Japan market brief · Document AI and AI-OCR · Updated 3 November 2026

The Japan market for document AI and AI-OCR in 2026

This page covers Japan's market for AI-OCR, intelligent document processing and document parsing for LLM and RAG projects: its size, where demand comes from, who sells now and who is arriving. It's for founders, CROs and investors planning a Japan entry or first Japan hire.

Why we publish this. We put this research together for companies planning to launch or grow in Japan. It is the groundwork we do before every Country Manager search, and every figure links to its source, most of them published only in Japanese.
¥63.6B
Japan OCR solutions market, FY2025 estimate, up 7.0% on FY2024
34.5%
AI OCR's forecast share of that market in FY2024, up from 30.1% in FY2023
61.9%
OCR users who process invoices, delivery notes and receipts, the top document type (2025)
85.1%
Companies with sales of ¥1 trillion or more using generative AI for language tasks (2025 survey)

Demand

Where Japanese demand for document AI comes from

Tax rules forced companies to digitise invoices from 2023, and growth is now moving to non-standard forms and to document preparation for generative AI.

Invoice and e-books rules drove 2023 and 2024

Since 1 October 2023, インボイス制度 (inboisu seido, invoice system) ties consumption tax credits to registered sellers' invoices. From January 2024, 電子帳簿保存法 (denshi chōbo hozon hō, e-books law) requires keeping electronic invoices and contracts as data. Deloitte Tohmatsu MIC credits both for FY2023 and FY2024 OCR growth and expects that demand to level off in FY2025.

Paper and fax are still the input

In JEITA's December 2025 survey of 310 OCR users, 48.1% used legacy OCR and AI-OCR together, 39.0% legacy only and 12.9% cloud AI-OCR only. Invoices, delivery notes and receipts led at 61.9%, overtaking order slips as non-standard invoices grew. Deloitte Tohmatsu MIC expects FY2025 growth from 非定型帳票 (hiteikei chōhyō, non-standard forms) older OCR couldn't read.

Banks and insurers automate paper intake

Hiroshima Bank gets about 12,000 mortgage pre-screening applications yearly, about 70% on paper. AI inside's DX Suite and RPA cut entry from 25 to 8 minutes each and moved it from 13 branches to one. In 2021 Nippon Life said on-premises DX Suite cut admin costs on new bank-counter contracts by about 40 to 50%.

Generative AI projects need clean documents

JUAS found 33.9% of 957 companies using generative AI for language tasks in late 2025, rising to 85.1% at companies with ¥1 trillion-plus sales. Deloitte Tohmatsu MIC puts enterprise generative AI implementation services at ¥44.3B in FY2025, forecasting ¥63.6B for FY2026. Macnica says many RAG systems become unstable in production because documents lose their structure.

Government is removing paper and in-person rules

A Digital Agency status report counts 12,205 legal provisions classed as アナログ規制 (anarogu kisei, analogue regulations) such as visual checks, posted paper notices and in-person viewing. Of the 8,162 that needed review, 8,017 had been reviewed by 30 September 2025.

Japanese forms are a technical barrier

AWS Textract lists printed Japanese, not its handwriting, invoice, receipt, ID, query or forms features. Google Document AI lists Japanese OCR, form and layout parsing, not handwriting. Microsoft lists printed Japanese for read and layout models. None mentions vertical text. AI inside reports its Japanese model PolySphere-3 averages 95.1% OCR accuracy across 50 form types.

Japan OCR solutions market, FY2020 to FY2025
¥ billions, Japanese fiscal years. FY2021 and FY2025 are estimates. Deloitte Tohmatsu MIC, 2022 to 2025 editions
Aspiring Partners
FY2020¥49.1B
FY2021¥50.1B
FY2022¥53.3B
FY2023¥57.3B
FY2024¥59.5B
FY2025¥63.6B

Who sells it in Japan

Who sells document AI and AI-OCR in Japan now

The hyperscalers and ABBYY compete with Japanese vendors that built models for Japanese forms and invoices. The leader shown runs the whole Japan business unless stated.

