AI agents
AI agents that workalongside your team.
For companies that want to delegate repetitive tasks in email, CRM and documents: we build AI agents that find the information, use only the tools you approve and see the task through. You decide what they can do on their own and where a person confirms.
From answers to tasks that get done
Many companies have already tried a chatbot that can hold a conversation. The next step is the actual work: searching through systems, filling in fields, passing things on. This is where an AI agent can save the most time.
The same checks, dozens of times a day
Someone looks up an order status, checks a customer in the CRM or confirms stock. Simple tasks, but they constantly interrupt important work.
Information is scattered across several apps
To answer a single question, a colleague opens email, the price spreadsheet, the CRM and a shared document. Context gets lost along the way.
A chatbot that only talks
A classic chatbot gives general answers, but it doesn't create a task, update a record or prepare a quote. For now, those actions stay with the team.
Leads and requests that wait too long
Enquiries that arrive in the evening or at the weekend are picked up only the next day, and qualifying them by hand delays the reply to the customers who are genuinely interested.
The natural need for control over AI
Nobody wants a system that sends emails or changes data without asking. With clear access and approval rules set from the start, control becomes part of the project, not an obstacle.
What an AI agent actually is
An AI agent is a software assistant that receives a goal, follows a workflow defined together with you and uses authorised tools (for example, searching the CRM, creating tasks, reading documents) to achieve it. Unlike a chatbot, it can carry out actions. Unlike classic automation, it can interpret a freely worded request and choose the next step from a set of approved actions. It does not have unlimited autonomy: it does only what you have allowed, within the limits you have set.
A defined workflow
We start from the real process: what triggers the task, which steps follow, when it stops. The agent doesn't improvise a new route each time.
Authorised tools
The agent gets access only to approved actions, through integrations and APIs (tool calling). Each tool has clear permissions: read, write or propose only.
Knowledge from company sources
For accurate answers, the agent consults the documents, databases and internal rules we configure, not just the model's general knowledge.
Human approval points
Actions with impact (an email to a customer, a price change, a payment) can wait for a colleague's confirmation before they are carried out.
Logging and monitoring
Every step the agent takes is recorded: what it received, which tool it used, what it decided. That way you can check, correct and improve its behaviour over time.
Types of AI agent
Each agent has a clear role, approved tools and points where a person signs off. Choose the situation closest to yours.
AI agent for sales
Prepares quotes, follow-ups and replies from your CRM data, while the salesperson reviews them and decides what goes to the client.
DetailsAI agent for customer support
Answers customers from approved information, checks order status in your systems and hands over to colleagues the cases that need a person.
DetailsAI agent for lead qualification
Asks the qualification questions agreed with your team, scores each lead against the agreed criteria and sends it to the right person.
DetailsAI agent for research
Gathers information about the market, competitors or suppliers from approved sources, with citations for every claim, while the analyst validates the conclusions.
DetailsAI agent for document processing
Takes in incoming documents, extracts the data, checks it against your rules and systems and sends it on, with a person handling the exceptions.
DetailsAI agent for financial reporting
Prepares periodic reports from your financial and operational data and flags variances, which the finance team then analyses.
DetailsAI assistant for recruitment
Supports recruiters with CV screening, interview scheduling and candidate communication, without any rejection decisions taken automatically.
DetailsAI assistant for e-commerce
Guides shoppers to the right product using your catalogue, answers questions about availability and orders, and prepares cart recovery messages only where consent exists.
DetailsAI agent for operations
Tracks internal requests, tickets and stock thresholds, prepares tasks for teams and coordinates handovers between departments, with a person responsible for every decision.
DetailsAI agent for data analysis
You ask a question about your company's data in natural language, and the agent writes the query, runs it on authorised sources and explains the result, showing how it got there.
DetailsAI assistant for your internal knowledge base
Employees ask how something is done in the company and get the answer from internal procedures and policies, with a reference to the source document.
DetailsAI agent for workflows
Runs multi-step processes that need decisions along the way, where fixed automation stops at the first exception, with human approval at the sensitive points.
DetailsMulti-agent systems
Several specialised agents, each with its own role, tools and permissions, coordinated in a single process and supervised by a person. Only where a single agent is no longer enough.
DetailsAI voice agents
Agents that answer and handle phone calls, with hand-over to a person.
Dedicated page
What your team can gain
- 01
Fewer interruptions
Routine checks can be taken over by the agent, and colleagues return to the tasks that require experience and negotiation.
