🎙️ Backstage Tech by George Helgesen

Jesper Bylund is the co-founder and Chief Product Officer of INXM — a Berlin company building an AI-powered process execution engine for enterprises. INXM's bet is compiled AI: instead of letting an LLM do the work directly, the AI writes a deterministic plan that runs exactly the same way every time — auditable, reliable and repeatable thousands of times. Jesper started in game design, then spent 15 years as the second designer startups called in after the first version failed. INXM raised €5.7M at pre-seed and works with enterprises in retail, industrial engineering and production.

In this episode, Jesper breaks down why LLMs fall apart at enterprise scale, and why delegating all knowledge work to AI is a market big enough for a €5.7M pre-seed round.

Topics covered:

• Why a 98% accurate LLM drops to a coin flip after about 35 steps — and why enterprises never do anything just once
• Compiled AI: an LLM writes the plan once, then the plan runs 10,000 times with guaranteed outcomes
• Semantic drift and wrong numbers aren't bugs — they're how LLMs work
• How enterprises actually roll out AI — forward-deployed engineers, service agreements and long procurement
• Why AI won't replace product designers, but will replace a lot of the effort
• Never trust an LLM with math — and the AI tools a CPO who tests a new one every other day actually keeps

If you're bringing AI into an enterprise, or building an AI product that has to work the same way every time — this episode is for you.

👉 Follow George:

• LinkedIn: https://www.linkedin.com/in/georgehelgesen/
• YouTube: https://www.youtube.com/@georgehelgesen

👉 Follow Jesper:

• LinkedIn: https://www.linkedin.com/in/jesperbylund/
• INXM: https://inxm.ai

Minh Hoang is a product leader and community builder, working on AI products for highly regulated industries like banking, insurance and government. Minh led product at OpenBB, the #1 open-source project in finance, used by 300+ organizations including some of the biggest banks and financial institutions. AI Builders Vietnam, the community Minh founded, brings together developers and founders shipping global products out of Southeast Asia.

In this episode, Minh breaks down why AI made building the easy part, and what your moat is once software becomes a commodity.

Topics covered:

• What to build and why — the 2 questions every builder has to answer now
• What it takes to turn a working product into a sustainable business
• How commercial open source makes money and gets sold to enterprises
• What your moat is when software becomes a commodity — and how AI is changing the engineer's job
• Building global products from Vietnam — and why more founders are moving there

If you're launching a new AI product, offering an open-source solution, or simply curious about product leadership — this episode is for you.

👉 Follow George:

• LinkedIn: https://www.linkedin.com/in/georgehelgesen/
• YouTube: https://www.youtube.com/@georgehelgesen

👉 Follow Minh:

• LinkedIn: https://www.linkedin.com/in/minhhoang1023/

Peter Damm is the co-founder and CEO of Granite — a company building modular data centers for the agentic cloud: shipping-container and rack-sized units placed right next to available energy, instead of waiting years for a hyperscale site. Granite came out of a venture studio exploring the agentic economy. The team tested agent products first, then realized the real bottleneck was underneath — power, hardware and who controls it. Granite plans 10,000 modular data centers across Europe over the next 5 years, with the first deployments going live in Denmark and Belgium.

In this episode, Peter breaks down why AI's shortage is really an energy problem, and why an open model you can audit beats a frontier model you can't.

Topics covered:

• The UK pays offshore wind farms billions a year to switch off — while data centers can't get power
• Up to 90% of AI compute is wasted on overhead — the Hummer vs Prius problem
• Who controls the weights controls the alpha of your company
• Chinese open models aren't more dangerous than private ones — the difference is you can audit them
• Shadow AI: 94% of briefed employees leaked something into a chatbot within a month
• Storage with 20% overhead instead of 400% — and why that makes it quantum safe

If you're building on someone else's models, running AI infrastructure, or wondering why your AI bill keeps climbing — this episode is for you.

👉 Follow George:

• LinkedIn: https://www.linkedin.com/in/georgehelgesen/
• YouTube: https://www.youtube.com/@georgehelgesen

👉 Follow Peter:

• LinkedIn: https://www.linkedin.com/in/peterhdamm/
• Granite: https://www.graniteinfra.ai/

Louis Nicholls is the co-founder of SparkLoop — the #1 recommendation network that connects the best newsletters across every platform so they can grow each other. SparkLoop exists because newsletters are the one content type with no discovery tool. Podcasts have Spotify and YouTube. Writing has the social platforms. Newsletters have nothing, so somebody had to build the layer that helps readers find them. It started as a favor to a friend who kept getting the same request from his customers, and turned into a separate company six years ago. James Clear was one of the first 50 customers. SparkLoop was acquired by ConvertKit, now Kit, in 2023.

