Deploying AI as close to the field as possible

Context

Founded in 2014, United Heroes is a B2B SaaS company whose mission is to mobilize and engage employees at client companies through collective, unifying experiences built around topics such as workplace wellbeing, company culture, charitable commitments, and transformation. United Heroes has already supported more than 1,000 clients and mobilized over 3 million employees.

Today the company has around forty employees, based mainly in Paris. Its internal structure is dual: roughly 50% of the workforce is in tech and product — profiles already familiar with AI and using it day to day — while the business teams (sales, customer success, HR & finance) were using it little or not at all.

Challenges

Mathilde Salobir, HR Director at United Heroes, identified a clear pattern: an AI usage gap was widening internally between the tech/product teams and the business teams. On top of that came a heavy workload and a patchwork of tools (Notion AI, Google AI, personal ChatGPT accounts, etc.) that made any collective, sustainable practice difficult.

Three challenges emerged:

• Develop the teams' employability and leave no one behind.
• Free up time and gain day-to-day efficiency with AI.
• Manage change so that the teams buy into the project.

"The point isn't to do more, it's to do better."
Mathilde Salobir, HR Director, United Heroes

The solution: why Teamstarter?

United Heroes chose Teamstarter for one main reason: the collective dimension of AI use. Where the practice remained individual and siloed, Teamstarter made it possible to share uses, best practices, and automations across a team and the wider company.

Internal preparation was thorough:

• Early communication to all teams about the desire to help them grow on AI topics, to spark interest and questions.
• Mobilization of the AI champions already identified internally — both to consult them on the choice of partner and to recognize their role.
• Empowerment of managers, in a supporting, encouraging, and reassuring role, tasked with ring-fencing about 4 hours over 2 months in calendars to identify automatable tasks, take part in training and workshops, and build the first skills / AI agents.
• Prior awareness-building with Mathilde by Teamstarter on the risks of AI (isolation, cognitive overload, CSR considerations) — seen as an essential step to anticipate objections and reassure the teams.

The launch took place in person at United Heroes' offices, a deliberate choice to bring the human element back to a subject that is digital by nature. The teams were curious, attentive, and engaged. Ambassadors were appointed in each team (in addition to the cross-functional champions) to carry the voice of the collective. Five teams were brought on board, covering all 40 employees, and Claude was deployed on every workstation.

The results

The first skills built illustrate the diversity of use cases:

• Sales: personalizing prospecting emails, automating post-meeting reports, tailored call scripts.
• Customer success (CSM): generating client communication kits, analyzing data to prepare quarterly reviews.
• HR: automating the summary of recruitment interview notes, passed on to the next manager so they quickly know what has been covered and what still needs to be explored with the candidate.

The rollout is tracked through weekly news updates on OKR progress. According to Mathilde, the Teamstarter approach made it easier to adopt the tool and quickly put the first skills in place.

Conclusion

United Heroes' experience illustrates one of Mathilde's convictions: in a company where half the teams already master AI and the other half have no access to it, the HR challenge isn't technological — it's about alignment and the collective. By preparing the ground (early communication, mobilizing managers, ring-fencing time, raising awareness of the risks) and choosing a bottom-up approach (where it's the employees on the ground who surface the ideas, projects, or solutions) centered on each role's concrete pain points, United Heroes started a rollout that addresses not just a few insiders but all 40 employees.

And above all, the direction is clear: AI isn't deployed to produce more, but to free up quality time — time the teams can reinvest in what truly matters, such as digging into clients' challenges or deepening the support they provide.