Retrieva

What Retrieva does

Retrieva is a Tokyo-based AI company founded in August 2016 as a spin-off from Preferred Infrastructure, built on a decade-plus foundation in natural language processing and information retrieval. The company operates three business lines: (i) AI technical services, including AI system development support covering RAG and AI-agent construction, LLM research and verification, and corporate generative-AI training; (ii) an AI product business anchored by YOSHINA, a subscription text-analytics SaaS used by 100+ Japanese enterprises for voice-of-customer, survey, and call-centre analysis, with add-ons YOSHINA Search and YOSHINA Chatbot; and (iii) a bioinformatics business comprising GGGenome (high-speed short-sequence nucleotide search distributed on offline SSDs), CRISPRdirect, ORIGENICS, and siDirectPlus, used by pharmaceutical firms, CROs, and academic institutes for nucleic-acid drug off-target evaluation.

Underpinning all three lines is proprietary Japanese-language NLP IP: RetrievaBERT, RetrievaEmbedding-01 AMBER (released March 2025 on Hugging Face), and a Japanese-trained T5 encoder-decoder LLM. The company monetises through subscription SaaS at ¥150,000–¥300,000/month for YOSHINA, single perpetual licenses of ¥4,800,000 or ¥2,400,000/year for GGGenome, ¥3,000,000/year for siDirectPlus, and unit-priced training sessions (¥600,000–¥700,000). Distribution combines direct enterprise sales with named channel partners including Ryoyo Electro, Fixstars, CAC, TMJ, NTT DATA, CIE, Transcosmos, and Gnext, and an AI-agent implementation practice built around Dify and the joint NEDO-track KOKOROE LLM-safety diagnostic with group subsidiary Everyleaf (acquired July 2019).

Retrieva is a private, venture-backed Japanese corporation with headquarters in Toshima-ku, Tokyo (WeWork Metropolitan Plaza, since February 2024) and a regional office in Miyazaki City (opened July 2022). Leadership transitioned from founding CEO Kawahara Kazuya to former CFO Taguchi Takuya in February 2023; research direction is led by CRO Nishitoba Jiro. The company holds AWS Partner Software Path certification (passed FTR in ~2 months in 2022), 14 consecutive ITreview Grid Leader awards for text mining (2023–2026), and patent JP 7640104 (granted February 2025) covering text-data visualisation in YOSHINA.

Name

Retrieva

Legal name

株式会社レトリバ

Website

https://retrieva.jp

Company type

Private

Founded year

2016

Operating status

Operating

Headcount range

11–50 employees

Short description

Retrieva is a Tokyo-based AI company providing Japanese-language NLP services and products: YOSHINA text-analytics SaaS, GGGenome/siDirectPlus bioinformatics tools for nucleic-acid drug design, and consulting on RAG, AI-agent (Dify) implementation, and generative-AI adoption for Japanese enterprises.

Ownership category

akta.pro rank

Product category

Enterprise AI Software (NLP and Text Analytics)

NAICS

Custom Computer Programming Services (541511), Software Publishers (5132)

SIC

Services-Computer Programming Services (7371), Services-Prepackaged Software (7372)

akta.pro primary industry

AI/ML Engineering & Model Development Services (BPAEAHAF)

akta.pro secondary industries

LLMOps & Generative AI Platforms (Prompt/Agent Orchestration, RAG) (HDAEANAD), Generative AI & LLM Solutions Services (RAG, Agents, Copilots) (BPAEAHAG), AI Observability, Monitoring & Evaluation Platforms (Drift, Quality, Safety) (HDAEANAF)

Keywords

• Natural language processing
• Text analytics software
• Genome sequence search
• AI system development
• Nucleic acid drug design