In this episode, Louis breaks down why the newsletter market is more fragmented than any other content category, and why publishers bidding against each other on Meta is a problem they can solve without Meta.

Topics covered:

• Why breaking past a million subscribers is a completely different game than getting to a million
• SparkLoop only lets in the top 1% of newsletters — and the flywheel that comes from turning everyone else away
• Google and social traffic drying up, and what publishers are doing instead
• How publishers buy a subscriber from Meta for $2 and earn it back the same day
• The blank slate moment: 2-3% conversion vs 40-50%
• How 1,000 subscribers turned into $1M in revenue

If you run a newsletter, or you own an audience you're trying to turn into a business — this episode is for you.

👉 Follow George:

• LinkedIn: https://www.linkedin.com/in/georgehelgesen/
• YouTube: https://www.youtube.com/@georgehelgesen

👉 Follow Louis:

• LinkedIn: https://www.linkedin.com/in/louisnicholls/
• SparkLoop: https://sparkloop.app

Jens Larsson is the founder & CEO of Raqt — a 🇸🇪 Swedish sports tech company and the leading pickleball platform, connecting players, events and communities across 38 countries. Raqt didn't start as software. Jens found pickleball around 2017, co-started the Swedish Pickleball Federation, and launched one of Europe's first pickleball equipment web shops because nobody in Europe was importing the gear. The platform came later, built by people who were running the tournaments themselves. Eight more countries open up within three months.

In this episode, Jens breaks down why launching on two national championships in a single weekend beat any MVP they could have shipped, and why rev share is a better model than SaaS for event software.

Topics covered:

• Why Raqt skipped "test small, fail fast" and went live on two national championships at once
• The scheduling module that placed 6,500 matches in 10 minutes — and squeezed in 15% more registrations
• How the right software cuts 20% of a tournament's staff
• Gary Vee's pickleball bet — from $400,000 teams to a $15,000,000 target valuation
• Why AI means anyone can build tournament software now, and why Jens is fine with that
• Pickleball vs padel — and why "which sport is growing faster" is the wrong question

If you're a founder building in a category that barely exists yet, or an operator trying to squeeze margin out of live events — this episode is for you.

👉 Follow George:

• LinkedIn: https://www.linkedin.com/in/georgehelgesen/
• YouTube: https://www.youtube.com/@georgehelgesen

👉 Follow Jens:

• LinkedIn: https://www.linkedin.com/in/jensflarsson/
• Raqt: https://raqt.com

▶️ Watch the full video on YouTube: https://youtube.com/@georgehelgesen

👉 Let's connect on LinkedIn — send me a request and mention the podcast.

In this episode of my micro-podcast, I break down the 3 business ops B2B software and service founders should automate first — the ones that quietly eat 40–50% of revenue.

Most founders scale the expensive way. More clients means more people to deliver, and operating costs climb right along with revenue. That's running faster on a treadmill — a lot of sweat, same spot.

After 8+ years in the service business, I see the same three areas bleed money in almost every company I look at. All three are mostly copy-paste work, and AI automation cuts 30, 50, sometimes 80–90% of that cost — without cutting people:

1. Proposals — stop rebuilding the same 80% across Google Docs, Canva, and PowerPoint every time. Your rep sends the offer while the deal is still hot
2. Contracts — when the client says "I'm ready," the contract goes out in 15 minutes, not two days. Pick options from a questionnaire instead of typing fields by hand
3. Invoicing — one click after signature: the invoice lands in QuickBooks, the payment link goes out, and the payment matches itself. This saved one client 20+ hours a month

I also walk through the 3-level scoring system I use to grade each area — Level 0, 1, or 2 — so you know exactly where your money and time leak today.

If you own, fund, or lead a B2B software or service company and your ops costs grow every time you land a new client, this episode shows you where to start.

👉 Want to talk through how to cut your ops costs and scale your company? Email me at [email protected] or connect with me on LinkedIn.

▶️ Watch the full video on YouTube: https://youtube.com/@georgehelgesen

👉 Let's connect on LinkedIn — send me a request and mention the podcast.

In this episode of my micro-podcast, I break down 7 roles where AI lets you grow output without growing headcount — pulled from real automations I've built with B2B founders.

Pulling from 8+ years working shoulder-to-shoulder with software founders, here's where AI takes the first pass so your small team can take the money pass:

1. Sales Rep — an AI voice bot calls warm leads minutes after they submit a form, before a human rep even sees the notification. The human takes over the second the lead is engaged
2. Web Designer — ship the MVP or landing page while your team's offline. AI does ~90%, a designer with taste refines the rest. Speed wins deals
3. QA Engineer — as dev speeds up, testing matters more. Point AI at the branch, check it against the specs, run it in the browser, and hunt for what's broken
4. Copywriter — feed it everything you've written, set the voice and audience, and it drafts blog content that actually sounds like you — not generic AI mush
5. Data Researcher — turn a name and a corporate email into real research on the company and the person, then pipe it straight into the first call or email
6. SMM Manager — one person runs the whole content pipeline at today's speed: clips, voice, video, lipsync, and graphics
7. Software Engineer — the one role you DO hire. Engineers don't write code by hand anymore. Your harness — not the model — decides how productive they are

If you own, fund, or lead a software product with a small team that struggles to deliver, this episode shows you how to move faster and win more business without growing your monthly burn.

