Disrupt or Defend

Less than 20% of Germany's "hidden champions" use AI systematically. That single stat sets the tone for this conversation. Daniel Kazani of Softup sits down with Dr. Annette Doms - EU AI Act strategist, serial founder, and one of the most influential voices on AI in the German-speaking world - to ask the question that most Mittelstand leaders are afraid to answer out loud: are we actually ready?

ㅤ

What comes out of the conversation is honest and specific. German SMEs aren't failing because they lack data or domain expertise - they have more of both than almost any competitor. The problem is the gap between potential and practice. Too much hesitation, too little experimentation, and a regulatory environment that adds uncertainty faster than it provides clarity.

ㅤ

👤 Guest Bio

Dr. Annette Doms is a Munich-based AI strategist, author, and serial founder with a PhD in art history and over 13 years at the intersection of technology and business transformation. She is CEO of ICAA Strategists GmbH, Vice President at the Bundesverband für KI-Transformation e.V., and Founding Partner of MindMash - a Munich community for forward-thinking technologists meeting every second Tuesday. Named a Top 50 Most Influential Woman in AI in 2026, she specializes in helping Mittelstand companies and executive boards build AI readiness strategies, including EU AI Act compliance. Her book "Von Gutenberg zu ChatGPT" frames the current AI era as a civilizational shift - not a product cycle.

ㅤ

📌 What We Cover

• Why the German Mittelstand is "AI curious" but not AI ready - and why that gap is dangerous when competitors are moving fast
• The "Rhinoceros" concept from Dr. Doms's book: how Mittelstand companies are built for durability, not unicorn exits - and what that means for AI adoption
• Why AI readiness starts with clean, structured data in ERP systems - long before any tool gets selected
• The step-by-step roadmap for a first AI implementation: assessment, bottleneck identification, scoped pilot, proof of concept, then scale
• How to think about AI as part of a company's DNA - not an add-on - and why change must be led from the top
• The data sovereignty problem: why Mittelstand companies are hesitant to train on US or Chinese models, and what federated learning and platforms like Catena-X offer as alternatives
• The EU AI Act's current ambiguity: how unclear penalty structures and fragmented European infrastructure are adding hesitation, not confidence
• Where the next three years lead: the companies that operationalize AI now will set the pace for everyone else

ㅤ

🔗 Resources Mentioned

• Dr. Annette Doms on LinkedIn
• ICAA Strategists GmbH
• MindMash - community meetup, every second Tuesday in Munich
• Catena-X - decentralized automotive data ecosystem for sovereign AI infrastructure
• Von Gutenberg zu ChatGPT - Dr. Doms's book on AI as civilizational shift (referenced in episode)
• Softup Technologies GmbH

Most internal tools are outdated before they ever ship. That's the problem at the heart of this conversation - and it's one that almost every technical founder has felt personally. Daniel Kazani sits down with Dario Di Carlo, CEO and co-founder of bricks.sh, to talk about why internal tooling is quietly draining engineering teams, and why the vibe-coding shortcut will make it worse before it makes it better.

ㅤ

Dario built bricks.sh after living the problem firsthand: spending months building admin panels at a previous startup, only to ship something already outdated. The platform auto-generates admin panels directly from your database and keeps them in sync as your schema changes - so your engineers stop rebuilding the same tables, forms, and dashboards over and over again. Softup has seen this same pain up close with dozens of clients, which makes this one of the more honest and grounded conversations on the show.

ㅤ

👤 Guest Bio

Dario Di Carlo is CEO and co-founder of bricks.sh, a Milan-based startup automating the creation and ongoing maintenance of internal admin panels. Before bricks.sh, he was a Product Manager at Nibol - Italy's leading workplace management platform - and spent several years exploring the intersection of AI and APIs. He holds an MSc in Innovation Management from Scuola Superiore Sant'Anna and co-founded bricks.sh with CTO Giuliano Torregrossa. The company raised a €1.6M pre-seed round in early 2026 and has signed up 500+ users across 40+ countries.