CompanyWhat it sellsIn JapanJapan leader (public record)
ABBYYIntelligent document processing and OCR platformABBYY Japan Co., Ltd., a KK, with a Japanese websiteTomohito Shimizu, named to lead ABBYY Japan in an undated ABBYY announcement
Microsoft (Azure AI Document Intelligence)Cloud OCR, layout and field extraction serviceSold through Microsoft Japan (日本マイクロソフト), established February 1986; Japanese listed for printed text in the read and layout modelsMiki Tsusaka (津坂美樹), President of Microsoft Japan since 1 February 2023
Google Cloud (Document AI)Cloud OCR, form, layout and specialised document parsersGoogle Cloud Japan G.K.; Japanese listed for its OCR, form and layout parsersTomoko Mikami (三上智子), Japan representative since 1 October 2025, succeeding Tomoyuki Hirate (平手智行)
Amazon Web Services (Textract)Cloud text and data extraction from scanned documentsSold through AWS Japan; Japanese listed for text detection, but handwriting, invoice and receipt features are English onlyAkihiko Shirahata (白幡晶彦), President of AWS Japan since November 2024
UiPath (Document Understanding)Document extraction inside its automation platformUiPath KK, founded in Tokyo in 2017Koichi Hasegawa (長谷川康一), Chairman and CEO of UiPath KK since 1 September 2025; he founded the KK
Box (Box AI)Content cloud with AI metadata extraction from documentsBox Japan, established August 2013; more than 20,000 Japanese customers and 24.0% of Box revenue in FY25 Q1 to Q3 (February 2025)Noriyuki Sato (佐藤範之), President of Box Japan since 1 February 2025
UpstageDocument Parse for complex documents and Syn Japanese LLMsUpstage AI KK (Upstage AI株式会社), Minato-ku, Tokyo; partners include HP Japan and AWS; hiring solutions architects and forward deployed engineers in JapanHiroyuki Matsushita (松下紘之), Representative Director and Country Manager
AI inside (DX Suite)AI-OCR and data entry platform, now positioned for AI agentsListed on TSE Growth (4488); revenue ¥4.75B to March 2026; 3,203 contracts, 77,975 users; OEM with NTT West, NTT Data, NTT EastTaku Toguchi (渡久地択), founder, Representative Director, President and CEO since 2015
Cogent Labs (SmartRead)AI-OCR for standard and non-standard documents, successor to its Tegaki handwriting OCRTokyo (Roppongi), founded April 2014; SmartRead launched in 2021Eric Hideyuki Whiteway (エリック・秀幸・ホワイトウェイ), Representative Director, President and CEO
Sansan (Bill One)Invoice receipt service that turns paper and PDF invoices into dataTSE-listed (4443); 4,599 paying Bill One contracts at 30 November 2025, up 38.9%; claims top revenue share in cloud invoice receipt servicesChikahiro Terada (寺田親弘), Representative Director and CEO

LayerX also sells AI-OCR invoice receipt in its Bakuraku suite. Hyperscience, Instabase, Rossum and LlamaIndex showed no Japan entity, distributor or Japan job ad in the sources checked.

Coming to Japan

Document AI companies with recent Japan signals

Public signals from 2026: distributor deals, Japanese partnerships, a Japan launch event and Japan job ads.

Company and backersWhat it sellsPublic signal
Unstructured
Menlo VenturesSeries B, March 2024
Converts documents into structured JSON for LLM and RAG pipelinesMacnica signed what it called the first Japan distributor agreement on 30 March 2026 and offers design and deployment support.
fileAI
SMBC Asia Rising FundSingtel Innov8Corporate venture investment, August 2026
Singapore AI platform that classifies files and extracts data from documentsJRE Ventures (JR East) partnership, 15 June 2026, to restructure legacy contracts and operating documents. Hiring Tokyo sales, engineering and customer success staff. Japan expansion funding raised 27 August 2026.
Lazarus AI
Document understanding AI for regulated industriesOKI announced a strategic partnership on 8 June 2026 to propose Lazarus AI's technology for social infrastructure, core corporate operations and defence. No Japan entity named.
INTSIG
TextIn document parsing and DocFlow document workflow productsAnnounced its full Japan launch on 19 September 2026, starting with insurance applications and claims, and hosted partners at Tokyo Midtown Hibiya on 29 September. No Japan entity named.
Apryse
PDF and document processing SDKs, including iTextEnterprise Account Executive, Japan (remote, English and Japanese required) open on its Lever board on 3 November 2026. XLsoft has sold iText by Apryse in Japan since 1 March 2023.
Iron Software
IronPDF, IronOCR and other document libraries for developersXLsoft signed a Japan distribution agreement and began selling the products on 18 August 2026. No Japan entity named.

