Skip to main content

Services · AI automation for SMEs

Turn one manual process into an operational system.

I map the real work, connect the tools already in place and deliver a measurable workflow your team can operate. Not an isolated demo: a documented, monitored and transferable production system.

Identify the process to automateView the case studies

One point of contact · one metric before building · human approval for sensitive actions

Field evidence

Technology matters. A usable outcome matters more.

Every engagement starts with observable friction and ends with a system the team can monitor, correct and hand over.

2,000+
n8n and Make workflows deployed
21
live automations in the Minia case
35–50 h
saved per month, the executive's estimate for Minia

Three levels of intervention

Choose the level that fits the problem, not the fashionable tool.

A precise process may need targeted automation. Scattered context may justify an AI agent. Several interdependent workflows need architecture first.

01

Process automation

I connect your applications through APIs, n8n or Make, then add the controls and alerts required for the workflow to hold up in production.

Best suited when
Your team re-enters, consolidates, follows up or checks the same information every week.
Target outcome
Remove repetitive steps without replacing tools that already work.

Typical deliverables

  • Target workflow map
  • Automation, testing and production launch
  • Execution logs, alerts and documentation

02

Human-approved AI agents

The agent gathers sources, prepares an answer or action, and asks for approval whenever the company is committed. Every important claim remains tied to its evidence.

Best suited when
Decisions take too long because context is scattered across messages, documents and business tools.
Target outcome
Prepare the right information and next action without blindly delegating the decision.

Typical deliverables

  • Sources, memory and access rules
  • Conversational interface or business integration
  • Human approval and exception handling

03

Systems and product architecture

I design the architecture, data flows, permissions and interface before orchestrating delivery of an application or complete system.

Best suited when
Several processes, datasets and roles need to work together in one coherent interface.
Target outcome
Move from a pile of tools to a maintainable, secure and scalable system.

Typical deliverables

  • Functional and technical architecture
  • Data model, APIs and permissions
  • Delivery plan and production launch

A short, verifiable method

Measure before you automate.

The goal is not to build the most impressive system. It is to solve the right problem with a scope the team can adopt.

  1. 01

    Observe

    Volume, time spent, errors, delays and people involved become the decision baseline.

  2. 02

    Scope

    We choose one outcome, one main path, the essential exceptions and explicit boundaries.

  3. 03

    Ship

    The workflow is connected, tested and launched with logs and alerts.

  4. 04

    Stabilise

    The team uses the system, gaps are corrected and the documentation is handed over.

Is this the right time to automate?

Good automation starts with a process frequent enough to measure and stable enough to scope.

The project is probably ready if

  • A business owner can explain the current process
  • Volume, time or errors can be estimated
  • The relevant tools provide the required access

I recommend waiting if

  • The process still changes every week
  • Nobody can approve rules and exceptions
  • The goal is simply to add AI without a specific business outcome

Start with one process, not a company-wide transformation.

In 30 minutes, we check volume, friction and potential value. If the case is not strong enough, I will say so before any paid scoping.

Assess my processExplore the Flow Pilot