Headquarters

HQ city

Tokyo

HQ country

Japan

HQ region

Asia

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

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

Revenue model

1. AI技術支援事業 (AI Technical Services): Professional services revenue from AI system development, consulting, researcher support, and engineer training. Includes RAG/AI agent implementation, LLM research support, and skills training programs for client engineers.
2. SaaS Subscription (YOSHINA): Monthly/annual SaaS subscription for YOSHINA text analytics platform. Pricing tiers: Light Plan from 150,000 JPY/month, Standard Plan from 300,000 JPY/month. Includes initial setup and support fees. High retention rate of 99.3% reported.
3. SaaS Subscription (YOSHINA Search): Generative AI-powered search tool subscription, part of the YOSHINA product family.
4. GGGenome Package (Hardware + Software): Portable SSD with pre-installed GGGenome software sold as one-time perpetual license (4,800,000 JPY/license, no time limit) or annual subscription (2,400,000 JPY/year including program updates and database updates). Includes customer support during subscription period.
5. siDirectPlus Subscription: Annual subscription for siDirectPlus siRNA design software package (3,000,000 JPY/year including GGGenome package version). Sold as SSD-based offline product.
6. Training/Education Programs: Generative AI training programs including AI seminars (beginner, technical, strategy), prompt training (basic, advanced). Online: 600,000 JPY; On-site: 700,000 JPY per corporate session. Campaign pricing for basic prompt training at 100,000 JPY.

Pricing tiers

Go-to-market motion3 records

Distribution channels5 records

Marketing channels11 records

Core offering

Retrieva develops and sells proprietary AI software products and provides AI technology services for Japanese enterprises. Its primary product is YOSHINA, a SaaS text analytics platform that classifies and generates reports from text data using NLP and generative AI. It also offers GGGenome, a high-speed nucleotide sequence search system for nucleic acid drug off-target evaluation, plus CRISPRdirect, ORIGENICS, and siDirectPlus for genome editing and siRNA design. AI services include RAG/AI agent implementation, LLM research support, and engineer training.

Product overview

Retrieva provides a unified portfolio combining AI technology services (AI system development, research support, and engineer training), an AI text analytics product (YOSHINA), and a bio-informatics product line (GGGenome, CRISPRdirect, ORIGENICS, siDirectPlus). Underpinning the portfolio are proprietary Japanese NLP models (RetrievaBERT, AMBER embedding model, T5 LLM) developed over 10+ years of research. Core technology areas are natural language processing, knowledge retrieval, generative AI, and AI agents, applied across enterprise knowledge management, customer voice analysis, and nucleic acid drug discovery.

Differentiator

Problem solved

Functional benefit

Brands

• YOSHINA: AI-powered text analysis and text mining tool for analyzing customer voice (VoC), surveys, and qualitative data. Includes generative AI auto-report and chatbot features.
• GGGenome
• CRISPRdirect
• ORIGENICS
• siDirectPlus
• KOKOROE
• Dify活用支援パッケージ
• AMBER

