AI in your company: steered, not just tried out.

I translate between leadership and engineering: where AI pays off, what it costs and how it gets into production safely. What I recommend, I build myself, with AI agents.

Start a conversation

20 minutes, free, I reply personally

TriedChecksProduction
From my own product: of 1,215 changes, 1,131 went into production. I stopped 83.

  • Since 2016

    taking AI products to market, in a global corporation and in my own company

  • Live

    My own software with paying customers

  • 80+

    startups and teams supported

Situations

Where AI gets stuck in companies.

It’s rarely the technology that’s missing. Usually it’s someone who translates between leadership, business teams and IT, so that a decision can be made.

Typical situations, paraphrased from workshops and conversations.

  1. Our competitors promise the moon with AI. How do we stand out?

    With honesty in the detail: saying clearly what your product does and what it doesn’t. That’s how I took the first AI product of a global corporation to market.

    Interim AI lead

  2. We bought the licences. We don’t have a plan for them.

    First we settle what you use AI for, where your data lives and who you want to depend on. After that the tool question answers itself quickly.

    Briefings, trainings, talks

  3. We need a chatbot that can do everything.

    Usually you need good software: adapting existing systems, adding modules and interfaces. AI goes only where it pays off.

    AI architecture review

  4. We have plenty of ideas. Nothing makes it onto the road.

    I break the project into milestones. At each one you see whether the direction is right and can correct course.

    Prototype sprint

  5. Our developers work with AI, but we don’t trust the result.

    Your team learns how I work: clear specifications, tests up front and a check that stops before anything goes live.

    Team intensive: building with agents

  6. Someone does AI on the side. Who decides, and who owns it when something goes wrong, is unclear.

    I take on ownership of AI for a while and build up your team until it can carry it.

    Interim AI lead

Recognise yourselves? Three sentences by email are enough.

Principle

Every second case is solved by good software.

The brief often says: we need a chatbot. Usually it’s enough to adapt existing systems, add modules and interfaces and work across departments in a new way. Ongoing AI only where it pays off.

CostTimeOngoing AI usageGood software, built with AIVendor raises the pricebuild oncecheaper from here
Cost over time, schematic

Ongoing AI usage

  • Cost on every use
  • The vendor sets the price
  • If it goes down, operations stop

Good software

  • Built once, it belongs to you
  • Keeps running on your systems
  • AI only where it pays off

This is where the biggest potential lies: staying independent and building your own strengths. And the biggest risk: becoming dependent. Preventing that is the core of my work.

Process

From the first conversation to production.

Six steps. After each one you hold something in your hands and decide whether to go on.

Example: a wholesaler with 80 staff, quotes from customer enquiries

  1. 01Conversation

    20 minutes, free

    You describe the situation in three sentences. Afterwards you know whether and how I can help. If I can’t, I’ll say so plainly.

  2. 02Review

    1–⁠2 weeks, from €1,500

    I check value, data, risks and dependencies: what you can do on your own, what connects to existing systems and where AI is really needed.

    In my own workI answer the same questions for every product of my own before anything gets built.

  3. 03Decision

    you decide

    Build, buy or stop: you decide on a clear basis. Stopping is a good outcome too.

    In my own workA deliberate no saves the most. For my own products it is written down what they won’t become.

  4. 04Build

    2–⁠4 weeks, from €7,500

    AI agents write the code, I steer what gets built and in which order. Every week there is a version to try. At every milestone you can correct course and sharpen the goal.

    In my own workIn my field sales product, agents write 95% of the code. More than 16,000 automated checks run on every change.

  5. 05Safeguard

    before every release

    Security, data, permissions and cost are checked automatically. If something is missing, everything stops until it’s fixed.

    In my own workOn one day the checks stopped me twice. The new feature still went live that evening.

  6. 06Production and handover

    3–⁠6 months as interim AI lead

    I stay with it in production until your team can run it: ownership, documentation, next steps.

    In my own workThat’s how my own product runs, with paying customers.

Offers

Five formats. Clear prices.

From a fixed-price review to interim AI ownership. Every format ends with a result you own, and with an architecture that keeps running when a vendor goes down.

AI architecture review Start here

A written report on use case, model choice and dependencies with a clear recommendation: build, buy or stop. Plus a 90-minute walkthrough.

Duration
1–⁠2 weeks
Price
from €1,500fixed price
Ask about it

Prototype sprint

A working prototype plus an architecture document, so it’s clear what production still needs and what it costs.

Duration
2–⁠4 weeks
Price
from €7,500fixed price
Ask about it

Interim AI lead

AI and product ownership for a limited time: roadmap, architecture, vendor choice and acceptance. I translate between leadership and team until your team owns it.

Duration
1–⁠2 days a week, 3–⁠6 months
Price
from €4,500per month
Ask about it

Briefings, trainings, talks

For leadership, teams, boards, associations and conferences: what AI and agentic building can do, what they can’t and what it costs. Available as AI literacy training under Article 4 of the AI Act.

