Automate a complex business process with an internal AI app

​

 Context

A wholesaler regularly receives supplier price lists in PDF or Excel. The formats change depending on the suppliers, the formats sometimes change for the same supplier, there are no two equivalent supplier formats and discounts are sometimes by item, type of item, or category of item, and the target ERP imposes a strict import format.

Problem

Updating prices required several hours of manual work: reading the supplier files, Excel reprocessing, applying discounts, calculating the selling prices, checking, and then importing into the ERP.

Created solution

Creation of an internal application that automates the flow: uploading the supplier file, extracting data, human validation, reconciling with the existing database, and then generating the ERP-compatible import file.

What this shows

This case shows that well-calibrated AI can be used on a very concrete process, with precise business rules, without replacing human validation. This allowed the company to reduce the time spent on updates from several hours of work to just a few minutes.

 

Connect small business tools to avoid double entry

​

 Context

A real estate agency uses several tools: A real estate CRM, a tool for leads and expert reports, and an internal database for operational tracking and Make.com for automations.

Problem

Information about properties, leads, statuses, and documents must flow between several systems. Without automation, the team wastes time re-entering the information for each property, does not always complete all the information, and misses opportunities.

Created solution

Implementation of an automation architecture centered on their internal database: Automatic extraction of information from expert reports, feeding the database, generating mandates, and updating the CRM.

What this shows

This case illustrates an approach suitable for a small team: no over-engineering, no AI agent "on autopilot", an automation with AI steps and human control at each stage.

Support a law firm in responsible AI adoption

​

Context

A Luxembourg law firm wishes to advance in AI while respecting its professional obligations: confidentiality, independence, and professional secrecy.

Problem

In the legal field, AI cannot be introduced as a simple productivity tool. Ethical constraints, data security, and human supervision are non-negotiable.

Solution

AI training for management and teams on the risks related to AI and on the basics to understand how AI works. Analysis of existing tools, identification of useful use cases, and definition of safeguards to avoid risky uses.

What this shows

Team training in this case is essential to avoid shadow AI and ensure proper use and understanding of the tools.

Monitoring product and technical alerts

​

  Context

A SaaS app needs quick signals: new users, API errors, chatbot interactions, messages to review…

Problem

Without alerts, incidents or user signals remain hidden in the database or logs.

Solution

A webhook connected to different tables in the SaaS app's database triggers notifications upon a new login, monitors errors, updates the knowledge base, and sends messages directly to support for immediate intervention.

What this shows

Automation can serve as the nervous system for an app: it captures the events, alerts the right people, and keeps a usable record.

 

Monitoring LinkedIn and Facebook

​

Context

A presence on LinkedIn, Facebook, or Instagram requires publishing, avoiding repetitions, tracking statistics, responding quickly to comments, and adhering to editorial guidelines.

Problem

Without a centralized history, there is a risk of republishing the same ideas, losing the statistics, or stepping outside the brand framework. Without a quick reaction, there is a risk of losing the visitor's interest.

Solution

Scenarios retrieve the history of posts, store drafts, review the guidelines, retrieve published posts and collect LinkedIn/Facebook statistics. An automation responds directly to comments to provide the requested resources.

What this shows

Marketing automation is not just about "generating a post." It must also manage memory, consistency, validation, and measurement.

 

 

Automatic treatment of receipts and supplier invoices AI.

​

  

Context

A small business uses several paid online AI tools.

Problem

Receipts and invoices arrive separately, by email, with different formats. Processing them manually takes time and increases the risk of forgetting.

Solution

Several active scenarios process receipts or invoices related to these tools, store the invoices in the company's drive, analyze the content of the invoices, extract the data to feed the accounting system.

What this shows

A good case for automation does not need to be spectacular. Automating small repetitive administrative flows frees up time and ensures the tracking is reliable.

 

Create an AI memory for a solo entrepreneur.

​

Context

A solo entrepreneur must manage many topics in parallel: strategy, clients, projects, ideas, content, tools, decisions, priorities, documents, and operational tracking.

Problem

When information is scattered across notes, emails, files, management tools, and AI conversations, it becomes difficult to find the right context at the right time. The risk is to repeat the same analyses, forget decisions, or waste time reconstructing the history of a project.

Created solution

Creation of a structured Second Brain in Obsidian, organized around projects, clients, apps, priorities, tools, ideas, and key decisions. This system serves as a central memory for the entrepreneur and is used as context by AI assistants to produce more accurate, coherent, and aligned responses with the reality of the business.

What this shows

This case shows that a good use of AI heavily depends on the quality of the context. A well-structured Second Brain allows a solo entrepreneur to better manage their activity, delegate more tasks to AI, have less to rework from AI output, and keep a record of important decisions as well as transform their knowledge into operational assets.

 

 

Transforming an entrepreneur's idea into a public app

Context

Lactose.help was born from the idea of Gauthier de Valensart, a Belgian entrepreneur who is himself affected by lactose intolerance. After years of personal research and structuring a method called Lacto-Score, he wanted to transform this expertise into a product accessible to the general public.

Problem

People with lactose intolerance often lack simple information when choosing a product. They sometimes avoid foods they could tolerate, pay more for “lactose-free” products, or have to analyze difficult-to-interpret labels on their own.

Created solution

The AI App Factory supported the transformation of this idea into a complete digital ecosystem: mobile app, web app, brand identity, user journey, scanning system, scoring logic, integration of food data, AI-assisted analysis, and connection with the book associated with the method written by Gauthier.

What this shows

This case shows how The AI App Factory can start from the idea and expertise of an entrepreneur to create a complete consumer application : product, brand, user experience, technical architecture, and content.