Startuprad.io ist die führende Quelle für Nachrichten und Einblicke in das dynamische digitale Ökosystem von Unternehmern und Investoren im deutschsprachigen internationalen Raum.
Unsere detaillierten Analysen und Interviews mit Gründern und Venture Capitalists aus der GSA geben einen Leitfaden für diese pulsierende Startup-Landschaft. Wir konzentrieren uns auf Startups aus Deutschland, Österreich und der Schweiz, die gemeinsam als DACH oder GSA bezeichnet werden, und liefern unseren exklusiven Inhalt wöchentlich direkt auf Ihr Gerät. Unsere Publikation in englischer Sprache hat zahlreiche Anerkennungen erhalten, darunter Platzierungen im Global Top 20 Entrepreneurship Podcast von Apple, im Global Top 100 Tech Podcast von Chartable und im Global Top 0,5% von Listen Notes. Begleiten Sie uns, um stets informiert und in dieser spannenden Gemeinschaft engagiert zu bleiben. Wir erreichen monatlich über 70.000 Unternehmer und Investoren mit unseren Podcasts, YouTube-Blog, Newslettern und Internetradiosender sowie über 250.000 monatlich über unsere Webseiten oder soziale Medien.
Hören Sie jetzt die neuesten Folgen von Startuprad.io Podcast
The Q3 2026 review for Germany, Austria, and Switzerland. Germany passed the Strategic Necessity Test; Q3 posed the Ownership Test.
In this episode: the public prediction card (1 confirmed at 8/10 confidence — in 11 days) · Helsing $1.8B at $18B · Quantum Systems $1.2B · Proxima Fusion €411M · Isar Aerospace reaches orbit Sept 5 · 354 rounds (−11%), Berlin pre-seed −65% · Uber’s $14.8B Delivery Hero offer, Cohere–Aleph Alpha, NVIDIA–Hugging Face · The Exploration Company: $450M + ESA in 48 hours · 3 new Q4 predictions with confidence levels.
E 779 — Leapsome: 10 Years to an Agentic HR Platform
E 779 — Leapsome: 10 Years to an Agentic HR Platform
Leapsome bootstrapped 5 years to a $60M Series A, rebuilt its core as an HRIS in 2024, and now runs EU-hosted AI agents for 2,000+ organizations. Co-CEO Jenny von Podewils explains the 10-year bet.
Chapters:
03:52 Five bootstrapped years
12:44 The 2024 HRIS rebuild
17:04 AI agents in production
30:48 The Density Gap: 1:53 vs 1:299
56:00 The 2028 forecast
Why this episode matters: Every HR vendor claims AI agents in 2026. The real question is who built the data foundation that makes agents work. Leapsome’s 2027 Workforce Trends Report (100 high-growth tech companies, 20+ public data points each, no surveys) puts numbers on it: the median company runs 1 HR professional per 53 employees. ElevenLabs runs 1 per 299 with a 4-person People team; OpenAI 1 per 243. That spread — the Density Gap — is the operating-leverage measure of AI-era organization design. This episode is sponsored by Leapsome.
Leapsome — founded Berlin 2016 by Jenny von Podewils and Kajetan von Armansperg; bootstrapped to reportedly $10M ARR; $60M Series A led by Insight Partners with Creandum and Visionaries Club (March 2022)
October 2024 — Leapsome HRIS ships: the data foundation precedes the agents
July 2026 — unified agentic platform: 5 EU-hosted AI agents, EU AI Act-compliant, Recruiting ATS in early access; customers include Notion, Spotify, and Mercedes-Benz
81% of high-growth companies with available data put AI into People functions — most still miss the expected productivity gains
Talent density under fire: metric or layoff euphemism? Jenny von Podewils answers on the record
Related episodes: E 401 — Jenny von Podewils on scaling exceptional organizations (2023): Apple Podcasts | Spotify · E 769 — Talent Without Recycling: The European Scale-Up Question, Part 4 · E 774 — Capital Gravity
Partner with Startuprad.io — sponsored episodes and growth partnerships for companies reaching founders and investors across Europe: https://www.startuprad.io/become-a-partner
Host: Jörn “Joe” Menninger — Startuprad.io, the startup podcast covering startups, tech, and venture capital from Germany, Austria, and Switzerland since 2014.
E 778 — Isar Aerospace, ESA & Europe's Sovereignty Stack
E 778 — Isar Aerospace, ESA & Europe's Sovereignty Stack
E 778 — Isar Aerospace, ESA & Europe's Sovereignty Stack
Europe found the money and the customers in September: Isar Aerospace reached orbit, The Exploration Company raised $450M and signed ESA for up to €760M, and Mistral raised €3B. Ownership is the trade-off — according to KfW, more than 3 quarters of Q1 venture money in German startups came from abroad.
Why this episode matters: Our startup podcast reads the startup scene in Germany, Austria, and Switzerland through one framework this month: capital, customers, ownership. For entrepreneurship and growth partnerships alike, the question is no longer only who funds a company, but who buys first and who owns the upside.