Duration
45–⁠120 minutes
Price
on request
Ask about it
  • Every format starts with a free 20-minute call. If none fits, I’ll say so plainly.
  • Startups: prototype sprint once funding is in place. For early stages, the AI Factory Austria programmes are the better start. My workshops there are separate from these offers.
  • All prices exclude VAT where applicable.
  • Process, prices and answers on one page: AI consulting in Vienna.

Proof

First AI product in 2016. First company in 2013.

Before my own products: a global corporation, advisory, more than 80 startups, over 150 juries and trainings. That’s why I can do both: build and explain.

Stations

  1. since 2026

    AI Factory Austria AI:AT

    Head of AI Acceleration & Growth

    The Start AI, Build AI and Scale AI workshops and the Executive Track for SMEs, startups and leadership teams. Mentor at the Public AI Hackathon 2.0 with Austria’s Federal Chancellery.

  2. ongoing

    FieldProFlow

    Founder

    Product, architecture, operations and sales of field sales software with paying customers, built with AI agents.

  3. 2026

    UNDP AltFinLab

    Technical Lead Emerging Tech (AI and blockchain)

    Helped build the SDG accelerator, including curriculum and team selection. Workshops on go-to-market and investor readiness for 20+ AI and blockchain teams.

  4. 2024–⁠2025

    Cointelegraph

    Head of Program Accelerator, then Web3 Business Director

    More than 40 startups in the accelerator, jury and pitch events with investors, workshops in Dubai, speaker at Paris Blockchain Week 2025. In 2025, 10+ partnerships with more than USD 4M in revenue.

  5. since 2022

    Valuex AG

    Co-founder, CEO for the first two years

    Strategy and fundraising advisory for tech startups: USD 1M+ revenue in year one, USD 10M+ valuation in year two. The startups we supported raised USD 25M+.

  6. 2019–⁠2020

    Deloitte Digital Austria

    Senior Business Development Manager

    Helped build the Salesforce and digital consulting business in Austria, digital transformation and CRM strategy for enterprise clients.

  7. 2016–⁠2019

    Siemens Healthineers

    Global communications, Senior Manager Strategic Marketing

    The company’s first AI product, the AI-Rad Companion, from launch: EUR 2M+ budget per year, international teams, launch at RSNA in Chicago, the AI story for the 2018 IPO.

  8. 2013–⁠2015

    rethynk

    Founder and CEO

    My first software company: SaaS for digital publishing, launched in Europe, the US and Australia.

The full profile: AI leadership
Paul Kruegel speaking on stage in front of an audience
Regularly on stage: talks, panels, juries and workshops at a wide range of events.

Open to keynotes, panels and juries

More than 150 juries, trainings and workshops, among others moderating at Proof of Talk in the Palais du Louvre, juries in Paris and Copenhagen, speaker at Paris Blockchain Week, workshops in Dubai and Zurich. Plus pitch and jury events in many other cities. Topics: steering AI in companies, staying independent of AI vendors, building with agents. In German and English.

Ask about a talk, panel or juryTopics and formats

Worked with and for

  • AI Factory Austria AI:AT
  • Siemens Healthineers
  • UNDP
  • Cointelegraph
  • Deloitte Digital
  • Valuex
  • adidas
  • Nestlé Schöller
  • AIDA Cruises
  • Sky
  • Carl Gross
  • Brand Club Austria
  • Guiding Innovators Hub Vienna
  • NUREG
  • rethynk
  • Playboy Deutschland

What others say

He has a rare ability to look at a project and quickly get to the core business risks and opportunities.

Chris ScibiorekCointelegraph Accelerator · LinkedIn recommendation, Nov 2025

His mastery of strategic conceptualization and actionable execution is second to none, parsing the signal from the noise.

Anthony FrancisCointelegraph · LinkedIn recommendation, Nov 2025

He gets the space, asks the right questions, and brings clarity when it's needed most.

Pawel LaskarzewskiNomad Fulcrum · LinkedIn recommendation, Jul 2025

Paul brings a rare blend of strategic clarity and hands-on leadership in marketing, positioning, and reputation-building.

Anna ShakolaADI Foundation · LinkedIn recommendation, Jun 2025

His remarkable ability to tackle complex challenges with innovative solutions makes him an invaluable asset to any team.

Oliver SchmittValuex, co-founder · LinkedIn recommendation, Jun 2025

Thanks to him the global launch of the AI-Rad Companion was an outstanding success.

Ivo DriesserSiemens Healthineers · LinkedIn recommendation, Dec 2018
All recommendations on LinkedIn

Approach

Agents write 95% of the code. The rest is the actual job.

With agents, almost anyone can build a demo in an afternoon. Software that customers pay for needs the part underneath.

Six rules I steer AI by

  1. The model delivers, the code does the maths.

    In my idea evaluation, several models judge separately and the code computes the result. No model can override it.

  2. AI only where it pays off. And it can always be switched off.

    In my peer-help app the AI assistant is off by default and runs on an EU model. My budgeting app has no AI advice at all.

  3. Answers only from approved knowledge.

    My support product answers only from reviewed knowledge. When it’s about money, accounts or privacy, a human takes over immediately.

