The opportunity is more people building useful AI on the enterprise’s existing infrastructure. Keeping an owner, approval path and monitoring attached to each project matters as usage expands.
See it taking shape ↓
Let business and technical teams reuse the same project.
Dataiku automated collection and processing, combined public and internal datasets, and made reporting data available in the lake.
The food-safety team manually collected data, managed multiple spreadsheets and generated reports.
The 2024 case concerns a food-safety reporting workflow, not all operations at Perdue.
Financial crime specialists, data scientists and engineers built STABLES, a scoring engine that ranks money mule risk; investigators decide what to do with each case.
Investigations of suspected money mule accounts could take months, and investigators reviewed alerts with limited prioritisation.
Dataiku customer story published 2026-09-21, tied to its 2026 award; figures are approximate and are the company's account of what the bank reported, for one use case.
An illustrative food-safety team automates recurring data preparation and reporting, grounded in Perdue Farms’ case. The same environment can later support models and agents.
The analyst identifies USDA sources and internal datasets. Perdue’s architecture used Azure and Snowflake rather than replacing both platforms.
Teams use visual preparation or code in a shared project. The aim is a repeatable pipeline instead of manually rebuilding each monthly report.
Visual and code development share one environment. LLM Mesh lets a team manage providers, quotas and monitoring when a workflow uses language models.
Registries, approvals and lifecycle monitoring accompany production. Perdue’s automated reporting freed its team to focus more on analysis.
Dataiku starts from a problem it says it sees in large companies. AI is spreading across clouds, models, agents and applications faster than most organisations can control it, and a company that buys from many vendors ends up with duplicated work, uneven performance and blind spots in governance. The chief executive, Florian Douetteau, puts it in one picture: ask a bank how many servers it runs and you get an answer to the decimal; ask how many AI agents it runs and you get a shrug. In a survey of 685 chief information officers that Dataiku commissioned and published in September 2026, 84% agreed that employees are creating AI agents and applications faster than IT can govern them. The sponsor sells the cure, so read it as the company's view of the market.
governing AI at every single step
Florian Douetteau · Co-founder and CEO · Company announcement of the Platform for AI Success · 2026-03-09
The belief is older than generative AI. Dataiku launched in Paris in 2013, founded by Florian Douetteau, Clément Stenac, Thomas Cabrol and Marc Batty, on the idea that data work should not stay with a handful of specialists. Stenac said as much in Seoul in July 2025: the founders wanted every kind of company, and every person inside it, to use data and get value from it. In 2026 the company states the idea in three parts. People: business experts build the AI inside guardrails that IT sets. Orchestration: data, models, agents and business rules run as one system. Governance: testing, approval and measurement happen at every step, so nothing ships without sign off. What separates it from a cloud vendor is one line from its March 2026 relaunch: it integrates with all of a customer's data platforms, models and agent frameworks, and depends on none.
Why now comes down to three pressures. Executives must show results: in a 2026 Dataiku survey of 900 chief executives in eight countries, South Korea among them, 80% said they actively question or challenge AI outputs. Agents are multiplying faster than anyone can list them. And in August 2026 Douetteau said restricted access to one frontier model had pushed enterprise customers to rethink relying on a single AI provider. All three point to a neutral layer above the vendors.
A case Dataiku published in November 2024 shows where the product sits. A food safety team at Perdue Farms, a US food company, spent hours every month gathering US Department of Agriculture data across spreadsheets and building reports by hand. To connect, staff upload their files to a shared site while API calls pull the government data, and both land in a Dataiku project running on the Azure and Snowflake systems Perdue already had. To prepare, analysts clean the data with visual tools while engineers work in code on the same project, so the monthly rebuild becomes a pipeline anyone can inspect. The next steps illustrate the current platform rather than attribute agents to this 2024 case: a team can build a model or agent, route language-model calls through the LLM Mesh, and attach an owner, approval path and monitoring. Perdue reports that tasks which took days now take hours.
A person still decides. In Dataiku's account of Standard Chartered's money mule scoring engine, AI identifies and prioritises the risk, and investigators decide what to do about it.
Most of the proof is first party: customer stories the company writes, awards it judges with outside experts, and surveys and a return on investment study it commissioned. That is normal in enterprise software, and a Korean buyer will still ask for a reference they can call.