Products and services

• YOSHINA AI-powered text analytics SaaS that automatically classifies, analyzes, and generates reports from text data such as customer support tickets, surveys, call center logs, and reviews. Features topic classification, sentiment analysis, auto-report generation, and generative AI-powered improvement recommendations. Used by call centers, marketing, and product development teams across 100+ enterprise clients.
• GGGenome Package High-speed, exhaustive nucleotide sequence search software for offline use, designed for short-sequence queries including mismatches and gaps. Used for off-target gene evaluation in nucleic acid drug development. Supports latest RefSeq and D3G databases across 9 species. Distributed on portable SSD for secure offline operation.
• CRISPRdirect Guide RNA design tool for CRISPR-Cas genome editing that designs guide RNAs minimizing off-target effects by considering experimental conditions and parameters.
• ORIGENICS Integrated nucleic acid drug design platform covering target gene collection, target sequence extraction, antisense or siRNA design, and safety evaluation via GGGenome integration. Supports siRNA, ASO, and CRISPR-Cas gRNA design from a single platform.
• siDirectPlus Offline siRNA design software package based on Tokyo University's proprietary algorithm. Enables design of highly effective and safe siRNA molecules with off-target evaluation powered by GGGenome. Supports seed region analysis, 3'-UTR search, and fully offline operation.
• RetrievaBERT Proprietary Japanese BERT model, the largest-scale Japanese BERT model with long-sequence support. Part of Retrieva's core NLP technology stack developed over 9+ years, used in enterprise AI applications and available on Hugging Face for commercial licensing.
• RetrievaEmbedding AMBER Japanese search-optimized text embedding model available in large and base sizes. Claims highest accuracy among compact Japanese embedding models for enterprise internal search and RAG use cases. Licensed for commercial use.
• Retrieva T5 Japanese-trained encoder-decoder LLM based on T5 architecture, pre-trained on large-scale Japanese corpora. Part of Retrieva's NLP research output for Japanese language tasks.
• KOKOROE LLM vulnerability diagnosis web application jointly developed with Everyleaf. Diagnoses whether LLMs generate safe responses and tunes LLMs based on diagnostic results using human-written guidelines for enterprise LLM safety evaluation across various industry-specific risks.
• Dify Utilization Support Service Implementation and consulting services for Dify, an open-source AI application development platform. Includes Dify implementation support (requirements definition, design, initial build, testing) and Dify consulting (quality improvement and operational optimization for existing Dify apps). Helps enterprises succeed with AI agent projects on Dify.
• AI System Construction Support Service AI system development services covering the full lifecycle from concept to deployment and operation. Specializes in AI agent and RAG system construction for business reform, including internal document structuring for RAG, search improvement with domain-specific embedding fine-tuning, and AI-powered proposal quality checking.
• AI Agent Construction Support AI agent building support service using cutting-edge generative AI and LLM technology to help enterprises succeed with AI agent projects. Addresses production-grade deployment challenges beyond initial PoC.
• AI Research Support LLM research and verification services combining paper research, proof of concept, and implementation. Addresses unexplored LLM use cases and deployed model improvements including accuracy, speed, data preparation, and safety enhancement.
• Engineer Training Support AI engineer training combining lectures and hands-on sessions. Project-based practical training guides teams to become self-sufficient in AI utilization within months. Covers generative AI training programs at beginner, technical, strategy, and prompt engineering levels.
• YOSHINA Bot ChatGPT-powered chatbot using Retrieva's proprietary 'knowledge set' technology combined with ChatGPT summarization. Delivers human-like responses by selecting answer candidates from a knowledge base and summarizing with ChatGPT.

Quantifiable outcome

• YOSHINA customer retention rate of 99.3% (Oct 2022–Nov 2023)
• +3 more outcomes

Named customers4 records

Segments4 records

Ideal customer profiles4 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration1 record

AI capability8 records

Feature9 records

Partnerships

Eight partnerships are on record, tiered minor and core.

• 東京大学 × ベンチャー企業 産学連携AI研究フォーラム (UTokyo × Venture AI Research Forum)minorStrategic or Co-development Partner · 29 June 2026Participating in UTokyo's industry-academia AI research forum, joining Fixstars as a cooperating organization. Aligns Retrieva's NLP, search, and generative AI expertise with UTokyo's cutting-edge research for Japanese AI technology development and social implementation.
• 株式会社万葉 (Everyleaf Inc.)coreStrategic or Co-development Partner · 30 March 2026Group company; jointly developed KOKOROE, an LLM vulnerability diagnostic web application. Everyleaf provides Ruby on Rails web application development. KOKOROE enables LLM safety diagnosis and tuning using human-written policies. Both companies collaborating on NEDO GENIAC-PRIZE safety initiatives.
• 菱洋エレクトロ株式会社 (Ryoyo Electro Co., Ltd.)coreStrategic or Co-development Partner · 29 October 2025Capital and business partnership (資本業務提携) established October 2025 to build an integrated support system for enterprise AI adoption. Combines Ryoyo Electro's GPU infrastructure and customer base with Retrieva's NLP and LLM implementation expertise. Joint sales, consulting, and co-development of AI solutions for manufacturing, telecom, and finance sectors. Pursuing Agentic AI and Physical AI deployment. Ryoyo Electro (part of Ryoyo-Electro Holdings Group) is NVIDIA's authorized distributor with strong GPU and medical device robotics presence.
• Dify (LangGenius, Inc.)coreTechnology or Integration · 1 June 2025Dify Implementation Support Packages providing Dify introduction services and Dify consulting services. Helping enterprises build and operate AI agents using Dify (open-source AI app development platform supporting 100+ LLM models). Services cover requirement definition, design, build, testing, quality improvement, and operational optimization. Japan office and Dify Association established 2025.
• NVIDIA (via RYOYO AI Techmate Program)coreTechnology or Integration · 12 November 2024Participating in Ryoyo Electro's 'RYOYO AI Techmate Program' with NVIDIA as a partner. Provides NVIDIA AI Enterprise platform, training support, technical support, and enterprise matching for manufacturing, medical, energy, and other industries. Retrieva contributes NLP expertise and AI talent education programs to the program.
• 国立研究開発法人 医薬基盤・健康・栄養研究所 (NIBIOHN)minorOthers · 29 November 2022NIBIOHN adopted GGGenome Package for artificial nucleic acid screening project off-target risk evaluation. Reported ~90% workload reduction. Using GGGenome for off-target gene extraction with multi-sequence simultaneous search and cross-database search.
• プラチナバイオ株式会社 (Platinum Bio Co., Ltd.)minorOthers · 1 November 2022Platinum Bio adopted GGGenome Package for their Genome Editing Cloud platform's safety evaluation. Citing GGGenome's simple static API, offline capability, and support for complex searches including degenerate notation and gap-type off-target searches.
• 株式会社ジーネクスト (Gnext Co., Ltd.)coreTechnology or Integration · 15 May 2022Business partnership since May 2022, API integration between YOSHINA and Gnext's Discoveriez customer management platform. Strengthened in August 2025 with generative AI integration: YOSHINA's AI automatically extracts important VoC from Discoveriez call logs and generates automatic reports. Enables real-time customer feedback analysis for enterprise CX improvement.