  4. Nothing reaches production without a check.

    Of 1,215 changes to my field sales product, I stopped 83.

  5. A deliberate no is a decision too.

    My peer-help app keeps a written list of what it won’t become: no chat, no ticketing, no leaderboard.

  6. Data stays in the EU. And only as much as needed.

    My apps store data in Frankfurt and work without tracking. The budgeting app doesn’t even need an account.

The numbers behind it: building software with AI agents
DemoWaterline
  1. Permissions
  2. Security
  3. Privacy
  4. Billing
  5. Integrations
  6. Resilience
  7. Operations
Demo: what fits in a video.Underneath: what customers pay for.

Products

Built myself: five products, one person.

What I recommend, I do myself: AI agents write the code, I decide what gets built and when it goes live. The status next to each one is honest.

Live

Field sales software for independent reps and brands in FMCG retail.

Plan visits, capture them right in the store and reuse the same data for orders, reports and invoicing. On iPhone and Android, with or without a connection.

100,000+
verified retail locations
Multiple principals
data kept separate per principal
Offline
works without signal in the store
E-invoicing
ZUGFeRD and XRechnung
fieldproflow.com
FieldProFlow in the browser: weekly planner of a field sales rep with demo data
FieldProFlow on a phone: the day plan with route and next store visit, demo data
Real product screens with demo data. The product is in German.
  • Pilot

    For home care services

    Care staff capture care documentation by voice or text, review it and export it ready to file. Less desk time, more time for people.

  • In build

    For companies with 20 to 500 people

    A digital notice board where colleagues help each other. In the company’s own look.

    knowhand.com
  • Alpha

    For households in Germany and Austria

    Shows what really remains to live on after fixed and yearly costs. No bank access needed.

    restklar.com
  • Pre-launch

    For teams with web products

    Answers customer questions from reviewed knowledge and hands over to a person as soon as money, accounts or privacy are involved.

    no-bs.io

Plus internal tools for idea evaluation, contacts and sales that you can’t buy like this. With them, one person runs all of it.

Build log

What I’m building and learning now.

On LinkedIn and X I write about what AI actually changes in companies, and what it doesn’t.

  1. NoteMind blown. I'm taking back part of my Saturday skepticism.LinkedIn
  2. NoteIf token costs live on the cost line, the AI budget fight is half lost already.LinkedIn
  3. NoteSprach-KI zieht auf den Laptop, und die Cloud-Kette wird kürzer.LinkedIn
  4. NoteEin Anbieter mit allen Modellen ist trotzdem ein Anbieter.LinkedIn
  5. NoteEu ai policy loves announcing a supercomputer.X
  6. NoteEine Kontrolle, die der Agent selbst verwalten darf, ist keine Kontrolle mehr.LinkedIn

Questions

Quick answers.

Who is Paul Kruegel?

Paul Kruegel (Paul Krügel in German) is an AI consultant and software builder in Vienna, Austria. He helps SMEs and mid-sized companies use AI deliberately, from use case and model choice to architecture and operations, translating between leadership and engineering. He builds a portfolio of his own software products with AI agents, including FieldProFlow, field sales software for FMCG retail, and is Head of AI Acceleration & Growth at AI Factory Austria AI:AT.

What does an interim AI lead do?

For three to six months they own AI in the company: roadmap, architecture, vendor choice and acceptance. They translate between leadership, business teams and IT until an internal team can carry it. With me from €4,500 per month for one to two days a week.

What does the AI Act require on AI literacy?

Since 2 February 2025, Article 4 of the AI Act requires companies that provide or deploy AI systems to take measures to ensure a sufficient level of AI literacy among their staff. How much is needed depends on use and risk. I train leadership and teams in practice; this does not replace legal advice.

What does an AI architecture review cost?

From €1,500 as a fixed price, excluding VAT where applicable. Within one to two weeks you get a written report on use case, model choice and dependencies with a clear recommendation, plus a 90-minute walkthrough.

Which tools and models do you work with?

For building, coding agents such as Claude Code and Codex; agents write 95% of the code while I own product, architecture, tests and releases. For clients I choose the model per task, often a smaller one that can run in Europe, and never with lock-in to a single vendor.

Do you give talks or sit on juries and panels?

Yes. More than 150 juries, trainings and workshops, most recently as moderator at Proof of Talk in the Palais du Louvre and on juries in Paris and Copenhagen. Topics are steering AI in companies, staying independent of AI vendors and building with agents, in German or English. Requests by email to post@paulkruegel.com.

Do you work with startups?

With funded startups, as a prototype sprint. For early stages, the AI Factory Austria programmes are the better start.

Where, and in which language?

From Vienna for companies in Austria, Germany and Switzerland, on site or remote, in German and English.

How do we start?

With an email to post@paulkruegel.com and a free 20-minute call. After that it is clear which format fits, or whether none does.

Contact

Where are you with AI?

A few sentences are enough: what it’s about, who’s involved, by when. I reply personally.

Start a conversation

Vienna. German and English. LinkedIn · X @paulkruegel

Portrait of Paul Kruegel
Start a conversation