Teklia

What Teklia does

Teklia is a French AI software company founded in 2014 by Christopher Kermorvant, specializing in document recognition and processing for cultural heritage, archival, and public-administration clients. The company builds deep-learning models for handwriting recognition (HTR), printed-text OCR, layout analysis, table extraction, named-entity recognition, media cataloging, and content-based image retrieval, applied to historical corpora at industrial scale.

The technological core is Arkindex, an open-source web-based platform that orchestrates complex pipelines of OCR/HTR, segmentation, classification, and NER models across millions of pages. Arkindex integrates with the IIIF standard for image interoperability, exposes a documented REST API and Python SDK, and supports Docker-based custom-model insertion. Teklia also develops Callico (collaborative annotation), PyLaia (HTR library), Doc-UFCN (document segmentation), Nerval (NER evaluation), and specialized platforms such as HikarIA (Japanese photographic heritage) and SocFace (French census database, targeting 400M records). Proprietary and open-source libraries are published on GitLab, PyPI, and HuggingFace.

Commercially, Teklia operates a hybrid model combining multi-year professional services (person/month and volume-based pricing) with a three-tier Arkindex subscription (Community free, Academia €1,500/month, Enterprise €4,000/month), managed infrastructure hosting, and custom model training. Customers span GLAM institutions (French National Archives, Library of Congress, Musée du Louvre, Musée Guimet, National Library of Norway), public administration (OFPRA, Service Historique de la Défense), research bodies (INED, Sorbonne, UQAC), humanitarian organizations (ICRC, Mémorial de la Shoah), and corporate heritage (Hennessy/LVMH). Geographic reach covers France, Norway, Switzerland, the USA, Canada, and Egypt. Headcount is approximately 12 core members, including three PhD holders.

Name

Teklia

Legal name

TEKLIA SAS

Website

https://teklia.com

Company type

Private

Founded year

2014

Operating status

Operating

Headcount range

11–50 employees

Short description

Teklia is a French AI software company that builds deep-learning document recognition and processing tools — anchored by its open-source Arkindex platform — for cultural heritage, archival, public-administration, and research institutions.

Ownership category

akta.pro rank

Product category

AI Document Processing Software

NAICS

Libraries and Archives (519210), Libraries and Archives (51921), Document Preparation Services (56141), Document Preparation Services (561410), Museums, Historical Sites, and Similar Institutions (712)

SIC

Services-Computer Programming, Data Processing, Etc. (7370), Services-Computer Processing & Data Preparation (7374), Services-Computer Programming Services (7371)

akta.pro primary industry

Content Governance, Compliance, Retention & eDiscovery (HDAEAGAJ)

akta.pro secondary industries

Records Digitization & Archiving (Legacy Files, Retention Support) (BPAAAKAG), Content Identification & Matching (Fingerprinting, ID Resolution) (MPAHALAE), Document Scanning, Digitization & Archival Reproduction (Records, Transcripts, Microform) (EDAEAJAI)

Keywords

• Heritage document processing
• Handwriting recognition software
• Document AI platform
• Cultural heritage AI
• Automated text recognition

Headquarters

HQ city

Raymond

HQ country

France

HQ region

Europe

Offices3 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Infrastructure, Operations, Marketing or Sales

Revenue model

1. Arkindex Software Licenses: Arkindex offers three editions: Community (free, open-source, self-hosted), Academia (1500€ per month, self-hosting or TEKLIA hosting, advanced ACL, HPC integrated), and Enterprise (4000€ per month, advanced features, support, integration with HPC clusters). All editions include unlimited document capacity and users.
2. Professional Services: Consulting, technical advisory, and full-service document processing projects. Services include early-stage project advisory, AI configuration, training and consulting support. Pricing based on person/month and volume to be processed.
3. Infrastructure Services: On-premises managed by TEKLIA, dedicated instance hosted by TEKLIA. Prices depending on maintenance options and volume to be hosted.
4. Custom Model Training: Custom models trained by TEKLIA for specific document and processing needs. Includes fine-tuning of generic models using project-specific data to improve performance.

