Data preparation is one of the biggest barriers to AI adoption, and when it's done manually, it can take months. This customer story shows how Kantar used Microsoft Copilot Studio to build a team of AI agents that translated documents, validated policies, and organised data automatically. The result: 4,000 artifacts cleaned and structured into 400 policy documents in six weeks. Read the story to see how a structured, agent-driven approach can accelerate your path to AI readiness.
How did Kantar use Copilot Studio to prepare data for AI?
Kantar needed to get a large volume of HR content into shape before it could confidently use AI agents. The team was dealing with 4,000 artifacts in multiple languages that had to be turned into about 400 standardized policy documents in just six weeks.
Instead of treating data preparation as one big, complex project, Kantar broke it into smaller, focused tasks and built a set of AI agents in Copilot Studio to handle each step. For example:
- One agent searched across company files to locate and surface relevant documents for review and categorization.
- A second agent translated all documents into English to simplify development and review.
- A third agent identified duplicate policies and generated a clean, correctly formatted starting version for each policy.
In total, the People Team effectively had a “team of 10 AI agents” working on data prep. These agents cleaned, tagged, and organized the content so it was ready to power an HR advisory agent and other AI use cases.
All of this was surfaced through Microsoft Teams, making it easy for employees to access and use the agents in their normal flow of work.
What business impact did Kantar see from automating data prep?
By automating data preparation with Copilot Studio, Kantar turned a highly manual, low‑appeal task into a structured, AI-assisted workflow that delivered measurable benefits:
- Faster data readiness: The People Team cleaned, tagged, and organized 4,000 artifacts into 400 policy documents in six weeks, a timeline that would have been difficult to meet with manual work alone.
- More consistent HR content: Translation, deduplication, and standardized formatting helped create consistent, accurate policy documents across 60 countries.
- Reduced HR workload: With an HR advisory agent now in place, Kantar expects a significant reduction in routine HR queries, allowing HR professionals to focus more on strategic work such as talent retention, training, workforce planning, and partnering with hiring managers.
- Improved recruitment processes: Simple agents like a job description builder and document finder helped standardize and speed up recruitment activities across different countries.
Overall, automating data prep has helped Kantar reshape how its People function operates, shifting time and attention from manual data work to higher-value, people-focused initiatives.
How did Kantar enable non-technical staff to build and use AI agents?
Kantar’s approach was to democratize access to AI rather than limit it to technical teams. Several elements made this work:
- Copilot Studio as a low-barrier tool: Leaders in the People Team, including those without a technical background, found building agents in Copilot Studio to be intuitive. They could design agentic workflows based on their domain expertise instead of coding skills.
- Grounding agents in organizational knowledge: Each agent was connected to Kantar’s own HR and organizational content, so non-technical experts could focus on what the agent should know and do, rather than how to program it.
- Publishing through Microsoft Teams: By publishing agents directly into Microsoft Teams, Kantar made them accessible in the tools employees already use every day, reducing friction and training needs.
This approach led to a broader cultural shift. According to Kantar’s leadership, the company has effectively “democratized technology” by giving everyone the tools to innovate, regardless of technical background. That shift is helping Kantar grow its capability and capacity, while making work simpler and more human for its people.