AI that belongs to everyone: How Versent embeds AI into everyday work

According to HubSpot CEO Yamini Rangan, 90% of companies use AI, yet only 6% achieve transformational results, revealing a significant gap. AI becomes valuable when people stop treating it as a separate experiment and start using it to improve the work they do every day.

So how did Versent manage to successfully embed AI into the everyday lives of it’s 620+ employees?.

---

From AI ambition to everyday practice

At Versent, we have rolled out tools, explored AI-native ways of working, tested agents and invested in capability across the business. Along the way, we have seen why early enthusiasm is not the same as lasting adoption — and why scaling AI requires more than choosing a model.

Our approach is straightforward: learn what works, build the capability to use it well, and engineer the foundations to scale it responsibly. We call this being Customer Zero.

---

Start with confidence, not just technology

The first step in our journey was to explore and prepare.

AI creates new possibilities, but it also raises important questions. What information should people share with an AI tool?. Where is human judgement essential?. How do we balance the freedom to experiment with the responsibility to manage risk?.

Our approach started with risk assessment, education and policy definition which was supported by an AI Governance Council which we established, with representation from across the business, not just our technology focused teams.These foundations gave people a clearer understanding of how to engage with AI and establish boundaries for its use.

The aim is not to constrain innovation or experimentation, it’s to make participation more confident. When people understand both the possibilities and the expectations, they are better placed to identify useful opportunities and explore them responsibly.

For us, AI readiness is as much an organisational and operational undertaking as it is about technical access.

---

Learn what works — including what does not stick

Our early experimentation targeted specific capabilities across the business. While initial interest was strong, some tools did not become a lasting part of people’s everyday work, and over time many returned to familiar ways of working.

We learned early that it is okay to recognise when an approach is not working as intended. A successful launch does not automatically create lasting change: access and enthusiasm are only part of the story. The harder work is making AI useful for recurring tasks and helping people build the confidence to use it consistently.

The unsuccessful experiences helped us distinguish hype from meaningful utility and adapt our approach. Because the technology keeps changing, we revisit opportunities as new capabilities and use cases become viable. What does not work today can still teach us where to focus next.

---

Put business value at the centre of experimentation

The people closest to the work often have valuable ideas about where it could improve.

We used one of our internal hackathons to focus on AI use cases, encouraging teams of 3 to identify opportunities within our own business and scoring ideas on technical viability, as well as business value. This made the experiments more feasible, they weren’t just tech demos but applications of AI that would genuinely have meaningful business impact.

We invested in AI enablement for our senior leaders. As the pace of change accelerated, leadership fluency across all business domains became an important part of helping the organisation make informed decisions.

To support a more repeatable path from idea to implementation, we developed an AI Experiments framework. It enables ideas to be submitted from across the business, prioritised, prototyped, tested, reviewed, and if successful, developed and deployed into the business. This framework gives our people’s amazing ideas somewhere to go, and a path towards realisation, whilst maintaining discipline and appropriate controls.

---

Build the foundations to scale what works

Scaling AI is not simply a matter of extending every pilot to more people. It requires decisions about shared platforms, knowledge, support and ways of working.

An example of this was when we selected Notion as our company-wide knowledge management and AI platform, bringing a significant change-management effort and an opportunity to review our data sources as we migrated to a single platform. This took a major cross-functional effort to make sure the information feeding our new context layer was consistent, current and clear, so that when our people pointed AI at it or built agents on top of it, they could trust the outputs.

Notion provides a foundation for everyday searching, summarising and drafting. Now Claude supports our team to complete more complex analysis, asset creation, engineering and delegated tasks. The focus was to build a complete ecosystem of supported tools for our people, and enable them to use them effectively in the work that they do, regardless of where in the business they sit.

In engineering, we extended beyond GitHub Copilot to Claude Code, giving our engineers options to select their preferred tools to do their work. To ensure best practice adoption and alignment across this group, we drove training through Anthropic’s Claude Partner network.

The human support network is equally important. Our AI Champions program, supported by the wider AI Guild, bring together people from across the business as points of contact for ideas, adoption and training.

---

Treat AI economics as part of AI fluency

As AI moves from fixed-price access towards token-based consumption, the economics become part of the design challenge. We have seen the commercial model of tools we use shift multiple times this year, and it’s important to be ready to adapt.

We have built frameworks to monitor and manage consumption by agents and people. Just as importantly, our fluency work helps people get the most out of the tools they have, being aware of context creep, model routing best practices, when a task calls for an agent vs a shared skill.

All this was informed by usage patterns and research with our people, allowing us to shape our tools to fits the way our people saw themselves operating with AI. This is a continuous improvement process, we have observability to learn where we can optimise how we support our people, and even look at where people may be turning to other tools to understand why. These insights all help to assess the fit of the tools, frameworks and training we provide and understand adoption and return on investment.

---

Bring real-world learning to customers

Being Customer Zero gives us a practical basis for the conversations we have with customers. The lessons we learn, and mistakes that we make are experience we can share with our customers who are starting down that same path.

Versent brings these lessons together with our engineering, data, security and delivery capabilities to help organisations move from experimentation to trusted, repeatable ways of working.

If your organisation is ready to move beyond experimentation, start with a real problem and the people closest to it. Versent can help you turn that starting point into AI that becomes a useful part of everyday work.

---