ㅤ

📌 What We Cover

• Why internal tools are the most-built, least-appreciated part of any product org - and why that gap creates real friction between engineering and business teams
• The difference between dashboards (read-only data access) and operational admin panels (where you actually update users, flag transactions, process refunds) - and why bricks.sh aims to cover both
• Why vibe-coding your internal tools with Lovable or Claude Code creates a short-term win and a long-term maintenance tax
• The guardrails argument: why giving business users free customization without engineering guardrails turns into shadow IT and shadow AI - fast
• How bricks.sh chose startups and scale-ups over enterprise design partnerships early on, and why that bet was the right one for product-market clarity
• The bricks.sh product thesis: 90% of internal tools across finance, healthcare, logistics, and ed tech look identical - speed of development will beat freedom of customization
• What the future of internal tools looks like: conversational interfaces, MCP-powered data unification, and a single source of truth across Stripe, Intercom, Linear, and Postgres
• The founder side: zooming in vs. zooming out, defending a thesis under pressure, and why emotional stability - not enthusiasm - is the real asset at the early stage

ㅤ

🔗 Resources Mentioned

• bricks.sh - the guest's product
• Retool - mentioned as a primary competitor and a tool Dario used extensively at his previous company
• Lovable - mentioned as a vibe-coding alternative with no guardrails
• Bolt - mentioned alongside Lovable as a low-code/no-code alternative
• Claude Code - mentioned as an AI-native coding tool in the vibe-coding conversation
• Supabase - bricks.sh's current lead database integration
• Stripe - mentioned as a data source for future admin panel unification
• Intercom - mentioned as a data source for future admin panel unification
• Linear - mentioned as a data source for future admin panel unification
• Softup - Daniel's company, referenced in the conversation as a concrete example of the internal tooling problem

The hype surrounding artificial intelligence forces investors to separate true innovation from mere noise. Daniel Kazani, co-founder of Softup, sits down with Blerina Sanocki to explore her exact framework for evaluating early-stage startups.

ㅤ

They discuss the difference between building an AI-first product and being an AI-forward company. Blerina shares why she actively seeks out underrepresented and immigrant founders with a "watch me do it" mentality.

ㅤ

The conversation also highlights how the bar for securing funding has shifted, requiring founders to validate their ideas faster using modern tools. They wrap up by evaluating the real cost of moving a startup to Silicon Valley and why health tech remains the most exciting sector for disruption.

ㅤ

👤 Guest Bio

Blerina Sanocki is the Founder and CEO of TideBorn Capital and an Operating Partner at Dardania Capital. With 14 years of experience driving strategic growth at Google, she brings deep enterprise discipline to startup execution. Today, Blerina invests in early-stage software companies and focuses on supporting underrepresented founders.

ㅤ

She actively deploys capital through The Bond Fund and Portfolia. Originally from Albania, she combines her multicultural background with a passion for inclusive technology and building scalable AI infrastructure.

ㅤ

📌 What We Cover

1. The three core pillars of Blerina's investment thesis: backing underrepresented founders, bettering humanity, and using tech as a force for good.
2. Why the immigrant mentality creates resilient founders who treat rejection as a challenge to succeed.
3. How to quiet the noise in a crowded market by distinguishing between AI-first products and AI-forward companies.
4. The rising expectations for early-stage founders to build, launch, and validate their minimum viable products before seeking capital.
5. When it actually makes financial and strategic sense for an international startup to relocate to Silicon Valley.
6. The massive potential and inherent risks of applying artificial intelligence to genome sequencing and health tech.
7. A call to action for more women to overcome their fears and enter the venture capital and angel investing space.

ㅤ

🔗 Resources Mentioned

1. TideBorn Capital
2. Blerina Sanocki
3. Daniel Kazani
4. Softup Technologies
5. Dardania Capital
6. The Bond Fund
7. Portfolia

The bottleneck in software is no longer building the product: it is getting it into the hands of users. As artificial intelligence compresses development timelines, distribution and go-to-market strategies are becoming the true competitive advantages.

ㅤ

Host Daniel Kazani, co-founder at Softup, sits down with Ben Robinson, CEO of Aperture.co, to examine this shift. Ben explains how his firm acts as an operational VC, embedding growth teams directly into startups to solve the complex challenge of selling to large financial institutions. They look at why the standard subscription model is losing ground to variable and outcome-based pricing, and how embedded finance opens new monetization paths for software founders.

ㅤ

Daniel and Ben also discuss the remaining defensive lines in tech. They emphasize that proprietary data, network effects, and deep-rooted trust are what protect companies in an increasingly fast-paced market.

ㅤ

👤 Guest Bio

Ben Robinson is the CEO and Co-Founder of Aperture.co, a Swiss-based growth and investment partner for the financial services sector. Before starting Aperture, Ben worked as Chief Strategy Officer at Temenos, where he launched the Temenos MarketPlace to connect banks with complementary fintech scale-ups. Today, Ben and his team invest in early-stage European B2B fintech companies. They operate differently from traditional venture capital by running hands-on go-to-market projects and embedding product and marketing experts directly into their portfolio companies.

