Ignite: Conversations on Startups, Venture Capital, Tech, Future, and Society

Metal's fundraising data suggests that process velocity, not traction or founder experience, is the strongest predictor of fundraising success.Gul Usman, founder and CEO of Metal, has experienced both sides of venture capital fundraising. After working at Mastercard, Marqeta, and DoorDash, he founded Airlift in Pakistan, raising $120 million in venture financing before the company shut down after failing to secure its Series C. Now, backed by a16z and Y Combinator, he's building Metal, an AI-powered fundraising platform serving hundreds of founders.In this episode of the Ignite Podcast, host Brian Bell sits down with Gul to examine why fundraising remains inefficient and what actually separates founders who close rounds from those who don't.Drawing on data from thousands of fundraising processes, Gul reveals that only about 7% to 20% of fundraising attempts succeed, depending on the stage. Surprisingly, Metal's analysis identifies process velocity as the strongest predictor of success, ahead of traction, founder background, and existing investors. Founders who move investors through their pipelines quickly, replace stalled conversations, and maintain momentum have a meaningful advantage.Gul also challenges conventional startup growth advice. After Airlift expanded aggressively during the 2021 funding boom and ultimately shut down, he adopted a different philosophy at Metal: prioritize long-term positioning over maximizing quarterly growth. He explains why independent thinking, customer retention, and strategic urgency matter more than simply moving faster.In Today's Episode We Discuss:00:01 Introduction to Gul Usman and Metal00:48 From Lahore to Mastercard, Marqeta, and DoorDash03:31 The DoorDash experiment that broke Salesforce08:52 Why career growth isn't linear10:37 Building Airlift and solving Pakistan's transportation problem13:25 Why Airlift shut down after raising $120 million16:05 The fundraising problem that inspired Metal19:25 Y Combinator, experimentation, and finding product-market fit22:47 Solving fundraising software's retention problem26:01 How Metal matches founders with investors30:03 Why sustainable growth beats short-term acceleration32:48 Strategic urgency versus bias for action35:29 Why most fundraising processes fail40:02 Making investor updates more effective42:12 Why warm introductions outperform cold emails46:02 The strongest predictor of fundraising success48:30 Closing thoughts and Metal's futureThe conversation includes Gul's DoorDash experiment that persuaded restaurants to offer 32% commissions by modeling incremental revenue, Metal's six-landing-page experiment during YC, and a practical technique for removing unresponsive investors from fundraising pipelines. Brian also explains how specific round-progress updates can revive investor conversations.The deeper lesson is that lasting entrepreneurial advantage comes from independent judgment, disciplined preparation, and knowing where to direct effort.Pull Quotes:"It's not your capital, it's not your traction, it's none of those things, is actually your process velocity.""The preparation goes into a round is is where the magic happens."Subscribe on Spotify: https://open.spotify.com/show/6Ga6v0YUsHotLhjap67uu5Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/ignite-conversations-on-startups-venture-capital-tech/id1709248824Follow Gul Usman on LinkedIn: https://www.linkedin.com/in/gulsf/Follow Gul Usman on Twitter: https://x.com/usmangulFollow Brian on Linkedin: https://www.linkedin.com/in/bblinkedin/Visit Our Website: https://www.teamignite.vc🎧 Watch, listen, and follow on your favorite platform: https://www.teamignite.vc/podcast💬 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcastPre-order my book LP: The Insider's Guide to Investing in Venture Capital Funds (https://www.thelpbook.com/)

