AGENT · 01
HR Agents
Automate onboarding, leave management, and personnel queries
Trusted by Leading Organizations
- ANWB
- Wolters Kluwer
- Belastingdienst
- Evangelische Omroep
- Currence · iDEAL / Wero
- TravelEssence
- AMI Kappers
- InShared
- Heijmans
- Act Legal Netherlands
- GLS
- Trimbos Instituut
- Milence
- Lepaya
- Spril
- Monarch
A · What we do
Four pillars.
The strategic partner for adopting and implementing AI Agents: from strategy and moonshots to data foundation, governance, pragmatic building and training.
N° 01
Strategy & Moonshots
What does your organisation look like when AI Agents do the heavy lifting? We explore moonshots and decide where people add the most value in the process.
N° 02
Data foundation & Governance
The right data foundation in Fabric, BigQuery, Snowflake or Databricks, governance that works and an agent register with a passport per agent.
N° 03
Build & Integration
Agentic harnesses, collaborating agents and decision models alongside generative AI, built pragmatically inside your own systems.
N° 04
Tools & Training
Adoption of Microsoft Copilot, Copilot Studio, Claude Cowork and Microsoft Foundry, with training and AI Champion programmes.
P · Every agent a passport
Next to your HR register, an agent register.
Just as every employee is in the HR register, every AI Agent deserves a digital passport: who owns it, which model and service does it run on, how is its output evaluated and which data sources can and can't it access?
Agent passportTAG · ID
AGT-0427
Name
Intake agent
- Owner
- Customer service lead
- Model
- Pinned per version
- Runs on
- Own cloud environment
- Last update
- After every change
- Evaluation
- Output checks + sampling
Access rights (CRUD)
CRM
R · U
SharePoint
R
Invoices
—
P<TAG<<INTAKE<AGENT<<<<<<<<<<<<<<<<<<<<<<<<<<
B · AI Operating System
AI as an Operating System.
We build the operating system for your organization — where humans and machines come together.
FOR PEOPLE01
AI
Tools
— Empowering people to work optimally
Give your team superpowers with AI Tools that seamlessly integrate into their daily workflow. From Claude CoWork for smarter collaboration to custom tooling that simplifies complex tasks.
- Claude CoWork
- Custom AI Tooling
- Workflow Integration
FOR PROCESSES02
AI
Agents
— Making processes work optimally
Autonomous AI Agents that run your business processes — from intake to reporting. They work 24/7, scale effortlessly, and get smarter through your own business data.
- Multi-Agent Systems
- Process Automation
- RAG Pipelines
— A BOUTIQUE STANCEFIG. 03 · MANIFEST
We don't write
prompts.
We build systems.
01
10x mindset
Engineers who listen like consultants and ship like consultants.
02
Beside you, not for you
No black box. We code alongside you, inside your stack.
03
From sprint to stack
Low-code where it fits, high-code where it counts.
50+
AI projects in production
10+
Years of software experience
1 day
From idea to proof-of-concept
300+
Organizations trained via ai.nl
D · The teamFIG. 04 · INDEX — TEAM
The people behind the system.
A boutique of consultants. One philosophy: stay seated next to the client.
FIG. 01B&W · DSLR · 35MM
FIG. 02B&W · DSLR · 35MM
FIG. 03B&W · DSLR · 35MM
FIG. 04B&W · DSLR · 35MM
FIG. 05B&W · DSLR · 35MM
FIG. 06B&W · DSLR · 35MM= ATELIER WITH CHARACTERTHE AUTOMATION GROUP · 2026
D · AI Agents
AI Agents.
Smart systems that don't just answer questions but autonomously execute actions across all your digital tools. From databases and ERP systems to Salesforce, HubSpot, and Excel.
FIG. 04Modular by design — compose to fit
AGENT · 02
Marketing Agents
Generate content, analyze campaigns, and personalize communication
AGENT · 03
Intelligence Agents
Collect, analyze, and combine data from multiple sources for strategic insights
AGENT · 04
Email Agents
Process, categorize, and intelligently respond to emails
AGENT · 05
Compliance Agents
Monitor regulations, check documents, and ensure compliance
AGENT · 06
Security Agents
Detect threats, monitor systems, and proactively protect your digital environment
Model selection & token optimisation
FIG. 05More with fewer tokens
Token costs are spiralling. We know how to keep them in check.
Once agents hit production, your LLM bill explodes. Per-role model selection, context engineering, prompt caching and hard token budgets per run — we configure your tools and agents to get better output at a fraction of the cost.
- · Frontier vs. fast/cheap per task
- · Context trimming — only what's needed
- · Prompt caching, batching, routing
- · Token & cost monitoring per agent
Request
Task arrives
A user question or a trigger from a system.
