Transactional approach
What does the user
need to do?
Create a contact. Update an opportunity. Log an activity.
mindBrain Bootcamp · 4 weeks
Four weeks to design with mindBrain, test your rules, and plan the next phase of your project.

Start with your business expertise.
No programming experience required.
A first proof of concept with sample data.
Chat tools powered by large language models (LLMs) have made natural language a way to interact with software: type or speak a question to get an answer instead of searching for a screen.
Connected to company data, this interface lets you ask questions such as, “Why is this deal stuck? Who needs to step in?” The answer may draw on contacts, opportunities, activities, and business rules.
How you frame questions becomes a strategic design choice: it defines what each role needs to understand, decide, and do. Transactional interfaces, built for entering and editing data, cannot cover all these conversational uses.
The design approach needs to evolve. You need to design the knowledge, relationships, and conditions that make answers possible, alongside operations and views.
Transactional approach
Create a contact. Update an opportunity. Log an activity.
Answer-oriented design
Which deal is stuck? Why? Who needs to step in? What consequences should we expect?
Both approaches work together. Operations supply the facts; the business model organizes the knowledge needed to answer questions.
One case, several perspectives
Prospecting, sales, finance, leadership: each role has a specific job to move forward. Choose a role.
The example on this page is a CRM model that makes a use case easier to understand. The bootcamp is open to all kinds of agentic applications.
The SDR’s perspective
When I qualify an opportunity, I want to identify what is missing so I can plan the next step.
“What information or approvals are missing before we can hand off this deal?”
The answer in this example
No follow-up is scheduled and the budget is not confirmed. Camille needs to prepare a conversation and check the budget before the handoff.
Facts used: owner, next activity, budget confirmation.
Fictional data, scripted answers. Explore the reasoning; no agent is connected to this page.
Explore needs at a larger scaleYou have just seen five perspectives on one opportunity. How many answer needs does that represent across your business?
The same business activity involves different responsibilities. Adjust the two sliders.
For example, 5 roles with 8 questions each represent 40 answer needs.
5 roles × 8 questions
40answer needs
to organize
Illustrative calculation: some questions are shared across roles. This total does not count screens or estimate development costs.
Organize facts, relationships, and rules in a shared business model. People and agents can then use answers whose conditions are explicit.
A reference question can support different phrasings and parameters. “Where should I step in?” and “Which deals need my help?” draw on the same knowledge. A new question may also require another source or rule: that gap needs to stay visible.
Schedule a follow-up for Atelier Nord. Observe what changes and what still needs to be checked.
Demo rule: a scheduled follow-up and a confirmed budget are required before proposing a handoff to the manager.
Scheduling a conversation resolves the missing follow-up. It does not confirm the budget.
The selected role’s answer in the roles module uses these same facts. An unknown budget does not mean a rejected budget.
An agent can prepare an action using the available answers. Conditions and approvals remain explicit.
Based on your situation
The follow-up still needs to be scheduled. Preparing a conversation does not yet allow the opportunity to be handed off.
The proposal has not been prepared yet.
Local simulation: no activity is created and no message is sent. Changing the facts cancels the prepared proposal.
Defining rules lets you design automated processes. When a fact changes, the answer to a business question changes too: the agent can then determine what it may or must do.
At a larger scale, answers and actions can form a sequence that moves an entire process forward. You define the conditions for execution and the points where a person steps in.
From detecting risk to informing the right people. The agent carries out actions that are already authorized.
Step 1 of 4 · Agent
The agent checks the business answer. Atelier Nord appears among the opportunities at risk: its deadline has passed.
The business answer provides the starting point for the process.
Click to explore the steps. Educational simulation: no message or quote is sent.
A fully agentic process
The agent follows business answers and carries out actions authorized by the rules.
The rules specify triggers, recipients, and the conditions that authorize these messages.
A combined agent and human process
The agent prepares the work. A person approves the proposal that authorizes the next step.
Human approval authorizes the next steps. The prospect’s signature remains a separate step.
In both cases, you design the connection between facts, answers, rules, and actions. These examples describe processes to build and test.
Test the important situationsIf a follow-up is scheduled but the budget is missing, the handoff must remain on hold. Test three situations from the same example.
