AI Strategy & Automation

We help you use AI in a secure and auditable way.

We pick one process your team does by hand and build the automation with them. Then we help you get it approved by IT and security, and checked by audit. The automation runs in Alteryx, the data platform your team can open and check step by step.

Check my use case Book free consultation

Premier Services PartnerCertified by Alteryx, one of five worldwide

Where do you start?

AI Strategy Claude/Codex + Alteryx Kickstart AI Use Case Discovery Claude/Codex Training Knowledge Hub Local LLM Value Dashboard

LLMs made everyone a builder. Working automation now appears overnight, written in chat and promoted to production by its own success.

The speed is real. The black box comes with it: nobody reviewed the result, nothing shows how it was made, nobody owns it.

We keep the speed and add the audit trail.

Shadow AI

AI is already in everyone’s company. With or without a plan.

People use ChatGPT, Copilot and Claude at work today everywhere. The question is whether the company is in control.

The risk

Confidential data in public tools
Unverified AI output in decisions
No owner, no rules, no audit trail

Nobody decided this. It happens anyway.

The opportunity

Business value from richer analysis and reporting
Better documentation and knowledge base
People who already want to use AI at work, waiting for a way that is allowed

The upside goes to companies that make AI usable.

Shadow AI is the unsanctioned use of artificial intelligence tools, applications or models by employees without the formal approval, oversight or knowledge of an organisation’s IT and security teams.

PolicyClear rules for real work

ToolsAn approved way to use AI

TrainingRelevant skills and support

See AI Strategy

The real choice is not whether your people use AI. It is whether they use private accounts with no rules, or company tools with rules.

What we do with AI

Our services to enable you in AI

If you have many candidates and no shortlist, start with the Discovery. If you already know the process, start with the Kickstart.

•

AI Strategy

For senior managers and directors who are responsible for AI this year and have no plan for it yet.

The situation. People already use private ChatGPT or Claude accounts for real work, and the company has no central approach to AI.

What we do. We write the strategy with you and then execute it: which data may go to which tool, which tool each group gets, how to get people using it.

What you get. One approved way to use AI, ready for your security review. For example Microsoft Copilot for everyone and a governed Claude/Codex environment for the specialists.

•

AI Use Case Discovery

Half a day to pick the first process to automate.

What we do. We go through the work your teams do by hand, about 30 minutes per process, and score each one on value, effort, speed and reuse.

What you get. A ranked list with hours and euros per year, and one process chosen to build first.

•

Claude/Codex + Alteryx Kickstart

Build your first automation on your own process in two days.

What we do. We build one real automation with the people who run the process. Claude Code or Codex writes the Alteryx workflow. Your team reviews every step.

What you get. The workflow, its documentation and a recording of the build, so your team can build the next one.

The rules. A person starts every run. The AI works in one folder under written rules. Your team needs its own Claude Code or Codex accounts.

•

Claude/Codex Training

One or two days of hands-on training in using AI at work.

What we do. Your people learn Claude Code and Codex on their own tasks, not on slides.

What you get. A team that knows what the tools can do, what they cannot, and the rules for using them safely.

•

Knowledge Hub for Small Teams

One place where your team and your AI tools read the same rules.

What we do. We build a knowledge base the AI reads before it answers: your processes, standards, templates and security rules. Corrections are written in and carry over.

What you get. The same answer for everyone in the team. The first version is written together with them. Works best for a tax or finance team whose rules change every year.

•

Setting Up Your Local LLM

An AI model on your own hardware, working on data that never leaves the building.

For whom. Banking secrecy, a works council, a plant with no outside connection.

What we do. We choose the open model (Llama, Mistral or Qwen), size the hardware, set it up and leave one process running on it. Your team is trained to keep it running.

What you get. One fixed fee for the whole package. Before we start, a cost efficiency analysis: which model, and how to run it.

•

AI-Powered Value Dashboard

See how much time your Alteryx automations save today.

What we do. We scan your Alteryx environment: which workflows run, how often, who owns them, and the staff time each one saves. Every workflow gets a score from 0 to 100 and a short description written by AI.

What you get. Thousands of workflows ranked without opening them, and a clear view of which ones are worth keeping when Alteryx One counts runs. Already done on more than 10,000 workflows for an enterprise client.

Our standard

Every automation has to pass four checks. We call it VURA.

VURA stands for Visible, Understandable, Repeatable and Auditable. Every automation has to pass all four before we hand it over, whether a person or an AI built it.

Visible

Every step is drawn on screen, in order. You can see what the automation does by looking at it.

Understandable

Each step has a plain name, like “Find the invoice number in the payment reference”. You do not have to read code to see what it does. Each exception says why: unpaid, partially paid, no exchange rate.

Repeatable

It runs again next month, on new data, on your own Alteryx schedule, without the person who built it being there.

Auditable

When someone asks how a number was produced, you open the automation and show them.

Why you need our help at all

Creating a solution with AI is easy. Making it production-ready, secure and auditable, at scale and every month, is the hard part. It has to run again in a month, on fresh data, for someone who did not see it built. An Alteryx workflow shows every processing step, so the record comes with the thing that runs.

How we work

We start from one of your processes, with your team in the room. Your team learns it while we build it. Everything is named to your standards, scheduled where it belongs and documented so your team can build the next one. It is the same way we deliver process automation and tax automation. If your data has to stay in a certain country, or security has to sign off, we plan for that from day one.

In 5 minutes we build an automation that runs every day.

It compares your invoices with the ERP, flags every difference with its reason and builds the dashboard. Claude Code and Codex write the whole thing. Nobody touches the canvas. The same workflow handles a hundred invoices or hundreds of thousands.

5 min

to build the whole automation, live, from an empty canvas

Daily

runs on your Alteryx schedule, without the person who built it

100,000+

invoices in one run: ERP against invoices, every difference with a reason

Built by AI, reviewed by your team · numbers are calculated by Alteryx, not by the AI

Our webinar

Your First Claude Automation with Alteryx, Without the Black Box

Live demo webinar with Alteryx, 23 September 2026, 10:00 CEST. Watch an AI tool build the reconciliation above in Alteryx, from an empty screen. You see every step as it happens.

Presented by Bart Smuga, Co‑Founder at Data Pal and Alteryx ACE.

Watch the webinar

Questions we get

Four questions about AI and your data

Can we use Claude or OpenAI models without sending our data to the United States?

Yes, with the right setup, and we configure it together with your IT and security team.

Claude runs through Amazon Bedrock on an AWS account in Frankfurt, configured to process requests in EU regions. OpenAI’s GPT models run through Azure OpenAI in an EU region. Your data is not used to train any model. AWS and Microsoft provide data processing agreements your security officer can review.

If no cloud is allowed at all, a model on your own hardware is the alternative, within limits we name up front.

Does the AI see our financial data?

Only what you decide. For most Alteryx work the AI needs to know what the columns are called and to see a few example rows. It does not need the full data.

We agree that data flow before the build, and where files are confidential, we pseudonymise them before they reach a model.

Which AI models and tools do you work with?

Claude Code and Codex, with the add-on Alteryx publishes for Claude Code. Claude on Amazon Bedrock and GPT models on Azure when data has to stay in the EU.

The AI features inside Alteryx. Open models such as Llama, Mistral and Qwen on local hardware.

And whatever your company has already approved: Microsoft Copilot, ChatGPT Enterprise or Google Gemini.

Can you build the automation without AI?

Yes. Every automation on this page can be built without an AI coding tool. Banks and works councils often ask for that. The only difference is that building takes us longer.