For someone joining in Korea, the product is a change in who builds AI inside a customer. LG Chem described in 2023 a no code platform for plant and business staff who want to analyse data without programming, with Dataiku as the analysis layer. Finding that ambition in a Korean company, and the budget behind it, is the work.
Dataiku sells one platform and keeps adding layers to it. For most of its life the company offered a single product, Data Science Studio, which its Seoul job posting still calls DSS. Analysts use visual workflows while data scientists and engineers write code in the same project, on top of the customer's existing data, tools and infrastructure. In Korea it is sold both installed on a customer's own servers and as a cloud service, and the company points to its ISO 27001 and SOC 2 Type II certifications as its way into finance, healthcare and the public sector. What it replaces, in the company's own examples, is spreadsheet work and older tools such as Alteryx workflows.
The LLM Mesh is the gateway between a customer's applications and the language models they call. An administrator connects providers such as OpenAI, Azure OpenAI, Anthropic, Cohere, Google Vertex, AWS Bedrock, Databricks, Snowflake Cortex, Mistral and Hugging Face from one settings screen, then routes requests, watches cost and applies safety rules centrally. The promise is that a company can adopt a new model or switch vendor without rewriting its applications. It carries Douetteau's argument against tying a critical process to one frontier model.
Agent Hub, introduced in October 2025, is where employees build and share agents: a form asks for the agent's instructions, its role and the base model it runs on. Cobuild, available since June 2026, lets a user describe a business objective in plain language and generates a whole project in the visual interface, with pipelines, models, agents and applications that stay traceable. In September 2026 the company opened Cobuild to developers through Dataiku Headless, which works from coding tools such as Claude Code, OpenAI Codex and Cursor and can move an Excel workbook or an Alteryx workflow into a governed project.
Agent Management, unveiled at the Succeed conference on 24 September 2026 with general availability announced for October 2026, is a standalone product that sits above other vendors' agents. It connects to AWS Bedrock, Databricks, Google Vertex, Microsoft Copilot Studio and Azure Foundry, Salesforce Agentforce, Snowflake Cortex and Dataiku itself, scans them into one inventory, measures business and technical performance and flags the riskiest agents. It is priced per instance per year, with monitoring metered per agent, which gives a seller a clear commercial unit. Dataiku sets it against monitoring built into a single vendor's stack, which it argues favours that vendor's own agents.
Monitoring tells you an agent is running.
Florian Douetteau · Co-founder and CEO · Company announcement of the 2026 CIO survey · 2026-09-24
| Product | What it does | Status |
|---|---|---|
| Dataiku platform (DSS) | Visual and code work on data, models and agents in one governed project | Core product since founding |
| LLM Mesh | Gateway to many model providers with routing, cost reporting and safety rules | Part of the platform |
| Agent Hub | Build and share agents inside the platform | Introduced October 2025 |
| Cobuild | Turns a plain language goal into a governed project; Headless mode for coding tools | Available since June 2026, expanded September 2026 |
| Agent Management | Inventory, performance and risk for agents built on any platform | General availability announced for October 2026; priced per instance yearly, monitoring per agent |
| Reasoning Systems | Packaged systems for specific operations | Manufacturing Operations available from March 2026 |
Named customers come with numbers, each from the company or the customer and each for one use case. At Standard Chartered, financial crime specialists, data scientists and engineers built a scoring engine for money mule risk; in its first 12 months it surfaced about 40,000 accounts, cut mean investigation time by about 75%, and the bank froze or prevented about $90M of laundering, by Dataiku's account. At Bank Hapoalim an agent handles more than 90% of incoming foreign currency transfers under human oversight, cutting handling time from 6 minutes to 2. ExxonMobil's supply chain experts automated workflows across 40,000 railcars and more than 800 vessels and saved more than $10M.
Korean customers speak for themselves in Korean press. HSAD, the advertising company of the LG group, chose Dataiku for unstructured text analysis and to avoid lock-in to any one technology; its AI strategy team lead said market research that took at least a month through an outside firm now takes two weeks, and employees built more than 800 agents in the first month. LG Chem picked Dataiku first and then Snowflake to sit underneath it for a citizen analyst platform that predicts product quality from plant data.
