Building an AI-Native HR services company

Client
Industry
B2B HR Services & Workplace Wellbeing

Scaling expert knowledge across 750+ service employees while maintaining quality

Attentia is Belgium’s leading integrated HR services provider, combining payroll administration, HR and compliance services, and workplace wellbeing expertise under one roof.

With approximately 750 employees serving 15,000 client companies across Flanders, Brussels, and Wallonia, Attentia handles everything from contract administration to occupational health checks, risk analysis, and psychosocial support.

Their mission: help organizations and their people "feel good, work great" by streamlining HR administration and proactively safeguarding employee health and safety through their proprietary DOTS platform.
opportunity

For Attentia's Service Employees

Complex, regulation-heavy advice across legal, payroll, and benefits required deep expertise
Inconsistent service quality when junior consultants could not access senior know-how
Slow onboarding as new hires had to absorb decades of agreements, rules, and edge cases
Repetitive questions drained time from higher-value advisory work
Less time for proactive services due to constant reactive workload

For Attentia's Clients (HR Managers & Employees)

Delayed responses to urgent HR questions as they waited for the right expert to become available
Inconsistent advice depending on which service employee they reached
Complex processes for routine requests like contract changes or capacity adjustments
Lack of proactive support: no dashboard insights or suggestions on absenteeism, payroll optimisations, etc.
solution

An internal-first AI “sidekick” and workflow automation platform

An internal-first AI journey

An AI Sidekick for HR, legal, and payroll experts
The Sidekick is integrated into the DOTS portal across email, chat, and forms
The AI drafts responses to client inquiries to speed up and standardize case handling
Experts review and approve every answer to ensure accuracy and build trust

A Human-in-the-loop workflow

Client inquiries triggered AI-generated drafts
AI retrieved internal knowledge and past cases
Applied multi-step reasoning
Included source references
Experts reviewed, corrected, and sent responses

Reinforced learning through human feedback

Expert edits were captured as feedback
System improved with every interaction
Senior knowledge became widely accessible
Increased consistency across responses

Interactive Dashboards

Brings the most important HR signals into one clear, interactive view
Teams can filter and segment insights by role, team, location, and time period to spot patterns fast
Makes it easy to move from high-level trends to the underlying drivers with drill-downs
Turns insights into action with shareable reporting and consistent metrics across stakeholders
Includes dedicated dashboards like Absenteeism, to track trends, recurring patterns, and follow-up progress in one place

Agentic Workflows

Expanding toward full workflow automation
Key use cases;
Onboarding (contracts, compliance, setup)
Capacity changes (calculations, approvals, execution)
Absence management (registration, follow-up)
Built with approvals, audit trails, and integrations

Omnichannel expansion

Expanded beyond the DOTS portal
Added WhatsApp and Microsoft Teams
Further channels in development
Consistent HR support across all touchpoints

Composable architecture

We built a multi-agent system on the Nimble AI Service OS, tailored to Attentia’s service workflows. A supervisor agent routes tasks to specialized agents for knowledge retrieval (RAG), legal, payroll, and workflow execution, while maintaining context. The system is grounded in structured knowledge with semantic chunking, metadata filtering, and embedded expert feedback. It integrates with DOTS, Salesforce, Evolution Apps (via computer use), and Microsoft 365, and is accessible via web, WhatsApp, and Teams.

The architecture emphasizes modularity and continuous improvement. Stateless agents and strict tool contracts enable safe iteration and flexibility. Observability is handled through LangFuse and OpenTelemetry, combining automated evaluations, user feedback, and test sets. Computer use enables automation on legacy systems, while a reusable procedure engine supports scalable workflow automation, forming the foundation of a growing AI service OS.

Outcomes

Delivering real value

95%
accuracy for payroll 
& compliance explanations
30+
NPS for SME users of the beta
750+
empowered HR & wellbeing experts
For Attentia’s service employees
The system led to faster case resolution through AI-assisted drafting and immediate access to institutional knowledge. It increased consistency in advice, as consultants could rely on senior-level expertise embedded in the system. New hires onboarded more quickly, gaining instant access to the collective knowledge base. At the same time, routine questions were handled more efficiently, allowing experts to focus on higher-value, complex work.
For Attentia's Clients
The experience improved through faster and more accessible support. Common HR questions could be answered instantly, across multiple channels and at any time. The quality of responses remained consistent, regardless of the expert or channel involved. Clients were also guided more proactively through complex procedures, with clear next steps and structured support.

“This project proved that becoming AI-native starts internally. Together with Nimble, we built a system that makes our experts stronger and our service more consistent, while staying rigorous on governance and auditability.”

Catherine De Baets, CIO Attentia

Why internal-first AI?

You de-risk fast.
You create a compounding feedback loop.
You standardize quality.
You build the operating foundation first.
You earn the right to go omnichannel.