ㅤ

📌 What We Cover

1. How artificial intelligence compresses venture capital timelines by making it faster and cheaper to build software.
2. Why distribution and go-to-market execution are replacing development as the main bottleneck for startups.
3. The necessary shift from fixed subscription billing to variable, outcome-based pricing models.
4. Why embedded finance changes the way non-financial companies monetize their core products.
5. The most common mistake founders make when scaling their sales teams and stepping away from direct selling too early.
6. Why network effects, proprietary data, and regulatory licenses remain strong defensive positions.
7. The importance of building trust in financial services and how neobanks succeeded by starting with simple wedge products.

ㅤ

🔗 Resources Mentioned

1. Aperture.co
2. Temenos MarketPlace
3. Revolut
4. Stripe
5. Open Solar
6. PayTech
7. EasyJet

Autonomous AI agents promise massive productivity gains, but with autonomy comes severe financial and legal risk. If an AI system deployed in banking falls victim to a prompt injection attack, it could falsify compliance records or authorize illegal transactions. A human might notice a mistake after one or two errors, but an AI could execute thousands of false transactions per second, potentially bankrupting a financial institution.

ㅤ

Host Daniel Kazani sits down with Patrick Munro to examine the legal implications of this technology. Patrick shares a cautionary tale about the dangers of disabling safety features and granting full autonomy without human oversight. They discuss the difference between human-in-the-loop systems and autonomous agents from a legal perspective.

ㅤ

The conversation also touches on data sovereignty, the challenges of running on-premise servers, and why the healthcare sector remains hesitant to adopt automated tools. Daniel and Patrick highlight the need for a pragmatic approach to risk and the importance of establishing clear usage policies.

ㅤ

Guest Bio

Patrick Munro is a technology lawyer operating at the intersection of artificial intelligence, cybersecurity, and IT regulatory compliance. He serves as Legal Counsel for Financial Services at Capgemini and as Of Counsel at the IT law boutique planit//legal. Patrick acts as a dual Subject Matter Expert for AI and Cybersecurity, actively developing legal technology tools like compliance dashboards and contract review assistants. He believes that AI gives lawyers better tools to focus on judgment-intensive work rather than replacing them completely.

ㅤ

What We Cover

1. The financial risks of agentic AI and the dangers of prompt injection attacks.
2. How a single hijacked AI agent could impact compliance records and sanction screenings.
3. The monoculture risk of multiple banks deploying the same vulnerable AI models.
4. Legal liability differences between human-in-the-loop decisions and fully autonomous smart contracts.
5. Personal liability implications for managing directors under the NIS2 Directive.
6. Data privacy concerns with US hyperscalers versus the logistical hurdles of on-premise AI servers.
7. Why highly regulated sectors like healthcare are hesitant about AI adoption due to strict criminal implications for data leaks.
8. The need to define clear use cases and usage policies before implementing AI within a company.

ㅤ

Resources Mentioned

1. Capgemini
2. planit//legal
3. Softup
4. MINDMASH Munich
5. Anthropic Claude and Claude Code
6. COBOL
7. Amazon Bedrock
8. Mistral
9. NIS2 Directive
10. Digital Operational Resilience Act (DORA)
11. EU AI Act
12. EU Omnibus Proposals

Artificial intelligence models are hungry for high-quality human knowledge. For years, developers scraped this data without asking permission or compensating the creators. Now, massive lawsuits are forcing the tech industry to rethink how it acquires training material.

ㅤ

Host Daniel Kazani of Softup sits down with Julie Trelstad to discuss how authors can protect their intellectual property and earn money in the modern era. They explore the shift from unauthorized scraping to legitimate licensing agreements. The conversation covers the critical differences between European and United States copyright laws regarding machine training.

ㅤ

They also examine how specialized, peer-reviewed data is becoming a premium asset for new software applications. Finally, Daniel Kazani and Julie discuss how writers can instruct digital assistants to act as collaborative editors rather than replacements for genuine human creativity.

ㅤ

👤 Guest Bio

Julie Trelstad brings over 35 years of experience to the publishing industry. Starting her career editing technical and architecture books, she navigated major technological shifts from desktop publishing to the rise of digital distribution. Today, Julie serves as the Head of U.S. Publishing at Amlet AI, a public registry dedicated to text and data mining rights. She also runs Paperbacks & Pixels, an author-support studio that helps writers build sustainable businesses and structured sales systems.