Every figure above is one workflow at one customer, measured by that customer or told by Dataiku. In a Korean sales conversation, cite a case with its scope attached: one money mule engine, its first 12 months, the bank's own measure.
ARR at each disclosure date. The 2025 points are thresholds the company announced as surpassed; earlier points come from a press report and from a research firm's summary of company disclosures. Not recognized annual revenue. These are dated disclosures, not a continuous growth record.
| Date | Value | Basis |
|---|---|---|
| 2022 | $150M | First public ARR figure, revealed to investors in 2022, as reported in December 2022. |
| 2022-12 | $200M | ARR at the end of 2022, from company disclosures as summarized by a research firm. |
| 2023-09 | $230M | ARR crossed $230M in September 2023, same basis. |
| 2025-01-16 | $300M+ | Company announced surpassing $300M ARR; the chart plots the threshold. |
| 2025-10-17 | $350M+ | Company announced surpassing $350M ARR; still the figure it cited in June 2026. |
TechCrunch · Sacra · Dataiku
Company value set at each private round. The December 2022 Series F was priced below August 2021. These are dated disclosures, not a continuous growth record.
| Date | Value | Basis |
|---|---|---|
| 2021-08 | $4.6B | Valuation at the August 2021 round, as reported in December 2022. |
| 2022-12 | $3.7B | $200M Series F led by Wellington Management at a $3.7B valuation. |
TechCrunch · Dataiku
Raised $200M at a reported $3.7B valuation.
Company disclosed over 700 customer organizations.
Company disclosed over 750 customer organizations.
Dataiku earns money from large companies that subscribe to its platform. A research firm describes the buyers as enterprises in financial services, pharmaceuticals, manufacturing and retail, reached through direct sales with systems integrators, consultancies and cloud vendors alongside, and describes pricing as based on the number of users and the computing resources consumed. The newest product carries a price structure the company states itself: Agent Management is priced per instance per year, with monitoring metered per agent. In Korea the platform is offered on the customer's own servers or as a cloud service, and the country manager said in 2025 that smaller companies were increasingly choosing the cloud version.
The company reports annual recurring revenue, or ARR: the yearly value of contracts that keep renewing, at one point in time. It first revealed the figure in 2022, at $150M. It announced more than $300M in January 2025, saying ARR had more than doubled over the previous three years, and more than $350M on 17 October 2025, the figure it still cited in June 2026. The January 2025 release explicitly reports more than doubling ARR over the prior three years; the dated thresholds themselves are separate from an exact growth rate. A research firm estimates $342.5M for September 2025, consistent with the company's October announcement. Customer organisations rose from more than 500 in December 2022 to more than 700 in January 2025 and more than 750 by October 2025, including one in four of the top 500 companies of the 2024 Forbes Global 2000, excluding China.
| Date | Figure | What it measures | Basis |
|---|---|---|---|
| 2021-08 | $4.6B | Valuation at a private funding round | Press report |
| 2022 | $150M | ARR, first public figure | Press report |
| 2022-12 | $200M at $3.7B | Series F funding and valuation | Company and press |
| 2023 (end) | about $500K | Average ARR per customer | Research estimate |
| 2025-01-16 | $300M+ | ARR threshold | Company |
| 2025-09 | $342.5M | ARR | Research estimate |
| 2025-10-17 | $350M+ | ARR threshold | Company |
| 2025-10-17 | 750+ | Customer organisations | Company |
| 2025-10-17 | 1,250+ | Employees, 13 offices and remote | Company |
Dataiku competes for the layer where companies build and run analytics, models and agents. In 2022 its main named rivals were Databricks and DataRobot, and the large clouds sell managed machine learning of their own. A research firm places Databricks closer to data engineering and infrastructure and Dataiku higher up, and describes the clouds' offers as carrying lock-in. The lines blur, because Dataiku connects to Databricks and Snowflake and now offers to manage agents built on them. Gartner named it a Leader in its 2026 Magic Quadrant for AI platforms for data science and machine learning, the fifth time in a row, and reviewers on Gartner Peer Insights gave it 4.7 out of 5 with 98% recommending it as of 4 May 2026, both as the company reports them.