Only public signals are listed: job ads, company registrations, partner announcements and press. A company on this list has not told us it is entering Japan, and none of them is named here as a client.

Hiring

What this means for your first Japan hire

Most document AI entrants start with a Japanese distributor and a small sales and technical team. These are the public facts on who leads and who gets hired first.

  • Document AI is one product inside the big platforms' Japan subsidiaries. Microsoft Japan president Miki Tsusaka (津坂美樹) joined from BCG in February 2023 (BCN). Tomoko Mikami (三上智子) succeeded Tomoyuki Hirate (平手智行) at Google Cloud Japan on 1 October 2025 (EnterpriseZine).
  • Japan heads have large-firm backgrounds. Tetsuo Minami (南哲夫), UiPath's country manager from February 2024, came from AWS Japan and Microsoft Japan (UiPath). ABBYY Japan's head spent 30-plus years at Ricoh (ABBYY). Upstage's Hiroyuki Matsushita worked at Panasonic and AWS (Upstage).
  • Distributors and SIs carry most early sales: Macnica for Unstructured, XLsoft for Apryse and Iron Software, OKI for Lazarus AI. AI inside also sells through NTT West, NTT Data and NTT East OEM products (AI inside deck).
  • First roles in job ads are sales and technical. Apryse's Japan account executive needs English and Japanese (Lever), and Upstage Japan asks for native Japanese in its solutions architect role (Upstage).
  • Named buyers in public case studies are banks, insurers and local government: Hiroshima Bank, Nippon Life, Sompo Japan and Nagaoka City appear in AI inside's FY2026 results deck (AI inside deck).

Our searches in this area

Glean First Japan Country Manager, work AICelonis Country ManagerDatabricks Solutions architecture leadershipFivetran First enterprise seller

Searches led by Murray Clarke, including work at the firms he built before Aspiring Partners. We hold back candidate names. Full track record

“Murray has done outstanding work for PARC in a difficult talent space. Thank you.”

Patrick Cook
Global Head, Xerox PARC

“Impressive research and a job well done. Thanks for the great work. Happy to be a reference.”

Josep Garcia
VP Asia-Pacific, SUSE Software

“Murray helped us lock in a key leader for our Japan business. He did an excellent job.”

Mathew Fleming
Global Sales Director, Fusion-io (now Sandisk)

Read all client references

Who our clients are backed by

Investors behind companies we have worked with.

Venture and private equity firms behind clients we have run Japan searches for. Under each firm is the client it backs. These firms are not our clients and have not endorsed us.

Andreessen Horowitz logo
Andreessen Horowitz
Databricks, Fivetran, dbt Labs
Sequoia Capital logo
Sequoia Capital
Glean, dbt Labs
GV logo
GV
mabl, dbt Labs
Benchmark logo
Benchmark
Chainalysis
Accel logo
Accel
Chainalysis
Kleiner Perkins logo
Kleiner Perkins
Glean
Lightspeed logo
Lightspeed
Glean
General Catalyst logo
General Catalyst
Glean
Coatue logo
Coatue
Chainalysis
Paradigm logo
Paradigm
Chainalysis
Altimeter logo
Altimeter
Glean, dbt Labs
Norwest logo
Norwest
Zenity
SoftBank Vision Fund logo
SoftBank Vision Fund
Zenity
Intel Capital logo
Intel Capital
Zenity
Geodesic Capital logo
Geodesic Capital
Glean
Insight Partners logo
Insight Partners
Veeam, Recorded Future
EQT logo
EQT
SUSE
Vista Equity Partners logo
Vista Equity Partners
mabl

Questions

Common questions about this market in Japan

How big is the AI-OCR and document AI market in Japan?