Pricing tiers

Go-to-market motion2 records

Distribution channels4 records

Marketing channels5 records

Core offering

Teklia develops AI-powered document processing solutions for cultural heritage, archives, and public administration clients. Its flagship product, the open-source Arkindex platform, orchestrates industrial-scale OCR/HTR, layout analysis, named entity recognition, table extraction, and content-based image retrieval across millions of historical and heritage pages. The company complements the platform with professional services, custom model training, and managed/hosted infrastructure on European (France/Germany) GDPR-compliant infrastructure.

Product overview

TEKLIA offers a unified platform-plus-services architecture centered on Arkindex, their open-source document processing platform, complemented by the Callico annotation platform. The product portfolio includes specialized deep learning libraries (Doc-UFCN, PyLaia, Nerval), CLI and API client tools for developers, and two major research platforms (HikarIA for Japanese photography analysis, SocFace for French census records). Services encompass six AI-powered document processing capabilities: Automatic Text Recognition (OCR/HTR), Document Layout Analysis, Table and Data Recognition, Named-Entities Recognition, Media Cataloging, and Image Search by Content (CBIR). All offerings are designed for heritage, archival, and cultural institution use cases, with open-source software at the core and commercial processing services built on top.

Differentiator

Problem solved

Functional benefit

Products and services

• Arkindex Platform Open-source, web-based platform for orchestrating large-scale AI processing of scanned or digitized documents (OCR/HTR, layout analysis, NER, classification, entity extraction). Supports millions of pages, unlimited users, and self-hosting, with editions tailored to community, academic, and enterprise use.
• Callico Open-source annotation and validation platform for digitized documents used to generate training data for Deep Learning models, with collaborative annotation campaigns, form-based data entry, and Character Error Rate evaluation metrics.
• Astrae Platform (Early Adopters Program) Multimodal AI software platform for heritage collection exploration, combining image analysis, text recognition, and semantic understanding across multiple data types, currently offered through an Early Adopters Program.
• SocFace Digital database compiling a century of French censuses (1836–1936), built through AI extraction of demographic data from historical census documents.
• HikarIA Specialized AI-powered platform for discovering Japanese photographs from the Guimet Museum's collection, featuring content-based search, visual similarity detection, semantic proximity search, and geolocation across 280 albums of historical photographs.
• Professional Services (Document Processing & Consulting) Full-package document processing, consulting, training, and technical advisory delivered by Teklia on cultural heritage, archival, and research projects. Includes early-stage project advisory, AI configuration, full-packaged processing under client supervision, and pricing based on person/month and volume.
• Custom Model Training Training of custom AI models (e.g., HTR, layout, NER) for specific document types, including fine-tuning of generic Teklia models using client-specific data.
• Managed Infrastructure / Hosting On-premises managed or Teklia-hosted dedicated Arkindex infrastructure, with pricing depending on maintenance options and hosted data volume, hosted in Europe (France/Germany) for GDPR compliance.

Named customers15 records

Segments4 records

Ideal customer profiles4 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration1 record

AI capability10 records

Feature8 records

Partnerships

Eight partnerships are on record, tiered core and minor.