driving financial efficiency
Adam Towns · Chief Financial Officer · Company announcement of the $300M ARR milestone · 2025-01-16
Three risks matter for a seller. First, the platforms Dataiku sits on are building the same layer: each cloud and data platform now runs its own agents with its own monitoring, and if customers accept one vendor's controls as good enough, the neutral layer weakens. Second, buyers must prove value fast: in a Dataiku survey of 600 CIOs published in February 2026, 71% said AI budgets were likely to be cut or frozen if targets were missed by mid 2026, and nearly three in four regretted a major AI vendor or platform decision made in the previous 18 months. Third, the December 2022 financing was priced below August 2021. A 2025 report described banks preparing a potential US listing, with timing subject to change. Those records help frame questions about an equity offer's terms and share class; the 2022 transaction price does not itself value every later offer.
Two habits help in an interview. Say ARR with its date, and keep it apart from revenue: $350M was a threshold of recurring contracts in October 2025, not income booked over a year. And ask how the Korea business is measured. The Korea figures given in public are growth rates, 200% or more in July 2025 and double digits in May 2026, so the useful follow up is the base, the period and the measure.
Douetteau founded Dataiku in 2013 with Clément Stenac, Thomas Cabrol and Marc Batty.
The founding principle was broader collaboration around data; the platform later added generative AI and agents while retaining governance.
“Teams built agents faster than anyone could count them.”
Florian Douetteau · Company announcement of Agent Management · 2026-09-24
Florian Douetteau co-founded Dataiku in 2013 and has led it since. He is the public voice of the company's argument, and in 2026 he has been specific about where the market is going. In a September 2026 television interview, as the broadcaster summarised it, he said enterprises are moving from renting intelligence through frontier models to buying their own with open weight models so they keep ownership of their intellectual property, and that companies should train their own people rather than hand AI to consultants.
Clément Stenac, co-founder and chief technology officer, carries the product argument. In March 2026 he said real enterprise decisions need data feeding models, models informing agents, and agents controlled by explicit business rules and human oversight. In September 2026 he described the gap Cobuild is meant to close: technical teams and business teams working on different platforms without shared data or shared guardrails, manageable with a few pilots and a liability at scale. He presented the company in person in Seoul in July 2025. The other two founders are Thomas Cabrol and Marc Batty, who sits on the board with investors from FirstMark and Battery Ventures.
No amount of prompt engineering replaces structured orchestration
Clément Stenac · Co-founder and CTO · Company announcement of the Platform for AI Success · 2026-03-09
Four decisions explain the company a hire would join. One platform: for most of its history Dataiku sold a single product and extended it from analytics to machine learning, generative AI and agents, instead of building a family of separate tools. Neutrality: it runs on the clouds, data platforms and models a customer already has, connects to Databricks and Snowflake, and competes for the layer above them. Discipline on money: in December 2022 it raised $200M in a Series F led by Wellington Management at a $3.7B valuation, below the $4.6B of August 2021, and its finance chief argued in January 2025 that it grows with financial efficiency, unlike the heavy operating costs of model providers. A commercial reset: in June 2026 it hired Maxwell Long as President and Chief Revenue Officer.
Long comes from Smartsheet, where he was President of Go-to-Market, led a team of more than 2,300, saw the company grow from hundreds of millions in ARR to more than $1B and helped lead its $8.4B sale to Blackstone and Vista. At Dataiku he leads global sales, customer success and partnerships and reports to Douetteau, which puts him at the top of the commercial line a Seoul seller sits in.
Max has done this before
Florian Douetteau · Co-founder and CEO · Company announcement of Maxwell Long's appointment · 2026-06-01
Closer to Korea, Andrew Boyd was announced on 31 March 2026 as Senior Vice President and General Manager for Asia Pacific and Japan. He oversees Korea, Japan, ASEAN, Australia and New Zealand, brings more than 20 years in enterprise technology in the region, and worked at Microsoft and AWS, where he helped Korean companies expand abroad. In May 2026 he told a Korean daily he would keep spending a lot of time in Seoul. Jo Goh joined in January 2026 as Area Vice President for APJ partnerships. In Seoul, 김종덕 has been country manager since November 2024, after 30+ years in servers, security and cloud.