ㅤ

📌 What We Cover

1. The ethical and legal fallout from training large language models on pirated books.
2. How Amlet AI uses the International Standard Content Code to create a digital fingerprint for written work.
3. Why most authors are currently opting out of machine training licensing due to wildly fluctuating payments.
4. The structural differences between creator compensation laws in Europe and the United States.
5. How peer-reviewed research will become a premium asset for developers building highly specific software tools.
6. Practical methods for instructing tools like Claude to slow down and preserve an author's original writing style.
7. The future of publishing contracts, in which machine language training becomes a standard secondary right.

ㅤ

🔗 Resources Mentioned

1. Amlet AI
2. Paperbacks & Pixels
3. International Standard Content Code (ISCC)
4. StreetLib
5. Claude
6. ChatGPT
7. Fraunhofer Institute

Traditional real estate systems often come with rules, limitations, and legacy baggage that can stifle creativity. Host Daniel Kazani speaks with Bobby Bryant, M.Ed.x2 about building a solution that bypasses the MLS entirely. Bobby explains the strategy behind hōmhub.ai: a peer-to-peer, AI-powered Real Estate Operating System. They discuss why he chose a non-MLS approach to allow for features like digital offers and commission transparency. Bobby also details the launch of the Global Agent Exchange to unify agents across borders and his plans to bring voice-activated home search to the world.

ㅤ

👤 Guest Bio

Bobby Bryant, M.Ed.x2 is the CEO of DOSS Group, INC. and the founder of hōmhub.ai. A veteran with over 25 years in the industry, he became the first African American to create and franchise a real estate brokerage brand. His work is backed by Amazon and Google. Bobby holds two Master’s Degrees in Education and previously served as a contributor to Forbes.

ㅤ

📌 What We Cover

1. Why hōmhub.ai operates as a non-MLS platform to avoid data restrictions and allow for more creativity.
2. The concept of a property-agnostic marketplace: handling sales, rentals, and wholesale properties in one unified system.
3. How the team applies a "Steve Jobs" philosophy by designing for the consumer experience first and working backward to the technology.
4. The specific features of their AI search: users can speak in over 100 languages or visualize new wall colors and flooring instantly.
5. Creating a "Carfax for homes": allowing owners to upload warranties, receipts, and documents directly to a property profile.
6. The Global Agent Exchange (GAE): a listing service built to standardize real estate practices for modern agents.
7. Plans to expand the platform into international markets like Canada, the UK, and New Zealand.
8. Using data to answer hyper-local questions about neighborhoods: from noise levels to air quality.

ㅤ

🔗 Resources Mentioned

1. hōmhub.ai
2. Bobby Bryant on LinkedIn

Manufacturing generates massive amounts of data, yet many factories still run expensive machinery on settings that have not changed in a decade. Daniel Kazani sits down with Dr. Jonathan Spitz, Founder and CEO of GaussML, to discuss why having data does not always mean having information.

ㅤ

Jonathan explains his "small data" approach to industrial optimization. Instead of requiring months of data cleaning and massive data lakes, his team focuses on rapid experimentation. By running a few targeted tests, operators can find the ideal parameters for processes like laser cutting and injection molding in a single day. Jonathan shares real-world examples, including how a 0.5-gram adjustment saved Coca-Cola 20 tons of plastic a year and how job shops eliminated Saturday shifts by increasing efficiency. The conversation also covers the role of the human operator as a pilot rather than a bystander.

ㅤ

Guest Bio

Dr. Jonathan Spitz is the Founder and CEO of GaussML. Before launching his own company, he served as a Research Scientist at the Bosch Center for Artificial Intelligence, where he applied machine learning algorithms to industrial optimization. He holds a PhD in Mechatronics, Robotics, and Automation Engineering from the Technion - Israel Institute of Technology. Jonathan specializes in "small data" solutions that help manufacturers improve efficiency without complex integration.