Scale indicators7 records

Recent moves9 records

Expansion highlights7 records

Direct peers

• Stockmark: Japanese AI company providing NLP-driven enterprise solutions including document analysis and knowledge management. Comparable in serving Japanese enterprise text analytics needs, competing for similar customer segments as YOSHINA.
• ELYZA: Japanese AI startup developing Japanese-specialized LLMs and enterprise LLM deployment services. Closely comparable to Retrieva's AI system development and LLM research support offerings, with similar Japanese NLP moat.
• Cogent Labs: Tokyo-based AI company offering NLP, text analytics, and document automation for Japanese enterprises (e.g., Tegaki, KAIDOKU). Directly comparable to Retrieva's YOSHINA in serving Japanese enterprise NLP needs, though with broader focus including handwriting and contract review.
• Lightblue Technology: Japanese AI startup providing NLP, image recognition, and predictive analytics for enterprise. Comparable AI services and NLP specialization for Japanese enterprise customers, similar size and stage to Retrieva.
• User Local: Japanese AI/text analytics company offering social media monitoring, VoC analysis, and natural language processing tools. Overlaps with YOSHINA's VoC and survey analytics use cases for Japanese enterprises.

Emerging players

• Sakana AI: Tokyo-based foundation model research company building Japanese language and multimodal AI models. Competes with Retrieva's proprietary models (AMBER, RetrievaBERT) and represents the Japanese-native frontier-model threat to Retrieva's embedding moat.
• Turing: Japanese AI company developing large Japanese foundation models (Fugaku-LLM partnership) and autonomous driving AI. Competes indirectly with Retrieva on the Japanese LLM frontier-model axis, representing an emerging threat to AMBER's positioning.
• LangGenius (Dify): Open-source AI application development platform and Retrieva's strategic partner. While Dify enables enterprises to build AI apps without Retrieva, the Dify support packages create both partnership opportunity and disintermediation risk.

Others

• Fixstars: TSE Prime-listed AI/GPU acceleration company and Retrieva's capital partner. Operates in adjacent GPU cloud and AI services space, providing infrastructure that Retrieva's AI services rely on while also offering competing implementation capabilities.

Broad incumbents

• AI inside: Listed Japanese AI company offering broader portfolio of AI products (OCR, NLP, document automation, LLMs) with established enterprise distribution. Competes with YOSHINA in text/document AI and is a larger, publicly traded incumbent.

Market position

Strengths4 records

Weaknesses4 records

Competitive moat6 records

Key risks6 records

Key highlights6 records

Customer concentration

• Xx.com/retrieva_jp
• YouTubeyoutube.com/channel/UC5mjAq3PgaWhwcRyun7gteQ
• Facebookfacebook.com/retrieva.jp

Compliance1 record

Revenue estimate

Valuation estimate

Number of profiles

Profiles7 records

Funding overview

Funding rounds3 records

Investors5 records

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

M&A1 record

Investments

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

Frequently asked questions about Retrieva