Skip to content
AgentsLabs

An AI agent that takes work off your hands.

Customer questions answered within a minute, leads followed up automatically, data that moves itself between systems. Not a chatbot with canned replies, but software that actually finishes tasks — including at half past nine in the evening.

WHAT IT IS

The difference between a chatbot and an agent

A chatbot answers questions from a list. When it gets stuck you get "please contact our customer service". An agent does more: it looks up the order in your order system, checks the return policy, generates the label and logs the follow-up in your CRM.

The difference is access and responsibility. An agent is connected to the systems where the answers actually live, and may perform actions there that you have defined in advance. What it is unsure about, it hands to a human — with a summary, so you do not have to read back the whole conversation.

Want to see what that looks like? The homepage runs a simulated agent console with an example scenario: incoming questions, actions performed, and the moment it escalates to a human.
  • Works on your channels — website, email and WhatsApp Business, in your tone of voice.
  • Knows your business — fed with your own documents, prices and terms.
  • May act, not guess — only actions you approved in advance.
  • Escalates when unsure — complaints and exceptions go to a human, with a summary.
  • Leaves a trail — every conversation and action can be reviewed and audited.

FIVE APPLICATIONS

What we build in practice

Almost every SME has one process that costs too much time and requires too little thought. That is where an agent starts — never with everything at once.

APPLICATION 01

Customer service agent

Answers questions about orders, delivery times, returns and opening hours. Connected to your order system, so with real data instead of generic answers.

View the case

APPLICATION 02

Quotation assistant

Reads incoming enquiries, fills your CRM and drafts a quotation. Your colleague checks and sends — from two days to two hours.

View the case

APPLICATION 03

Lead follow-up

Follows up enquiries automatically when nobody responds, schedules calls and flags which leads are cooling off before they disappear.

APPLICATION 04

Document processing

Extracts data from invoices, packing slips and forms and puts it in the right system. The end of retyping nobody enjoys.

APPLICATION 05

Internal knowledge agent

Answers your own team's questions about procedures, prices and agreements — so not everyone has to interrupt the same colleague.

APPLICATION 06

Appointment agent

Books, moves and confirms appointments in your calendar, sends reminders and reduces no-shows.

HOW IT WORKS

Connected to the systems you already use

An agent is only useful once it can reach the right data. So every project starts with your existing systems, not with a new system you have to maintain on top.

  • CRM — retrieve customer data, create leads, log follow-ups.
  • Order or stock system — look up status, delivery time and availability.
  • Accounting — prepare invoices and check payment status.
  • Calendar and planning — book appointments without double bookings.
  • Your own documents — terms, manuals and price lists as a knowledge source.

WORKS WITH COMMON PACKAGES AND WITH CUSTOM SYSTEMS VIA API OR WEBHOOK

01

Scope it

Together we pick one process with enough volume and few exceptions. That is where an agent pays off fastest — and where the damage is smallest if it disappoints.

02

Connect and feed

The agent gets access to the systems and documents it needs, and nothing beyond that. We record which actions it may perform itself.

03

Shadow period

At first the agent only proposes; your colleague approves. That shows exactly where it is good and where it is not yet good enough.

04

Release and measure

What demonstrably holds up, it may do on its own. We measure how much it handles, how often it escalates and what that saves you.

LIMITS AND CONTROL

Humans stay accountable

We do not build an agent that decides everything by itself. That is technically possible and commercially unwise — one wrongly promised discount costs more than the agent saves.

WHAT THE AGENT DOES AND DOES NOT DO ALONE
Situation Agent Human
Standard question with a clear answerhandles it
Looking up data in a systemhandles it
Drafting a quotation or invoicepreparesapproves
Complaint or unhappy customerflags and summarisestakes over
Discount, exception or goodwilldecides
Legal or medical questiondecides
No data adventures. Your data is not used to train third-party models, and we process no more than the task requires. What exactly happens is recorded in a data processing agreement.

PRICE

Fixed price from €2,950

You get a fixed price up front for the whole project: integrations, training the agent, documentation and handover. On top of that you only pay the monthly usage cost of the model and hosting.

STEP 1 — ORIENTATION

AI scan · €195

We map which process suits an agent, what it would save and what it costs. You get a payback calculation on your own numbers. If you proceed, we deduct the €195 from the project.

Request the scan

STEP 2 — BUILD

Agent from €2,950

Fixed price up front. Includes integrations with your systems, knowledge base, shadow period, documentation and handover. Lead time three to six weeks.

View pricing

STEP 3 — RUNNING

Management from €49/mo

Monitoring, tuning answers and small extensions. Cancellable monthly. Model usage costs are charged separately and transparently.

Costs in detail

FREQUENTLY ASKED

About AI agents

Will my customers notice they are talking to an agent?

That is up to you. We advise being transparent — it builds trust and prevents irritation. In practice customers care more about a fast, correct answer than about who gives it, as long as reaching a human stays easy.

What if the agent says something wrong?

That is why it shadows first: it proposes, a colleague approves. Only what demonstrably holds up may it do alone. For sensitive topics — complaints, discounts, legal questions — it always escalates.

Does this work for a small company?

Yes, provided there is volume. At ten questions a week an agent does not pay for itself; at a hundred it does. In the AI scan we calculate that on your numbers, and if it does not add up we say so.

Does my data stay mine?

Yes. Your data is not used to train third-party models and the agent only gets access to what it needs. We record this in a data processing agreement.

What does it cost per month to run?

The model charges per processed message. For a customer service agent with a few hundred conversations a month that is typically between €20 and €80 per month, plus optional management from €49. Actual usage appears on your invoice.

Can the agent take on more tasks later?

That is usually the plan. We start with one process, measure what it does and expand from there. Every extension gets its own fixed price up front.

What would an agent save you?

Tell us which process costs too much time. In 30 minutes you will know whether an agent is the right answer — and if not, what is.

Book a free consultation info@agentslabs.net