Isar Aerospace — Spectrum reaches orbit from Andøya, 5 September
The Exploration Company — $450M Series C, then ESA Nyx contract worth up to €760M
Mistral AI — €3B Series D at more than €21B; Scaleup Europe Fund co-lead
WIN initiative — more than €25B targeted by 2030
Cohere and Aleph Alpha — definitive agreement; sovereign AI on STACKIT
Quantum Systems — In-Q-Tel opened the US market, co-CEO told Reuters
E 777 — AI Agents, Missing APIs and Mittelstand Germany
E 777 — AI Agents, Missing APIs and Mittelstand Germany
Code costs fell. Software licences did not. Delia Hallberg of Amadeni AI on what actually makes a Mittelstand Germany company fire a vendor it has used for years — and it is not the price.
Part 2 of 2. Horizontal tools stay; everyone needs email and a browser. The exposure sits in rigid vertical products sold to companies whose business model falls between two industries.
The missing API — no open interface means no automation, and every improvement bought from the vendor
The Berlin engineering firm — one feature requested for over a year, no API, proactive move to custom
Langdock — clients who cannot connect their own data because the incumbent is closed
The switching threshold — when the cost of staying exceeds the cost of changing
The 70/30 rule — standard product for most of it, small add-on over an open API for the rest
Digital sovereignty in Europe — a June 2026 export order, models disabled worldwide, and why model-agnostic became a requirement
The leadership risk — three managing directors, two months in, nearly a split
Chapters (estimated — confirm against the final cut)
00:00 Recap and the question
01:30 The developer as head of a department of AI agents
02:40 What breaks first in bought software
07:10 Which products are most exposed
10:30 Where the standard product is structurally better
14:00 The switching threshold and the missing API
17:30 Data ownership and what changed in June
21:00 Switching models in hours
24:10 When change experts belong in the room
28:00 Who owns the project
31:20 Centralised strategy, decentralised usage
33:40 The belief a client project disproved
36:00 Investors, hiring, and the one question first
E 776 — AI Agents Inside a German Mittelstand Company
E 776 — AI Agents Inside a German Mittelstand Company
AI agents are running an IT department at one of the newer German startups — and its co-founder says most companies adding AI to their processes are about to make things worse.
Delia Hallberg founded Amadeni AI in January 2026 and builds AI-supported custom software for Mittelstand Germany. Part 1 of 2 is the diagnosis: software, process or leadership, and why a bad process with AI on top breaks faster, louder and more expensively.
The naming test — if nobody can describe the process without naming a colleague, it is habits, not a process
Impuls KBB — 8 hours to 2 on development reports, €21,000 a month, four-week return
The bakery — 20 systems, 40 media breaks, 35 hours a week of duplicate work
AI agents at work — 80 percent of marketing, plus research, sales handover and multi-agent coding
Not vibe coding — what an elaborate agent process looks like next to a weekend prototype
Ostdeutscher Bankenverband — small tasks, unconnected models, full source traceability
Shadow IT — the private spreadsheet and the WhatsApp group that is really the project tool
Chapters (estimated — confirm against the final cut)
00:00 Cheap code versus bought software
01:10 Software, process or leadership
04:05 Sociology, symptoms and an ERP nobody asked for
06:40 Which AI agents do what four people did
10:15 Quality control on agent output
12:00 Lean company or hidden fragility
15:20 When not to build custom software
18:40 Impuls KBB — 8 hours to 2
23:10 Triage with 40 media breaks
26:30 The one metric worth trusting
29:40 The naming test and shadow IT
33:00 Designing against hallucination
Part 2 lands Tuesday, September 22nd, 2026 — missing APIs and digital sovereignty in Europe.
E 775 — EDIP, SAFE & the EU Defence Fund: Who Can Sell
E 775 — EDIP, SAFE & the EU Defence Fund: Who Can Sell
Germany has the largest defence budget in the European Union — 90.6 billion euros in 2024 — and was not among the 19 member states that applied for SAFE. Austria was not among them either, and the Court’s 2021-2027 country breakdown records no EU defence funding for Austria. Switzerland is not on the eligibility list for any of the 5 instruments, while Liechtenstein qualifies for all of them. Three neighbours, three positions, and almost no coverage of any of it.
Why this matters for founders and investors: Jörn “Joe” Menninger reads the European Court of Auditors’ Review 01/2026 — the EU external auditor’s first look at defence since 2019 — and finds the rulebook sitting inside it. Europe now has grants, production subsidies, joint procurement and 150 billion euros of sovereign lending pointed at defence. Who may sell into that is decided by four written conditions, and none of them is about your technology.
In this episode:
What Review 01/2026 is, who publishes it, and what triggered it
Which instrument is for whom — and why SAFE lends to states rather than funding companies
The 4 conditions: establishment, control, component origin, design rights — and the document behind each
Germany, Austria and Switzerland on three different sides of the same rules
Why the Council of the EU’s own SAFE page lists Switzerland, and why Regulation (EU) 2025/1106 does not
The Monday checklist: what to pull, who owns it, and how long it takes to fix