- 02
Faster response to requests
Requests can be picked up, classified and prepared as soon as they arrive, even if the final decision stays with a person.
- 03
Processes applied more consistently
The agent follows the same steps every time, which can reduce the variations and omissions of manual work.
- 04
Visibility over repetitive work
The logs show the volume and type of tasks taken over, which helps you decide what is worth automating next.
- 05
Gradual expansion, without major risk
You can start with an agent that only proposes, then grant it more permissions as its results are validated.
The benefits depend on each company's processes, data quality and available integrations. We estimate them together during the analysis stage and do not guarantee them.
What an AI agent could look like in your company
The scenarios below are conceptual examples, presented to illustrate what can be built. They do not describe implementations delivered to clients.
- B2B distribution
An agent for quote requests
It reads the request received by email, identifies the products and quantities, checks stock and the price list, then prepares a draft quote that the salesperson reviews and sends.
Conceptual example - Online store
An order support agent
It answers questions about order status and the returns policy by consulting the e-commerce platform, and opens a ticket for the team when a case requires a decision, such as a refund.
Conceptual example - Professional services
A lead qualification agent
It picks up the forms from the website, adds public data about the company, scores the lead against the criteria set by the team and creates the opportunity in the CRM, with a summary for the consultant.
Conceptual example - Recruitment
An agent for screening applications
It extracts the relevant information from CVs, compares it with the job requirements and prepares a shortlist with reasons, which the recruiter validates. The selection decision remains entirely human.
Conceptual example
What the project can include
Each agent has a precise role. We can build them individually or have them work together, depending on how complex the process is.
Customer-facing agents
- Agents for sales and quote preparation
- Customer support agents, with handover to the team
- Agents for lead generation and qualification
- E-commerce agents: orders, returns, recommendations
Agents for internal operations
- Agents for operations and administrative tasks
- Document agents: reading, filling in, checking
- Agents for recruitment and application screening
- Research agents: market, competitors, suppliers
Agents for data and decisions
- Data analysis agents working on configured sources
- Periodic reporting agents, with explained summaries
- Alerts when certain values fall outside the set range
Architecture and control
- Multi-agent systems, with separate roles and coordination
- Agents with controlled access to tools and APIs
- Tiered permissions: read only, propose, execute
- Human approval for actions with impact
Indicative prices
Packages and prices
Price benchmarks for this service, in euro, excluding VAT. The final offer is set after a conversation about the project.
See all pricesAI assistant on your website
+ from €90/month operation · 2–4 weeks
from€1,500
- Answers from your documents and pages
- Takes requests and passes them to the team
- Installed on the website, WhatsApp or Messenger
- Monthly monitoring and improvement
- Recommended
Automation
n8n or Make · 1–3 weeks
from€900per workflow
- Analysis of the process and the data
- Connecting the applications you already use
- Parallel testing before going live
- Documentation and training
- Optional operation, from €150/month
AI voice agent
+ per-minute costs · 4–6 weeks
from€2,900
- Handles calls, bookings and frequent questions
- Transfers to a person when the situation requires it
- Integration with your calendar or CRM
- Disclosure and consent in line with the law
AI provider costs (OpenAI, Anthropic, telephony) are billed on consumption, transparently. The online AI assessment is free.
How we build an AI agent
Step 01
We choose the right task
We identify a repetitive process with fairly clear rules and enough volume, where an agent can add value without major risks.
Step 02
We map the workflow and its limits
We set out the steps, the tools needed, what the agent can do on its own and what must be approved by a person. We also note what it will not do.
Step 03
We build a first working agent
We connect the data sources and approved tools and test on real cases, usually in the mode where the agent only proposes.
Step 04
We evaluate its behaviour
We check the answers and actions against a set of test cases: accuracy, compliance with the rules, situations where it should have stopped.
Step 05
We launch in a controlled way
We start with a small group of users or a limited volume, follow the logs and make adjustments before expanding.
Step 06
We improve based on the data
We regularly analyse where the agent got things wrong or asked for help and adjust the instructions, tools or escalation rules.
It connects to the tools you use
An agent is only useful if it can work where your data lives. We connect it through APIs and integrations to your existing applications, with permissions limited to what it needs.
The availability and limits of each integration (API, permissions, licence costs) are checked for each project individually.
