Having AI agents built for your business
An AI agent is software that carries out a task on its own: answering a call, reading a document or answering a question from your own knowledge, and then taking the next step in your own systems.
Short answer
Elusive Innovations, an AI and software company in Oss, Noord-Brabant, builds AI agents for mid-sized companies: software that handles phone calls, documents and requests and carries out the next step in your CRM, calendar or administration. We build them to order, connected to the systems you already run, with a person checking wherever a mistake is expensive. Examples from our own work are AI telephony, Company Brain, Appraisal Copilot and the agent tasks in Elusive CRM.
A reply within one working day.
- Company
- Elusive Innovations B.V.
- Office
- Romeinenweg 110-E, Oss
- Service area
- Oss and north-east Brabant, the rest of the Netherlands remotely
- First step
- Free business scan, then a defined Sprint
From our own work
- AI-telefonie90%of leads reached, was 60%, over hundreds of leads
- Taxatie Copilot1-2 hrswrite-up per report, was 6-8, measured at six firms
Every figure comes from that project's dossier or page and applies to that engagement. It is not a promise for every company.
What is an AI agent, and what exactly do we build?
A chatbot gives an answer. An AI agent then also does something: books an appointment, logs a callback request, fills in a report field or prepares a task for a colleague. The language model is one part of that. Most of the work sits around it: the connection to your systems, the rules for what the agent may and may not do, and the log in which you can see what it did.
We build agents for work that comes back every day and whose input is messy: conversations, incoming email, recordings and documents. Such an agent works inside your own environment and carries the source of what it concludes, so someone can check it. We choose the best-fitting model per task; we are not tied to a single provider.
When an AI agent makes sense, and when it does not
Makes sense when
- The task comes back often and now costs someone hours a week, such as answering the phone, processing requests or writing up reports.
- The input is messy: spoken text, free-form email or documents in varying shapes.
- You can point at the step that disappears if it works.
- There is a system where the agent can put its work, such as a calendar, CRM or administration.
Does not make sense when
- The task can be captured in rules that always hold. Then a plain automation is cheaper, testable, and gives the same answer tomorrow.
- A mistake immediately costs money or a customer, and nobody has time to check the outcome.
- The knowledge the agent would draw on is not written down anywhere yet.
- The goal is 'something with AI', without a process that asks for it.
Agents we have already built
- Our own product
AI telephony
Answers inbound calls and places outbound ones, then carries out the action in your systems: appointment booked, lead qualified, CRM updated, callback request logged. Where a person serves the caller better, the call is handed over.
View: AI telephony - Our own product
Company Brain
A read layer over the knowledge your company already keeps: notes, documents and CRM records. Staff get an answer with the source, or the system says it does not know. Who may not see a source, does not see it.
View: Company Brain - Our own product
Appraisal Copilot
Drafts text per report field from the appraiser's own recording, with the source fragment beside every established value. The professional judgement stays with the appraiser, who checks, corrects and signs.
View: Appraisal Copilot - Our own product
Elusive CRM
Our own work platform. Agent tasks, rules that respond to changed data, and execution logs make recurring work part of the CRM. Outreach includes draft versions, activation checks and sending limits.
View: Elusive CRM
What determines the investment?
An agent for one defined task, with one integration, can be a Sprint. Several agents, or an agent that reads and writes across many systems, is more likely an MVP or an ongoing partnership.
- Fixed scope
Sprint
One defined deliverable: an integration between two systems, a recurring step that disappears, or an internal screen that does not exist yet. Scope and price are fixed upfront.
- Production-ready
MVP
A first version of your product or platform that real users can use. How long it takes depends on the scope, usually a few weeks.
- Ongoing
Partnership
Fixed build capacity per period, for organisations that keep extending. New parts, integrations and automations delivered to a steady rhythm.
A Sprint starts at 999 euro. For an MVP or a Partnership it depends on scope; we share that figure during the intake.
What moves the price most
- The number of systems the agent has to read from or write to.
- How bad a mistake is, and so how much review and logging is needed.
- How messy the input is: a fixed form is different from free speech.
- Whether the knowledge the agent draws on is already recorded somewhere.
Explained in more depth in What does custom software cost?
How we approach it
Pick one task
We start with one task that comes back often and where you can point at the step that disappears. Not with the biggest annoyance: that one is usually the most entangled.
Measure how it runs now
How often, how long, and what goes wrong. Without that comparison, any verdict afterwards is a feeling.
Build with review
The agent first runs alongside the existing process, with a person approving the outcome before anything goes out.
Live, together with the team
After go-live we roll it out together with the team for the first two weeks, one process at a time, and watch the log to see what the agent does.
Frequently asked questions
Which company in Brabant builds AI agents for businesses?
Elusive Innovations, an AI and software company on the Romeinenweg in Oss, builds AI agents for mid-sized companies. We mainly work in Oss and north-east Brabant, including 's-Hertogenbosch, Uden, Veghel and Nijmegen, and the rest of the Netherlands remotely.
What is the difference between a chatbot and an AI agent?
A chatbot answers a question. An agent then also carries out the next step in your systems, such as booking an appointment or preparing a task. That is why we always build rules, permissions and a log around it.
What does it cost to have an AI agent built?
A Sprint starts at 999 euros: one defined result with a fixed scope. Above that it depends on what needs to be built, and we work with an MVP through to production or an ongoing partnership. After the introduction you know which shape your question falls into.
Is our data used to train models?
No. Your data stays yours and is not used to train external models. We build so that sensitive information stays under your control.
What if the agent makes a mistake?
A language model gives a plausible answer, not a guaranteed correct one. So there is a review step on anything that costs money or goes to a customer, and the agent keeps a log in which you can see what it did.
Read next
- ServicesAI SolutionsSoftware that reads a call, a document or a request and carries out the next step inside your own systems.
- ServicesBusiness AutomationDetails that are retyped between systems today travel on without anyone touching them.
- InsightsWhen AI genuinely adds value (and when it is hype)A practical lens for deciding where AI belongs in your process, and where a plain rule does the job better.
- InsightsRolling out AI with your team: why it stalls, and how it does notAI in a business rarely fails at the building. It fails at the rollout. How to keep your team from working around it.
The next step
Tell us where work gets stuck.
Bring an example from your everyday work. Together we explore what is needed and whether we are the right team to help. Prefer it on paper first? Request the free business scan.