Processing orders, preparing quotes or checking product data: much of the work follows the same route until an exception occurs. An AI agent takes over the recurring work and involves your people when their knowledge or judgment is needed.
A Pilot is an AI agent that solves a defined problem, connects to your daily operation and keeps running as long as it delivers value. Integrations, human oversight and management are included.
First determine where to start? Explore the workflows that are suitable, the metric to measure in advance and what your existing systems continue to do in the Knowledge hub.
OrderPilot, QuotePilot and ProductDataPilot already exist. The foundation is there, but every wholesaler has different systems, agreements and exceptions. We do not assume another customer’s results will transfer one-to-one.
Reads and checks orders from email, PDF, portals or Excel and prepares them for the ERP. People decide on exceptions.
Uses order files, customer agreements and ERP data. The demonstrated result at INSPIRED: 9x faster processing and 60–80% lower processing costs for this specific workflow.
Explore OrderPilotReads supplier and competitor price lists and matches products to the company’s own range. Powered by MatchPoint.
Uses PDF, Excel, catalogue data and ERP integrations. Results are qualitative for now: less manual work and more consistency.
Explore QuotePilotMakes supplier files and product data usable and flags missing fields. Powered by Origo.
Uses supplier files and product sources. We will publish the customer name only after approval.
Explore ProductDataPilotWe show results only after they have been proven live. OrderPilot figures come from INSPIRED Pet Nutrition.
In these workflows, we see recurring checks, scattered information and exceptions that require expertise. Our team turns the work that keeps piling up into a clear workflow and builds an AI agent around it that fits your people, data and systems.
For recurring customer questions about orders, deliveries and products, with your team handling the exceptions.
For signals around stock, demand and availability that currently remain scattered across systems and people.
For invoices, credit notes and comparison with orders or receipts before a person decides.
Then we build a custom AI agent around your workflow, exceptions and existing systems. During the Prove session, we clarify what can responsibly be automated and what should stay with people.
We continue monitoring how it performs, where exceptions arise and what changes in the process. This keeps it working and allows focused improvement. It can then take on more work, or another agent can address a different problem. That is always a separate choice.
Dennis connects the operational challenge, the business case and the people who need to work with the AI agent. He is your first point of contact for AI agent questions.
No. We call our AI agents Pilots, but a Pilot is a working AI agent that solves a defined problem in your daily operation. The Prove session determines what it should do in advance. We then build it and put it to work.
rb2 works in accordance with the GDPR and a data processing agreement is standard. Your data remains yours. For every agent, we determine in advance what data is needed, where it is processed, who has access and what oversight is required. The exact setup depends on your systems, data and security requirements. rb2 is working towards ISO 27001 certification, which is expected in October 2026.
That is the starting point. During the Prove session, we determine which systems, data sources and integrations are needed without replacing your IT landscape unnecessarily.
Security, monitoring, management and human oversight are part of delivering every agent and keeping it running. We track whether it does its job, where errors or exceptions arise and when human attention is needed. We agree the division of responsibilities between rb2 and your team in advance.
Yes. We determine in advance what work the agent handles independently and when an employee reviews or decides. Routine work goes to the agent; judgment stays with your people.
It continues doing the agreed work. Based on practice, we improve it, give it more work or build another agent. Every extension is a separate choice.
People who understand the process and its exceptions, access to relevant information and someone who can decide on scope and working methods. During the Prove session, we make the required time commitment concrete.
The Prove session has a fixed fee that you know in advance. The outcome is a substantiated decision: build, first improve data, a process or an integration, or do not proceed. A separate proposal with a fixed build fee follows only after a positive build decision. After that, a fixed monthly fee per agent covers management and improvement. There is no obligation to continue.
When the work has too little volume or repetition, the data is missing or no one owns the process. The outcome of the Prove session may therefore be to improve data, a process or an integration first, or to do nothing. We will say so plainly.
In the free introductory conversation, we check whether there is a recurring workflow with potential for an AI agent. In the paid Prove session, we quantify the current situation, the desired result and the gap between them. If the work is not suitable, we will say so.
Plan a 30-minute conversation