- Google Workspace
- Microsoft 365
- Slack
- HubSpot
- Pipedrive
- Salesforce
- WooCommerce
- Shopify
- SmartBill
- Notion
- Airtable
- n8n
- OpenAI API
- Anthropic API
Security & control
Limited autonomy, clear control
An AI agent can make mistakes, like any system. That is why we design it so that mistakes are visible, reversible and, where it matters, stopped before they reach the customer.
Personal data and GDPR
We establish what data the agent is allowed to process, on what legal basis and for how long it is kept. Data sent to external AI model providers is limited to what the task requires, under their data processing agreements.
Minimum permissions
Each tool has separate rights. An agent that only reads from the CRM cannot delete anything, and access can be revoked at any time.
Human approval for sensitive actions
Sending messages to customers, financial changes or decisions about people can be configured to wait for a colleague's confirmation.
A complete log of actions
Every step is recorded, so you can later find out why the agent made a particular choice and correct its behaviour.
What it does when it doesn't know
The agent is instructed to recognise situations that are unclear or outside its role and hand them over to a person, instead of making up an answer.
Useful guides
Before you decide, read
- AI & agentsChatbot or AI agent? The difference that matters for your businessA chatbot answers questions; an AI agent sees a task through using your tools. How to choose between them and where to keep human control.Read the guide
- LLMs & dataWhat is RAG? A simple explanation for AI assistants on company documentsRAG means the AI assistant first searches your documents and only then answers, citing the source. What it solves, what it doesn't and how to evaluate it.Read the guide
- ComplianceThe AI Act explained for businesses: what the EU regulation on artificial intelligence requiresThe AI Act sorts AI systems into risk levels and sets different obligations for providers and users. What it means in practice for a business that uses chatbots or AI agents.Read the guide
- LLMs & dataClaude AI for businesses: what it is, what it's for and how it compares with ChatGPTClaude AI is the family of language models developed by Anthropic. What it can do for a business, the difference between the app and the API, how data is handled and the limits you need to know.Read the profile
Frequently asked questions
What is an AI agent for businesses?
An AI agent is a software assistant that takes over a concrete task in the company's workflow, for example preparing a quote or qualifying a lead, and sees it through using only the tools and data you have given it access to. Actions with impact can wait for a person's confirmation.
AI agent, automation or document assistant: which suits me?
If the steps are always the same and the rules are clear, automation is usually simpler and more predictable. If the team mainly needs answers from company documents, the starting point is an assistant based on an LLM and RAG. An AI agent makes sense when the task requires interpreting a request and using several tools. For phone calls there are AI voice agents. In the initial conversation we decide together on the right option.
What is the difference between a chatbot and an AI agent?
A chatbot holds conversations and provides information. An AI agent follows a workflow and can use authorised tools to carry out actions, for example creating a task in the CRM or preparing a quote. Even so, the agent acts only within the limits you set.
Can the agent make decisions on its own?
Only small, well-defined decisions that you have approved in advance. For actions that affect customers, money or people, we recommend a human approval point. We set the level of autonomy together, and you can change it over time.
What happens if the AI doesn't know the answer?
The agent is configured to recognise situations where it doesn't have enough information or the request falls outside its role. In these cases it hands the conversation or task over to a colleague, with a summary of the context, instead of guessing.
How is the data protected?
The agent has access only to the sources and tools it needs, with minimum permissions. Together we decide what data can reach AI model providers and on what legal basis, and we keep a log of actions. GDPR compliance is reviewed for each project together with your data protection officer.
How do I measure an AI agent's effectiveness?
We define the indicators before launch: how many tasks it takes over, how often the result is accepted without corrections, how often it asks for help, the time to first response. We compare them with the situation before and track them regularly from the logs.
Will the AI agent replace people on the team?
The aim is not to replace people but to take over repetitive tasks. Decisions that require judgement, empathy and experience stay with the team, which gets more time for them.
How much does developing an AI agent cost and how long does it take?
It depends on the complexity of the workflow, the number of tools and integrations, data quality and the level of control needed. An agent with a single role and standard integrations is much simpler than a multi-agent system. After an initial conversation we can propose an estimate for your case.
Can’t find the answer? Write to us and we’ll reply directly.
Talk to us about an AI agent for your company.
Tell us which task takes up most of your time. We will assess whether an AI agent is a good fit, what it should be able to do and where human approval is needed.
Related services
- LLM & custom AIAI assistants that answer from your company's documents and data, with source references and controlled access.
- AI automationAutomated workflows between the apps you already use, with AI where it helps and human review where it matters.
- AI voice agentsCalls answered outside working hours too, appointments confirmed and conversations summarised straight into the CRM.