These test cases are independent of your earlier choices.
The case to check
Keep this distinction in your own rule.
It compares a local demo rule with the expected result for a specific case. In the bootcamp, your model’s scenarios help you observe results, identify gaps, and gather feedback from people who know the business.
The role of mindBrain
mindBrain connects business information, concepts, and rules to answer questions. It can complement existing software or support the model of an application you build.
The CRM example makes the explanation concrete. mindBrain adapts to any business model, whatever kind of agentic application you build.
In production, mindBrain Pro provides connectors to manage data synchronization. The scope depends on the software and available access.
For mindBrain Personal, connectors need to be adapted. Existing MCP servers can speed up this work by providing access to tools and data. Field mappings, data flow directions, and update frequency still need to be defined. MCP access alone does not provide continuous synchronization.
Another way to use the application
An agent such as Hermes or OpenClaw can connect your application to Telegram, WhatsApp, Discord, or Slack, depending on the integrations you set up.
You no longer need to design a fixed screen for every use case. The agent can handle transactions, answer questions, and trigger workflows. When a view or form helps move the work forward, it can also generate an interface on demand and send its URL in the chat.
An example use case to build
“Which opportunities are at risk, and what should I do?”
The bootcamp starts with a model and scenarios using synthetic data. Live connections and production deployment come later, based on the scope you have tested.
Prepare my projectYour turn
Start with a simple scenario to understand the question-centered approach. Choose a starting point, then adapt it to your project.
Changing the template replaces its suggestions. You will be asked to confirm before replacing your edits.
You have customized this brief. Loading the selected template will replace your text.
Your first model outline
This brief is a starting point. Scope and feasibility still need to be reviewed.
Your text stays on this page. It is not sent with your email address.
Your first project
During the bootcamp, you will define this scope, build the model with the Architect, and check how it works with the Operator and people.
Four weeks of practice with mindBrain on a focused process. You bring the business knowledge; the program gives you the method.
To build
You define the questions, facts, and rules. The Architect handles the technical build. You review it and request changes.
To use
You use your model through messaging, ask business questions, and observe the possible next steps in the scenarios you prepared.
Set the scope of a process, its roles, priority questions, and expected outcome.
Organize information, relationships, and rules with the Architect.
Prepare synthetic data and try the expected cases, unknowns, and exceptions.
Review results, gather human feedback, and decide on corrections and next steps.
Business consistency: the people involved recognize the concepts and rules. Functionality: the model behaves as expected in the tested scenarios. Usage: the answers help people move forward.
A proof of concept with sample data provides observations within a stated scope. A pilot with users and real sources can then assess adoption, integrations, and operating costs.
Weekly sessions over 4 weeks. No programming experience required, with definition work and practice between sessions.
Live connections to third-party software come later. The schedule, support, mindBrain access, and price will be specified before you enroll in the bootcamp.
Get the program and datesScope, commitment, and the next step.
No. You describe the expected behavior in natural language. The Architect handles the technical build. You still need to define your needs, review proposals, and practice with the Operator.
The goal is a first working business model in mindBrain, within a focused scope, tested with synthetic data. Your scenarios and observations will help you decide what comes next. A larger project will require additional work.
No. They are local illustrations using a fictional case. They make the principles interactive. The bootcamp teaches you to build and use your own model with the Architect and Operator.
The announced program uses synthetic data to test situations and rules. Live connections to third-party software and a pilot with real sources come later.
The program will specify pricing, the schedule, support, the duration of mindBrain access, and any exercise-related costs. It will also specify the supported messaging channel among Telegram, WhatsApp, or Discord.
Yes. Start with a team process or a service you want to offer clients. The first goal is a demonstrable, tested scope. A pilot, client deployment, and commercial preparation are next steps to consider.
No. The brief fields stay in your browser and can be downloaded as text. The form sends only your email address to the existing mailing list service. If you do not download it, your brief is lost when you reload or close the page.
No. You are requesting the program, dates, pricing, and details so you can decide. This signup does not reserve a paid seat or commit you to a purchase.
Get the program, upcoming dates, pricing, and details to choose your session.
Weekly sessions over 4 weeks to build and test a useful scope.