👉 Want to talk through how to scale your software company? Email [email protected] or connect with me on LinkedIn.

Thomas England is the CTO of Origin Smart Controls — a company building software for IoT device management in commercial buildings. Origin came out of the mechanical and electrical engineering world, not the software world. That's the whole point: mechanical engineers and software engineers build the platform together, so it solves the problem the way a building engineer understands it, not the way a developer assumes it. Their clients include the NHS and a major UK broadcaster.

In this episode, Thomas breaks down why velocity is the most dangerous thing AI gave engineering teams, and why point-solution SaaS companies should be watching their pricing very closely.

Topics covered:

• Why most IoT dashboard platforms fail — they give you a toolbox, not a solution
• The engineering triangle: quality, accuracy, velocity — and why AI breaks the balance
• Why you can't blame Claude or Codex when your production breaks
• Two types of AI resistance in engineering teams and how to handle each
• Why point-solution SaaS is in trouble — and what Atlassian should be worried about

If you're an engineering leader figuring out AI adoption, or a founder building in an old-school industry — this episode is for you.

👉 Follow George:

• LinkedIn: https://www.linkedin.com/in/georgehelgesen/
• YouTube: https://www.youtube.com/@georgehelgesen

👉 Follow Thomas:

• LinkedIn: https://www.linkedin.com/in/thomas-england-09334280/
• Origin Smart Controls: https://originsmartcontrols.com/

▶️ Watch the full video on YouTube: https://youtube.com/@georgehelgesen

👉 Let's connect on LinkedIn — send me a request and mention the podcast.

In this episode of my micro-podcast, I'm breaking down everything I learned in 8 years building software with B2B founders — 10 lessons in 12 minutes.

Pulling from 8+ years working shoulder-to-shoulder with software founders, here's what separates the products that scale from the ones that stall:

1. Integrate from day one — don't ask people to change how they work. Plug into their existing tools. People don't resist technology, they resist change
2. Distribution is not an afterthought — your product won't sell itself. Pay partners for customers who subscribe AND pay, not just signups
3. Prototype at zero cost — build the interface in Claude, show real users, validate, THEN spend on development. No more $50K MVPs nobody wanted
4. Every dollar needs an ROI — run the numbers on every single feature. 5x return? Ship it. No clear return? Skip it, no matter how cool it is
5. Product plus service — sell the service first, then let the product scale what the service delivered. Vsimple and Dig Insights both lead with service
6. Use external teams as extensions — outsource the boring stuff so your core team stays excited. Happy engineers build better products
7. Become AI-native — record every call, automate everything you can, but keep a human in the loop. AI is your co-pilot, not your autopilot

If you own, fund, or lead a software product, this episode gives you a practical 10-lesson framework to audit how you build — before you burn budget on things nobody wanted.

👉 Want to talk through how to scale your software company? Email [email protected] or connect with me on LinkedIn.

▶️ Watch the full video on YouTube: https://youtube.com/@georgehelgesen

👉 Let's connect on LinkedIn — send me a request and mention the podcast.

In this episode of my micro-podcast, I'm breaking down 7 things every software founder must do to survive while AI eats the SaaS market around them.

Pulling from 8+ years working shoulder-to-shoulder with software founders, here's what separates the products that become indispensable from the ones that get copied over a weekend:

1. Educate your customers — if they can't figure out your product without emailing support, your onboarding is broken. Build real walkthroughs, not help docs
2. Fix customer experience — pick up the phone. Email support is a churn accelerator. Adyen does it right. Stripe doesn't. That gap is a competitive advantage
3. Be more than software — stop selling a login, start selling outcomes. Dig Insights and Vsimple bundle services with their product. That's how you become a growth partner, not a vendor
4. Sell value, not features — Leadfeeder calculated my 25x ROI in the first five minutes. ContactOut just sold me contacts. One closed the deal, one didn't. Train your team accordingly
5. Remove UX bottlenecks — your users aren't telling you where they're stuck. They're just leaving. Use heatmaps and session recordings. Fix what's broken before it kills retention

If you own, fund, or lead a software product, this episode gives you a practical 7-point checklist to audit your product right now — before your customers start building their own version.

👉 Want to talk through how to make your SaaS indispensable? Email [email protected] or connect with me on LinkedIn.