Routing per role
Frontier model for reasoning, fast and cheap model for routine work.
Context
Trim the context
Only send the parts the model actually needs.
Prompt caching
Reuse repeated instructions and documents instead of resending them.
Execution
The agent works
Call tools, take steps, deliver the result.
Token budget per run
A hard limit per run, so one runaway task doesn't blow up your monthly bill.
Control
Monitoring per agent
Tokens and cost visible per agent and per task.
Adjust
Tune model selection and context on what the numbers show.
E · Why us
Why The Automation Group.
01N° 01
AI Agents That Actually Work
Smart systems that don't just answer questions but autonomously execute actions across all your tools: from ERP and CRM to Salesforce, HubSpot, and Excel.
02N° 02
Agentic Consultants with a 10x Mindset
Our Agentic Consultants are technical consultants who leverage hyper-effective AI agents to do the work of entire teams. No large project teams, but surgical precision with maximum impact.
03N° 03
Data Security Guaranteed
We work together with your IT team. Data stays within your own systems, accurate and cost-efficient. Plus: training and knowledge transfer so you can continue independently.
STACK · WHAT WE BUILD WITH
- 01Microsoft Copilot Studio
- Copilot Studio agents
- Microsoft 365 Copilot
- Teams & SharePoint
- Dataverse
- Power Automate
- Azure AI Foundry
- 02Build
- Claude Code
- Claude Agents
- OpenAI Codex
- Cursor
- Claude CoWork
- GitHub
- 03Orchestration
- LangGraph
- CrewAI
- Pydantic AI
- n8n
- Temporal
- FastAPI
- 04Memory & Data
- Weaviate
- Qdrant
- Mem0
- Postgres / pgvector
- Supabase
- Obsidian
- 05Cloud & Models
- Azure
- AWS
- Anthropic Claude
- OpenAI GPT
- Google Gemini
- OpenClaw (on-prem)
PRAGMATIC AI ADOPTION · AI CHAMPIONS
We accelerate your AI ambitions — pragmatically, not theoretically.
AI succeeds or fails on adoption, not on technology. We run AI Champion programmes, train AI Champion Leads per department and make sure agents land inside the systems you already use — Microsoft Copilot, Teams, SharePoint, your ERP and CRM. That turns AI into daily work instead of a pilot that stalls.
- 01 · AI Champion programme
- 02 · AI Upskilling by role
- 03 · Agents in existing systems
CORE EXPERTISE · MICROSOFT COPILOT STUDIO
Copilot Studio agents — production-ready, inside your Microsoft tenant.
For organisations fully invested in Microsoft 365 and Azure, we build agents that live where the work happens: Teams, SharePoint, Outlook and Dataverse. We combine Copilot Studio, Power Automate and Azure AI Foundry with custom Python orchestration where needed — governance, DLP and Purview properly configured.
- 01 · Agent design & topics
- 02 · Graph & Dataverse connectors
- 03 · Governance, DLP & rollout
Frequently asked questions
What clients ask us
The most common questions about Agentic AI, our way of working and collaboration.
What is Agentic AI and how is it different from a chatbot?+
A chatbot answers questions. An AI Agent performs tasks. Agentic AI combines language models with tools, memory and reasoning steps so an agent can autonomously take actions in your systems — processing emails, analyzing data, creating orders — within clear boundaries and with human oversight where needed.
How fast can we see a first result?+
For low-code projects we typically have a working proof-of-concept on your data within 1 to 3 weeks. For more complex multi-agent systems or legacy integrations we work in 2 to 4 week iterations, each delivering demonstrable functionality.
Is Agentic AI safe to use with our company data?+
Yes. At The Automation Group we default to EU hosting, role-based access, audit logging and — where required — on-premise solutions inside your own network. Your data stays yours and only leaves your environment when you explicitly want it to.
Will agents replace our employees?+
No. The Automation Group builds agents that take over repetitive, rule-based or administrative work, freeing your team for strategy, customer contact and creative work. Human-in-the-loop is our default: humans remain accountable for high-impact decisions.
Which tools and models do you work with?+
We are model-agnostic: OpenAI, Anthropic Claude, Google Gemini and open-source models like Llama. For orchestration we use LangGraph, CrewAI, Microsoft Copilot Studio, n8n and custom Python. Copilot Studio is a core expertise when clients already run on Microsoft 365 and Azure and want agents deeply integrated into Teams, SharePoint and Dataverse. The stack we choose depends on your requirements around cost, performance, privacy and existing infrastructure.
Continue reading
Discover more about our approach and solutions.
Start the conversation
Explore How We Can Accelerate Your AI Ambitions?
AI technology only works when people actually use it. We ensure that everything we build is practical, usable, and aligned with your business.