), commercial cloud services (Amazon, Azure, Google), and multimodal LLM-based engines (GPT-4o, Mistral OCR, Gemini 2.0 Flash). The white paper does not publish quantified performance metrics in its publicly available form, so the comparison methodology cannot be independently assessed. Buyers requiring verified accuracy benchmarks should request test results on representative document samples before procurement.

Technical specifications

SPOC-AI research project

In June 2025, Planet AI secured €11 million in European Regional Development Fund (ERDF) support for the SPOC-AI (Single Point of Contact) project, in collaboration with ALTOW Digital Innovation and the University of Rostock. The project focuses on developing a document hub for intelligent document processing, with scope extending into medical technology, environmental sciences, and real-time image analysis. Whether SPOC-AI research translates into commercial product features remains unconfirmed. The collaboration positions Planet AI within European AI research infrastructure and provides access to academic talent at a time when the broader IDP market, tracked at 456 companies by Deep Analysis in July 2025, is consolidating around vendors with differentiated technical foundations.

Resources

• Planet AI website
• IDA 5.3 release announcement
• JAIDE product launch
• JAIDE product page
• Planet AI OCR benchmark white paper
• Bechtle AI solutions
• SPOC-AI research project announcement

Company information

• Parent company: Bechtle AG
• Headquarters (operations): Rostock, Germany
• Bechtle headquarters: Neckarsulm, Germany
• Founded: 2015 (spin-off from Planet intelligent systems, est. 1992)
• Employees: approximately 60 (as of 2025)
• Bechtle stake: 51% (acquired October 2023, call option on remaining 49%)