ㅤ

What We Cover

1. The difference between being data-rich and information-poor in manufacturing
2. Why traditional deep learning often fails in factory settings due to the need for massive datasets
3. How the "small data" approach works: finding optimal machine settings with minimal experiments
4. Real-world wins: Reducing cycle times by 50% in machining and saving raw materials in bottle production
5. The Coca-Cola case study: How a tiny weight reduction per bottle resulted in massive material savings
6. The "Copilot" philosophy: Why AI should augment the operator's intuition rather than replace it
7. Overcoming the "worker gap" by making expert-level machine operation accessible to newer employees
8. Why is failing during the testing phase necessary to find the true limits of a machine

ㅤ

Resources Mentioned

1. GaussML (Official Website)
2. Optimyzer (Product)
3. Dr. Jonathan Spitz (LinkedIn)
4. Daniel Kazani (LinkedIn)
5. Softup Technologies
6. Bosch (Company)
7. TRUMPF (Company)
8. Coca-Cola (Company)

Real estate runs on data, but most of it is trapped in PDFs, lease agreements, and siloed legacy systems. In this episode, host Daniel Kazani sits down with Dr. Nino Paulus, Co-Founder and CPO of AlphaPrompt, to discuss how generative AI is bringing order to this chaos. Nino explains how his team moved from building simple dashboards to creating an AI that functions like a senior analyst—capable of reading entire data rooms, extracting complex lease terms, and spotting risks that humans might miss.

ㅤ

They discuss the reality of deploying AI in a traditional industry, sharing a story in which their software identified 7 active leases for a property the owner didn't even know they still owned. Nino also opens up about his "live demo" sales strategy and shares his thoughts on the future of autonomous AI agents, including the emergence of "Moltbook," a social network where bots communicate with each other. This is a practical look at how Softup and other tech builders can learn from AlphaPrompt's approach to automation and data structuring.

ㅤ

👤 Guest Bio

Dr. Nino Paulus is the Co-Founder and Chief Product Officer of AlphaPrompt. He holds a PhD from the IREBS International Real Estate Business School, where his research focused on Natural Language Processing (NLP) in the real estate sector. At AlphaPrompt, he leads the development of GenAI solutions that automate due diligence and data structuring for asset and property managers. His work bridges the gap between academic AI research and the practical, messy reality of real estate documentation.

ㅤ

📌 What We Cover

1. The Data Problem: Why the biggest challenge in real estate isn't a lack of data, but the fact that it is unstructured and stuck in "silos" that don't talk to each other.
2. Automating "Monkey Work": How AlphaPrompt uses GenAI to handle the tedious tasks—like typing out rent rolls or checking lease addendums—so analysts can focus on decision-making.
3. The "Live" Sales Pitch: Nino explains why he throws a prospect's actual data room into the tool during sales calls instead of using a canned demo.
4. Red Flag Reports: Moving beyond just data extraction to "risk alerts," such as spotting a break clause that allows a tenant to leave early.
5. The "Lost" Property Story: A case study where the AI found seven active leases in a small German town that the portfolio owner thought they had exited years ago.
6. Bottom-Up Adoption: Why AI initiatives fail when they are top-down mandates and why you need to involve the people doing the daily work to make it stick.
7. The Future of Agents: A look at "Moltbook" (Moltbot), a social network for AI agents, and what happens when bots start communicating and learning from one another without human input.

ㅤ

🔗 Resources Mentioned

1. AlphaPrompt (Guest Company)
2. Moltbook (AI Agent Social Network mentioned by Nino)

Brands are losing visibility because they rely on click-based attribution in a world moving toward answer engines. Daniel Kazani speaks with Malte Landwehr, CPO and CMO of Peec AI, about the reality of AI search. Malte explains why 25% of leads might come from LLMs even when tracking software shows 0%. They discuss the transition from SEO to GEO, why static dashboards are disappearing, and the hard truth that average marketing work is becoming valueless.

ㅤ

Guest Bio

Malte Landwehr is the Chief Product Officer and Chief Marketing Officer at Peec AI. He has over 20 years of industry experience, including roles as VP of SEO at idealo and VP of Product at Searchmetrics. Peec AI helps marketing teams understand and optimize their visibility in LLM-based search engines.

ㅤ

What We Cover

1. Why click-based tracking fails to capture the user journey inside ChatGPT and Perplexity.
2. The specific role of "grounding sources" such as Reddit, YouTube, and LinkedIn in AI responses.
3. How Peec AI simulates user behavior to track brand mentions and sentiment.
4. The shift from static data dashboards to on-demand AI chat interfaces.
5. Why is average marketing becoming free while top-tier marketers become exponentially more effective?
6. The challenge of finding product managers with strong product taste in Europe.

ㅤ

Resources Mentioned

1. Peec AI
2. Malte Landwehr (LinkedIn)
3. Daniel Kazani (LinkedIn)
4. ChatGPT
5. Perplexity
6. Google Gemini
7. Granola
8. HubSpot