• FINLAM ConsortiumcoreStrategic or Co-development Partner · 1 January 2026Research project on use of AI for processing BNF (French National Library) collections, particularly automatic structuring of historical press articles. Partners include LITIS and National Library of France.
• HAIKIDOminorStrategic or Co-development Partner · 1 January 2026Research project for development of specialized models combining tabular structure and multimodal approaches. Partners include INED, Paris Observatory, and École des Ponts.
• Paris School of EconomicscoreStrategic or Co-development Partner · 1 January 2023Research partner in SocFace project for demographic analysis of French census data.
• DAI-CréTDHI ConsortiumminorStrategic or Co-development Partner · 1 January 2023Research project on demography and anthroponymy of French population from 16th to 19th century through nominative tracking of individuals.
• Østfold University College (Norway)coreStrategic or Co-development Partner · 1 January 2022Project leader for Hugin-Munin project developing open-source handwriting recognition models for Norwegian cultural heritage. Partners include National Library of Norway, Lumex, and TEKLIA. Funded by Research Council of Norway.
• National Library of NorwaycoreStrategic or Co-development Partner · 1 January 2022Partner in Hugin-Munin project for developing Norwegian handwriting recognition models and NorHand dataset for historical documents.
• INED (National Institute for Demographic Studies)coreStrategic or Co-development Partner · 1 January 2021Collaborative research project (SocFace) conducted with INED, Paris School of Economics, and SIAF to extract individual demographic data from French census lists 1836-1936. Supported by French National Research Agency (ANR).
• SIAF (French Interministerial Archives Service)coreStrategic or Co-development Partner · 1 January 2021Collaborative research partner in SocFace project for processing French census data. Also partner in SIMARA project for French National Archives.

Scale indicators8 records

Recent moves13 records

Expansion highlights7 records

Direct peers

• Transkribus (read&co): Transkribus is the most direct competitor to Teklia, providing AI-powered handwriting recognition (HTR) and document transcription tools specifically built for historical documents and archival materials. Both companies target GLAM institutions (museums, libraries, archives) and historical research communities with specialized models for handwritten text recognition.

Broad incumbents

• Google Cloud Document AI: Google Cloud Document AI is a broad incumbent offering document processing, OCR, form parsing, and entity extraction as part of a wider enterprise AI portfolio. While not specialized for heritage, it competes for general document AI workloads and serves many of the same institutional customers through Google Cloud contracts.
• ABBYY: ABBYY is an established document AI and intelligent document processing vendor offering OCR, data capture, and process automation across enterprise customers. It overlaps with Teklia in document recognition capabilities but addresses broader financial services, insurance, and supply chain workflows rather than heritage specialization.
• Microsoft Azure AI Document Intelligence: Microsoft Azure AI Document Intelligence (formerly Form Recognizer) is a broad cloud incumbent providing prebuilt and custom models for document extraction, layout analysis, and entity recognition. It competes with Teklia for institutional document AI workloads, though it lacks heritage-specialized HTR models.
• Hyperscience: Hyperscience is a document processing automation vendor offering handwriting recognition and form extraction for enterprise workflows including insurance, financial services, and government. It competes with Teklia's enterprise document processing capabilities, though oriented toward modern rather than historical documents.
• OpenText: OpenText is a large enterprise content management (ECM) vendor offering document management, archiving, and content services for regulated industries. It addresses the broader enterprise content management category that includes Teklia's governance, retention, and archival capabilities, though at much larger scale.

Emerging players

• eScriptorium: eScriptorium is an open-source platform from the École nationale des chartes (PSMN) for HTR training and document transcription. It overlaps directly with Arkindex on open-source heritage AI tooling and competes for academic and small institutional users in the same digital humanities community Teklia serves.
• Rossum: Rossum is an AI document processing company specializing in invoice and business document automation with a cloud-native platform. While focused on transactional documents rather than heritage, it shares the AI-powered document extraction value proposition and enterprise SaaS delivery model.
• Nanonets: Nanonets is an AI document OCR and data extraction platform serving enterprise workflows with pre-trained and custom models. It overlaps with Teklia's machine learning-driven document processing but focuses on commercial document automation rather than heritage and historical archives.
• Ephesoft: Ephesoft provides intelligent document processing and data capture solutions targeting government, financial services, and healthcare. It competes with Teklia in document AI infrastructure but focuses on forms processing and compliance-driven workflows rather than heritage or historical document analysis.

Market position

Strengths1 record

Weaknesses1 record

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• GitHubgitlab.teklia.com/arkindex

Compliance1 record

Revenue estimate

Valuation estimate

Number of profiles

Profiles12 records

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Teklia