The published values are short. Accountable camaraderie means thinking, acting and celebrating as a team regardless of function. Dedicated to learning means staying curious about new technology and growing by helping others grow. The careers page counts progress, not hours, says benefits differ by country and describes a global, hybrid team. Open roles sit on the company's own Greenhouse board, which carried 19 jobs on 8 October 2026.
Signet's reading: Korea sits in an APJ region led by an SVP appointed in March 2026, while global sales report to a President and CRO appointed in June. Those dated changes identify the regional and commercial leaders to ask about Korea's account priorities, decision authority and support during onboarding.
The AE owns the sales cycle and coordinates Marketing, Product and Engineering around business value and customer issues.
Country Manager, Korea
The October 7 LinkedIn Korea filter showed 13 associated profiles. Groups summarize observed commercial, solutions/support and marketing functions; they are not a staffing census or reporting hierarchy. The country manager is separately verified in Korean press.
Dataiku moved into Asia after Europe and North America, and its co-founder and CTO said in Seoul in July 2025 that it entered Korea about three years earlier because Korean demand for AI was large; a 2026 newspaper profile dates the start to 2021. The fit is specific. Korea's large manufacturers and financial firms want AI and also want it controlled, often on their own servers or closed networks, and Dataiku installs there: Samsung Fire & Marine chose it on condition it run on a closed network. The APJ head said in May 2026 that Korean CEOs are very open to AI and face strong demands to comply with regulation, and in Dataiku's own 2026 survey 93% of Korean CEOs said their job could be at risk without visible AI results by year end, against 80% globally. Korean customers also cite the neutrality argument directly: HSAD chose Dataiku partly to avoid lock-in to one technology. The results so far are a team that doubled after November 2024, double digit revenue growth reported in May 2026, and a Dataiku Summit held in Seoul in June 2026.
The CTO said in July 2025 that Dataiku entered Korea about three years earlier, after Europe and North America; a 2026 newspaper profile gives 2021. Manufacturing customers came first.
An LG Chem AI team lead describes a no code analysis platform built with Dataiku and Snowflake, Dataiku chosen first.
김종덕 joins as Korea country manager after 30+ years in servers, security and cloud.
Clément Stenac and the Korea team present in Seoul; the country manager says the team has doubled and customers now reach finance, retail and advertising.
About 20 teams build shipbuilding AI ideas in Geoje on a Dataiku pilot program.
Data Solution is recognised as a Dataiku partner of the year for Asia Pacific and Japan, 2026.
Andrew Boyd is announced as SVP and General Manager APJ, overseeing Korea, Japan, ASEAN, Australia and New Zealand.
A months long program with POSCO's DX Strategy Office ends in a hackathon by about 30 engineers, with Dataiku designing, mentoring and judging.
The company holds its Summit in Seoul, a flagship series also run in Paris, Tokyo and London.
Dataiku came to Korea as part of its move into Asia after Europe and North America. Its co-founder and CTO said in Seoul in July 2025 that the company had entered about three years earlier because Korean demand for AI was large; a Korean newspaper profile in 2026 dates the start of the business to 2021. The reasoning that runs through the company's Korean statements is that Korea's big companies want AI and also want it controlled. The APJ head said in May 2026 that Korean chief executives are very open to AI and face strong demands to comply with regulation. Korean manufacturers and financial firms often run systems on their own servers or closed networks, and Dataiku installs there: Samsung Fire & Marine chose it on condition that it run on a closed network, and was moving toward the cloud in 2025.
a system that can scale and control
김종덕 · Country Manager, Dataiku Korea (한국 지사장) · Press conference ahead of Dataiku Summit Seoul 2026, reported in Korean; translated by Signet · 2026-06-24
Korean customers started in manufacturing and spread. LG Chem and POSCO are named among the first, and Doosan Enerbility built a model forecasting electric furnace steel output at about 98% accuracy that won an IDC Future Enterprise award in 2023. By July 2025 the country manager listed banks, insurers, card companies, retailers and advertisers among newer customers. Two programs show how Dataiku gets in the door at large manufacturers: a 2025 hackathon at Samsung Heavy Industries in Geoje, run on a Dataiku pilot program with about 20 teams, and a months long AI program with POSCO's DX Strategy Office in 2026, in which about 30 engineers built models for equipment anomaly detection, quality assurance and process optimisation, with Dataiku designing, mentoring and judging.
