Summary

• Paris-based Mendo has raised €12m in Series A funding led by Ventech and Educapital, with Tomcat and OVNI also participating.
• The company’s platform helps large organisations drive adoption, analytics, and governance around tools including Microsoft 365 Copilot, ChatGPT, Gemini, Mistral AI, and internal AI agents.
• The round reflects a shift in enterprise AI spending from licence purchase and pilots towards workflow change, employee adoption, analytics, and measurable usage.

Mendo has raised €12m in Series A funding to expand its enterprise AI adoption platform across Europe, as large organisations move from buying generative AI tools to working out whether employees are using them often enough, safely enough, and well enough to change daily operations.

The Paris-based startup’s round was led by Ventech and Educapital, with Tomcat and OVNI also participating. The new funding follows a €3.5m seed round in October 2024 and will support product development, sales expansion, analytics, and work around agentic AI deployments.

Founded in 2021 by Quentin Amaudry and Alexandre Pinon, Mendo builds software that sits inside the AI tools employees already use, including Microsoft 365 Copilot, ChatGPT, Gemini, Mistral AI, Claude, and internal AI systems. Rather than adding another standalone application, the platform guides workers towards role-specific use cases, tracks adoption, and helps organisations identify where agents or structured AI workflows may be worth deploying.

Mendo says its platform is used by more than 100 large organisations, including PwC, Novo Nordisk, Crédit Agricole, Groupe Rocher, and Edenred. Its product materials describe more than 200 use cases and a user NPS of 64, while Ventech says the company has supported nearly 100,000 employees across Europe.

Many employers are now discovering that AI adoption is not solved by procurement. Productivity suites have added copilots, departments have experimented with chatbots, and individual teams have found pockets of value, but usage often remains uneven. Employees may use AI for summarisation, drafting, translation, or search, while more valuable changes to process design, knowledge work, customer operations, or internal administration remain limited to early adopters.

Mendo is selling into the space between licence activation and operational change. Its software treats AI adoption as an ongoing management problem, with attention paid to user behaviour, training, repeatable workflows, analytics, and feedback from the people using the tools. That approach fits a market where companies need clearer evidence that AI is changing how work gets done, rather than sitting in a software bill as another underused productivity feature.

Model capability is only one part of that equation. A company can buy access to strong AI systems and still see weak returns if employees do not know when to use them, managers cannot identify productive use cases, or governance teams slow experimentation because usage is opaque. One-off training sessions may help employees understand the tools, but they rarely provide enough structure for departments to redesign work around them.

Agentic AI adds another layer of difficulty because systems that act across workflows require more control than individual chatbot use. A sales assistant that drafts follow-up emails, a procurement agent that compares suppliers, or an internal service agent that handles routine requests needs rules, permissions, monitoring, escalation routes, and evidence that it is doing useful work. Without those controls, agent deployment risks becoming either too timid to matter or too loose to govern properly.

Mendo’s analytics layer addresses that operational gap. Large organisations need to know which teams are using AI, which use cases are spreading, where training is weak, and which processes might justify deeper automation. Those signals can help management decide whether to expand licences, invest in agents, update policies, or stop funding tools that employees are not using.

Usage data still needs careful interpretation. A higher number of prompts does not automatically mean better output, and daily active usage is not the same as productivity. Organisations will still need to assess quality, time saved, employee trust, risk, and whether AI use improves outcomes rather than simply shifting effort from one interface to another. Adoption software can make usage visible, but it cannot replace management judgement or clear process ownership.

European buyers also face a more demanding governance environment than early AI pilots suggested. Deployments now sit alongside the EU AI Act, GDPR, sector regulation, works councils, labour consultation, and rising scrutiny of software suppliers. Platforms that track employee AI usage and help deploy agents will need to handle privacy, transparency, and internal accountability carefully, particularly where analytics are used to make decisions about teams or individual behaviour.

Mendo’s growth will depend on whether large organisations continue to treat AI adoption as a persistent operating discipline rather than a launch-phase obstacle. As generative and agentic systems move deeper into office workflows, the harder question is no longer whether the tools are available. It is whether they can be used consistently, safely, and productively across the organisation without turning adoption into another layer of managerial guesswork.