Deloitte Tohmatsu MIC puts the Japan OCR solutions market at ¥59.45B in FY2024 and estimates ¥63.59B for FY2025, up 7.0%. AI OCR was 30.1% of that market in FY2023 and was forecast to reach 34.5% in FY2024. The rest is older template OCR, which the firm describes as flat.

Why do Japanese companies still need OCR in 2026?

Paper and fax are still common inputs. The qualified invoice system from 1 October 2023 and the Electronic Books Maintenance Act rules from January 2024 pushed companies to turn invoices into data. In a December 2025 JEITA survey, 61.9% of OCR users processed invoices, delivery notes and receipts.

Who are the main AI-OCR vendors in Japan?

AI inside (DX Suite) claims the top market share, citing Deloitte Tohmatsu MIC's 2025 edition, and had 3,203 contracts in March 2026. Cogent Labs (SmartRead), Sansan (Bill One) and LayerX (Bakuraku) compete on forms and invoices. ABBYY, Microsoft, Google, AWS, UiPath and Box sell through their Japan subsidiaries.

Do AWS, Google and Microsoft document AI services handle Japanese?

Partly. AWS Textract reads printed Japanese, but its handwriting, invoice, receipt and ID features are English only. Google Document AI lists Japanese for its OCR, form and layout parsers without Japanese handwriting. Microsoft lists Japanese for printed text in its read and layout models.

How are foreign document AI startups entering Japan?

Mostly through Japanese partners. In 2026 Macnica signed Unstructured, XLsoft signed Iron Software, OKI partnered with Lazarus AI and JR East's JRE Ventures partnered with fileAI. Upstage set up a Tokyo KK with a Japanese country manager.

Is there demand for document parsing for RAG in Japan?

Yes. JUAS found 85.1% of Japanese companies with sales of ¥1 trillion or more using generative AI for language tasks in late 2025. Macnica says many RAG projects that work in a proof of concept become unstable in production because document structure is lost, which is why it took on Unstructured.

Case studies

How companies like these built their Japan business.

More Japan market briefs

Next step

Talk to the partner who would run the search.

Thirty minutes on your Japan plan, the market for the role and what it will take to hire well. You leave with our view of that market whether or not it becomes a search.

Murray Clarke
Murray Clarke
Founding Partner, CEO · Aspiring Partners
  • Placing Country Managers and presidents in Japan since 2003
  • More than 70 Country Managers and go-to-market leaders placed
  • ManpowerGroup (NYSE: MAN) bought his last search firm and made it Experis Executive
  • Data and AI searches include Glean, Databricks, Fivetran and Celonis

murray@aspiringpartners.jp · +81 90 8383 9025

Hiroshi Torobu
Hiroshi Torobu
Partner · Consulting and professional services
  • Eleven years recruiting in Japan, on executive and senior appointments across consulting, technology and manufacturing
  • Partner appointments into a Big 4 consulting firm in Japan
  • Country Manager appointments for overseas firms setting up in Japan

Partner, Managing Director and Director searches inside consulting firms.

How this page was researched. We produced this page with AI under Murray Clarke's direction. Every figure links to its original source, and the page was checked against those sources on 3 November 2026. Most sources are Deloitte Tohmatsu MIC press releases, JEITA and JUAS surveys, National Tax Agency, MHLW and Digital Agency pages, vendor filings and press releases on PR TIMES and TDnet, and Impress (Cloud Watch, IT Leaders) articles. Murray Clarke is accountable for what the page says. If you find an error, tell us at hello@aspiringpartners.jp and we will correct it.

Aspiring Partners K.K. Licensed by the Ministry of Health, Labour and Welfare for fee-charging employment placement, licence 13-ユ-318925. Company names on this page describe the Japanese market and are not clients unless we say so. Nothing on this page is legal, tax or investment advice. aspiringpartners.jp