Automation & AI

Fewer repetitive tasks. More time for your business.

I automate repetitive tasks for teams that copy data, repeat notifications or lose information between the website, email and other tools. First we document the current process and agree a shorter flow, with people responsible and a manual way out when something fails; AI only comes in where it adds value, and with human review. You work directly with me, with the scope in writing before we start.

See what is included
  • A small, testable flow
  • AI with human review
  • Clear costs
  1. Process
  2. Flow
  3. Control

In short

What it is
Automation, integrations and AI for repetitive tasks.
Who it is for
Teams that repeat tasks or lose information.
How I work
Remotely, with one person responsible.
Price
Fixed quote in writing; tool costs clarified beforehand.

Technologies and platforms depending on the project

  • OpenAI / ChatGPT
  • Claude
  • Gemini
  • WordPress
What is included

What does automating your tasks include?

  • Process map

    We draw the process as it is today: what triggers it, who is involved and what result you expect.

    Trigger, input data, decisions, people responsible, exceptions and expected result. The current process is documented before it is automated, in order to agree a shorter flow with clear responsibilities and a manual way out when something fails, and to decide what is automated and what is left in a person’s hands.

  • Automated flow

    I describe step by step the form, notification, document or status change that is automated.

    Specific forms, notifications, documents or status changes described step by step. A small, testable flow is proposed first, and the scope lists tools, inputs, outputs, permissions and tests, so it is known what the flow does at each step before it goes live.

  • Agreed integrations

    I connect the tools you already use that can be connected, such as your website, email or a CRM.

    Available connections between WordPress, email, CRM or other accessible tools. We agree which tools are involved and which data passes from one to another, and the scope sets the planned response when an external integration is unavailable, so the flow does not depend on everything always working.

  • AI with human review

    I use AI only where it adds value, and a person approves the step that matters.

    Sources, instructions, limits and validation point defined when AI adds real value. We decide which information the AI works with, what it is asked to do and what it cannot do, and a person reviews the result at the agreed point before it moves on. Depending on the project, models such as ChatGPT, Claude or Gemini are used.

  • Errors and recovery

    If something fails, it is logged, you get an alert and there is a manual fallback.

    Logging, alerts, edge cases and a manual fallback prepared so that failures are not hidden. Cases outside the norm are planned for, as is what happens when an external tool does not respond, so the failure is logged and flagged and the work can continue by hand while it is resolved.

  • Permissions and usage costs

    I ask only for the access that is needed and make clear what each tool costs to use before anything goes live.

    Minimum access, run limits and tool costs clarified before the flow goes live, with people responsible and controls defined. We agree which permissions each tool grants, which limits the flow has and who takes care of each control, and the flow is not started until everything is clear.

A small, testable flow is proposed first. The scope lists tools, inputs, outputs, permissions, tests and the planned response when an external integration is unavailable.

Proof

A flow you can check step by step.

This is how a flow is set out: small, testable and with a manual way out when something fails.

Permissions and costs

  • WebsiteOnly read the forms received.
  • EmailOnly send the flow’s notifications.
  • AICost per use.To clarify before going live
  • External toolsEach tool’s plan price.To clarify before going live

Illustration with example data; it is not a client’s flow.

Published work

Real websites, live.

See the whole portfolio
How I work

Five steps from the repetitive task to a flow that works.

  1. Understand the starting point

    You tell me what you copy, which notifications you repeat and where you lose information between the website, email and other tools.

  2. Define scope and proposal

    We document the current process and agree a small flow: tools, inputs, outputs, permissions and costs, with a fixed quote.

  3. Map the process

    We describe the trigger, input data, decisions, people responsible, exceptions and expected result.

  4. Build and test

    I build the flow step by step, with logging, alerts and a manual fallback, and test it before it goes live.

  5. Go live with control

    I switch the flow on with minimum access and run limits, and make clear who is responsible and when human review comes in.

FAQ

Questions about automation and AI.

Price, timing, AI, failures and usage costs: what people usually ask me before automating a task.

How much does it cost to automate a process?

It depends on the scope: how many tools are involved, how many steps the flow has and whether it includes AI. Before we start, you receive a fixed quote in writing, and any change of scope is also agreed in writing. The scope lists the tools, inputs, outputs, permissions and tests included, and the usage costs of the tools are clarified before the flow goes live.

How long does it take?

It depends on the scope and on the process, the access and the tools being ready. A small, testable flow is proposed first, and the dates and milestones are set out in writing in the proposal before we start.

How is it paid?

The payment terms are set in the proposal, together with the scope and the fixed quote in writing. The usage costs of external tools are clarified before the flow goes live, with people responsible and controls defined. Any change of scope is agreed in writing.

Who owns the flows and accounts that are created?

Access is limited to the minimum needed and the people responsible for each tool are defined before the flow goes live. Exactly what is delivered, including the flows and their documentation, is set out in writing in the proposal.

What do you need to get started?

How the process works today: what triggers it, which data comes in, which decisions are made, who is responsible and which exceptions come up. Also which tools you use, such as the website, email or a CRM, and what result you expect. If you are not sure, we document the current process together before closing the scope.

Which tasks can be automated?

Specific forms, notifications, documents or status changes that you do by hand today or copy between accessible tools, which are described step by step before they are built. If the process needs an interface or logic of its own, a web app may be a better fit than an automation.

Do you use AI for everything?

No. AI only comes in when it adds real value. In that case the sources, the instructions, the limits and the human validation point are defined. Which model is used, such as ChatGPT, Claude or Gemini, depends on the project and is agreed in the scope.

Can the AI get it wrong?

Yes, which is why I do not leave it on its own in the steps that matter: a human validation point is defined, with explicit sources, instructions and limits. I do not promise the AI will always be right; what I prepare is a flow in which failures are logged and flagged instead of hidden.

What happens if an external tool stops working?

The scope includes the planned response when an external integration is unavailable: a log of the failure, alerts, edge cases and a manual fallback prepared so that the work does not stop and the error is not hidden. That manual way out and who triggers it are agreed before the flow is started.

What permissions and costs does it involve?

Minimum access is requested, with defined run limits. The costs of the tools and of any AI used are clarified before the flow goes live, with people responsible and controls defined. The flow does not go live until that cost is clear to whoever is responsible for it.

Which tools can it connect to?

It can connect to WordPress, email, a CRM or other tools that offer a connection, as agreed in the scope. The scope lists tools, inputs, outputs and permissions. If an external integration is unavailable, the flow has a planned response, with a manual fallback, so the process does not depend on everything always working.

What happens after it goes live?

If something fails, it is logged and flagged, with a manual fallback ready. Improvements, extra steps or later support can be agreed separately, depending on what the flow needs, and any change of scope is agreed in writing.

Tell me which task eats up your time.

Tell me on WhatsApp what you repeat by hand and I will propose a first small, testable flow.