The public record identifies commercial leaders and local functions. 김종덕 joined as country manager in November 2024; in July 2025 he said the team had doubled, the business had grown 200% or more year on year, and the new hires were in marketing, partner management, post sales support and data science. Korean press also shows a solutions engineering lead, a sales executive and a senior solutions engineer. In May 2026 the APJ head said Korean revenue was growing at double digits. Deals start with ideation workshops for decision makers, then a prototype, a proof of concept and a contract. Partners carry part of the delivery: Data Solution was recognised as a Dataiku partner of the year for Asia Pacific and Japan in 2026, and the APJ head wants fewer, deeper partners in each market. Dataiku held its own Summit in Seoul on 25 June 2026, part of a flagship series also run in Paris, Tokyo and London.
On 8 October 2026 Dataiku's own board listed 19 open jobs worldwide and one in Korea: Enterprise Account Executive, Seoul, first published on 16 January 2026. The posting gives the whole sales cycle to one person, with marketing, product management and engineering behind them, and lists revenue from existing customers near the top. It asks for 5 to 7 years of software field sales or other relevant sales experience, a record of selling new concepts to large enterprises and the ability to build a shared vision with C level buyers, and prefers a background in big data, analytics, data science, BI and data warehousing or data integration. In Tokyo the same board splits two Enterprise AE roles by industry, financial services and retail with telecoms; the Seoul role is written for the whole market.
On the fear of a satellite office: Korea has had a country manager since 2024, a team that doubled, named customers in chemicals, steel, energy equipment, insurance and advertising, and its own Summit in 2026. On the fear of a stale posting: the role has been public since January 2026 and was live on 8 October 2026. Ask in the first call whether it is a new seat or a backfill, and which accounts come with it.
Current Signet catalog · 2026-10-19
Prepare a full-cycle enterprise deal with champions, trials and expansion.
Dataiku presented one orchestration layer for data, models, agents and governance.
He leads global sales, customer success and partnerships, reporting to the CEO.
Agent Management was announced for October general availability; Headless connects developer coding tools to governed projects.
Read against the thesis, the last year has three threads. The first is the product. In March 2026 the company relaunched itself as the Platform for AI Success, an orchestration layer over other vendors' clouds, models and agents. Cobuild followed in June, and in September it unveiled Agent Management, with general availability announced for October 2026, which counts, measures and flags agents built anywhere. Each step raises the same bet: if agents spread across many vendors, someone has to govern them all, and Dataiku wants that job. In a September 2026 interview, as the broadcaster summarised it, Douetteau said enterprises are shifting toward owning their AI with open weight models.
The second thread is people. A President and Chief Revenue Officer arrived in June 2026 with a growth mandate, an APJ head in March and an APJ partnerships lead in January, after a Singapore office opened in November 2025. Gartner named Dataiku a Leader for the fifth consecutive time in June 2026. For Korea, this places the country in a region described in a June 2026 interview with the APJ head as one of the company's fastest growing and most invested.
The third thread is money. The October 2025 ARR threshold of more than $350M was restated in June 2026. An October 2025 report described banks preparing a potential US listing, with timing subject to change. Three things will show whether the product bet translates into customer adoption and commercial results.
Signet's reading: the new products and dated commercial appointments give a seller specific questions to ask. Which Korean accounts will evaluate Agent Management and Cobuild, what support will the seller have, and who approves the implementation scope? Connect a customer reference to the workflow and the team that can deliver it.
Food-safety analysts can spend their time on findings while a repeatable pipeline assembles monthly data. Dataiku lets business and technical teams prepare data, build models and agents, and govern their operation.
The AE owns the sales cycle and coordinates Marketing, Product and Engineering around business value and customer issues.
Employer postings were checked on 2026-10-08. Availability can change. Mock applications request a review from Signet.
Follow the company from its vision through products, evidence and people to the Korea team and current roles. Use the table of contents to move within the page, and expand the sections for more context.