Administrative AI agents
Reading documents, invoices, contracts and emails; managing deadlines, internal reports and repetitive tasks with organised, verifiable outputs.
AI consulting for established businesses
The AI superpower inside your business
Under control, with privacy at the centre and a team trained to become independent.
I am not selling you yet another software product. I help you build an internal capability: trained people, useful AI agents, custom applications, documented processes and a clear strategy for using artificial intelligence in a practical, manageable and sustainable way.
Local AI when needed It can also work without the Internet, with data kept inside the business.
Less external dependency Not every activity needs to involve tokens, subscriptions or SaaS.
Privacy and governance Processes, logs, roles and human oversight designed from the start.
A more independent team Practical training to design agents and internal applications.
AI will not replace people. But people who use AI will outperform those who do not.
The real advantage is not having an account on a platform. It is having employees who can use AI in everyday work: administration, sales, marketing, management accounting, production, document analysis, internal software and decision support.
Before buying another tool
Build internal expertise
Many businesses are experimenting with AI without a clear direction. One department uses ChatGPT, another tries an automation tool, someone creates content and someone uploads sensitive documents without knowing where they end up. The result is often confusion rather than transformation.
The programme I offer brings structure: start with real processes, assess where AI creates value, define the right architecture and train people internally so that they can continue the work after my involvement ends.
More controlData, access, roles and workflows are designed before agents or automations are built.
Less dependencyThe aim is to enable your team to move forward, rather than tying you to one provider.
More speedWith agentic AI, you can create prototypes, websites, apps and internal software much faster.
More securityNot everything needs to leave the business: some solutions can work locally or on premise.
Let’s discuss your AI roadmap
An initial discussion with no obligation to establish whether to start with training, automation, agents or local AI.
What could become possible
From training to business AI agents
After a thorough programme, the business can move beyond using AI to write text and start building its own tools: agents, applications, software, dashboards, document systems and operational solutions working with internal processes.
Custom business management software
Internal applications for specific workflows, project management, archives, procedures, databases and tools that standard software does not cover well.
Websites and landing pages delivered quickly
Create pages, content, SEO structures, prototypes and digital materials independently, reducing reliance on long development lead times.
Marketing, sales and CRM
Prospect analysis, sales content, lead classification, email support, sales scripts, reports and follow-up automation.
Management accounting and decisions
Readable reports, predictive analysis, summaries of business data and support for quicker, more focused operational decisions.
Production and Industrial IoT
Sensors, edge gateways, local models and AI agents to monitor machines, inefficiencies, anomalies and signs of wear before they become problems.
I want to understand what I could build
From websites to internal management systems, document agents to sales support: we start with the most useful processes.
Practical training
The team learns by doing, rather than listening to theory.
Training that produces tools
More than a course. A transfer of capability.
Training should not end with slides forgotten in a folder. It should produce real expertise: understanding how an agent works, how to write an operational prompt, how to connect a model to documents, databases or APIs, how to validate an output and how to avoid exposing sensitive data.
Workshops on real processesWe work on actual business cases rather than abstract examples.
Templates and proceduresPrompts, checklists, workflows, controls and models the team can reuse.
Agents and proofs of conceptThe programme produces concrete prototypes, not just general awareness.
Progressive independenceThe team learns to build, test, correct and improve.
Organise training for your team
Training makes sense when it produces tools, procedures and operational independence.
Local AI, privacy and control
When data is strategic, it should not leave without a reason
There are cases where the cloud is practical. But there are also processes where business data, internal documents, price lists, production data, emails and archives should not be uploaded to different platforms without an assessment.
That is why I always consider local, hybrid and on-premise architectures too: models installed on business machines, internal knowledge bases, accessible logs, access rules, human oversight and traceable decisions.
No tokens for every requestWith local models, where suited to the use case, you do not pay for every individual query.
Offline operation tooSome solutions can work without an Internet connection.
Data under controlInformation remains in business systems or environments designed for its use.
Alignment with the AI Act and privacyAn inventory of AI uses, roles, checks, human oversight and operational documentation.
Let’s assess cloud, local or on-premise options
There is no ideological choice: the decision is based on data, risks, performance, costs and actual usefulness.
Do not buy yet another piece of software. Build your own internal AI system.
Standard software often forces you to adapt the business to the product. A well-designed AI programme works the other way round: it starts with your processes and builds tools around how your business actually works.
The method
From a curious business to an AI-ready
The programme can run from a few months to a year, depending on complexity. The aim is to create value immediately while keeping a long-term perspective.
01
Operational audit
Analysis of processes, data, bottlenecks, repetitive activities and risks relating to privacy, roles and access.
02
AI Roadmap
Setting priorities: what to automate, what should remain human, and what to develop in the cloud, locally, in a hybrid setup or on premise.
03
Proof of concept and training
Building useful prototypes while the team learns the method, tools, prompts, agentic logic and output checks.
04
Internal independence
Progressively transferring skills so that the team can continue, improve and create new agents and applications.
Start with the operational audit
First, measure where AI can make a real difference. Then build.
Production and IIoT
AI is useful beyond the office. It also belongs close to the machines.
In manufacturing businesses, value grows when artificial intelligence meets sensors, production-line data, maintenance, consumption, anomalies and machine history. The aim is to create intelligent layers for interpretation, diagnosis and operational support, rather than replacing PLCs or real-time systems.
TinyML and sensorsLocal detection of anomalous signals: vibrations, temperature, consumption, noise and cycles outside expected patterns.
Edge gatewayCollecting and normalising data on Raspberry Pi devices, industrial mini-PCs or edge servers.
Local agentsOperational diagnostics, log reading, threshold comparisons, reports and suggested interventions.
Predictive maintenanceIdentifying wear trends before they become downtime or inefficiency.
Edge AI and local data
Intelligence closer to real processes.
The practical difference
Scattered tools versus a governed AI system
Isolated experiments
- Each department tries different tools without shared rules.
- Data is copied into external platforms without a proper map of the risks.
- Skills remain concentrated in a few people.
- There is no governance framework, human oversight or validation criteria.
An internal AI system
- Processes analysed and priorities set before development.
- An informed choice between cloud, private cloud, local or on-premise infrastructure.
- Employees trained to build and manage AI agents.
- Documentation, logs, roles, policies and human oversight designed from the start.
FAQ
Before you start
Do we need an IT department already?
No, but you need to be willing to involve people internally. The programme can start with a contact in operations, administration, marketing or management.
Can everything work offline?
Not always. It depends on objectives, models, hardware and the performance required. However, many business use cases can be designed to work locally or in hybrid mode, reducing dependency on cloud services and tokens.
How long does the programme take?
An initial assessment can be quick. A structured programme covering training, roadmaps and skills transfer may take 6 to 12 months.
Is the training theoretical?
No. Theory is included only where needed. The core is practical: real processes, agents, prototypes, applications, use cases and internal procedures.
First step
Understand where AI can create real value
Leave your name and email. The initial discussion helps establish whether it makes sense to work on training, AI agents, internal applications, local or on-premise AI, automation or IIoT.
When does it make sense to talk?
- You have repetitive processes that consume hours every week.
- You want to use AI while taking care over sensitive data.
- You want to train people internally and reduce ongoing dependence on external consultants.
- You want to build custom agents, software, dashboards or IIoT systems.