Reliabl founder Annie Brown discovered that a social media platform with 100,000 users needed something traditional AI moderation systems couldn't deliver: contextual accuracy.Annie Brown is the founder and CEO of Reliabl, an AI training and evaluation platform built around collaborative data annotation. Her unconventional path spans history and gender studies, microfinance, AI bias research, and communications roles at Y Combinator-backed startups Superside and SafetyWing. Her first social media platform reached 50,000 users in six months and 100,000 within a year, exposing critical weaknesses in automated content moderation. That experience led to a patented approach to improving AI through participatory labeling and expert knowledge.In this episode of the Ignite Podcast, host Brian Bell explores Annie's argument that more training data cannot fix an AI system built on flawed assumptions. The problem often begins before model training, when teams define labels, establish taxonomies, and decide whose interpretation of the data matters. Annie describes an audio dataset in which Black women's neutral voices were repeatedly labeled as angry, demonstrating how annotation errors can become systematic model failures.Annie also challenges the economics of traditional data labeling. Instead of outsourcing enormous volumes of repetitive annotation work to underpaid workers, Reliabl brings domain experts and affected communities into the process. Her central claim is that smaller, carefully labeled datasets can train models just as quickly, or faster, while producing better results.In Today's Episode We Discuss:00:00 - Annie Brown's Background and Journey into AI04:19 - From Y Combinator to Founding Reliabl09:01 - How Collaborative Data Annotation Works11:04 - The Hidden Costs of Bad AI Training Data13:13 - Why Expert Knowledge Beats Crowdsourced Labeling14:59 - Reliabl's Role in AI Training and Evaluation17:34 - Human-in-the-Loop AI and Community Mapping19:18 - Building Software Through Co-Design and User Feedback21:36 - Testing Chatbots for Gender Bias25:28 - How Taxonomy Design Shapes AI Bias27:28 - Rethinking Data Labeling and Worker Compensation30:07 - Will AI Data Workers Automate Their Own Jobs?33:06 - Reliabl's Future and AI for Cities35:37 - Will AI Eventually Eliminate Human Annotation?40:06 - AI Existential Risk and Responsible Development46:04 - Changing Perspectives on Politics and Community Impact50:55 - Why More Data Won't Fix Broken AI52:30 - Curiosity, Mentorship, and Building Startups54:34 - Closing Thoughts and Where to Find AnnieThe deeper lesson is an old one: better judgment begins with better definitions, and technology inherits the assumptions of the people who build it.Pull Quotes"if you have less data but it is labeled better, your model will train just as fast, if not faster.""you're pushing more data through a tainted system and then it's coming out tainted and you're wondering why."Subscribe on Spotify: https://open.spotify.com/show/6Ga6v0YUsHotLhjap67uu5Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/ignite-conversations-on-startups-venture-capital-tech/id1709248824Follow Annie Brown on LinkedIn: https://www.linkedin.com/in/andreafrancesb/Follow Annie Brown on Twitter: https://x.com/andreafrancesbFollow Brian on Linkedin: https://www.linkedin.com/in/bblinkedin/Visit Our Website: https://www.teamignite.vc🎧 Watch, listen, and follow on your favorite platform: https://www.teamignite.vc/podcast💬 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcastPre-order my book LP: The Insider's Guide to Investing in Venture Capital Funds (https://www.thelpbook.com/)

Big Context abandoned a seven-figure AI platform growing nearly 30% monthly to build an AI-native services business instead.Zac Choi, founder of Big Context, has spent over two decades working in enterprise data, analytics, and AI. His career spans Wharton, General Mills, Clorox, Green Dot's IPO, and McKinsey, where he launched Fuel, a practice serving growth-stage technology companies. He previously founded String AI, an AI-powered messaging platform acquired by a telecommunications company. Today, he is building Big Context and developing Open Pragmatics, an open standard for enterprise AI context.In this episode of the Ignite Podcast, Brian Bell and Zac examine why enterprise AI projects repeatedly fail despite increasingly capable models. Zac argues that the fundamental obstacle is not model intelligence or access to more data. It is the missing business context that allows AI systems to interpret information correctly. A CFO, salesperson, and marketer can all define revenue differently, yet most enterprise systems never capture those distinctions in a form AI can reliably use.Zac challenges another foundational assumption: data gravity. He argues that enterprises remain dependent on their existing systems because the meaning attached to their data is difficult to move, not because moving the data itself is inherently expensive. His solution involves making context portable, opening the underlying standards, and turning complex implementation work into repeatable, AI-powered services.In Today's Episode We Discuss:00:00 - Introduction to Zac Choi and Big Context00:32 - From Wharton and McKinsey to Entrepreneurship04:37 - Building and Selling String AI07:19 - The Origins of Big Context11:21 - Why Enterprise AI Pilots Fail13:00 - The Missing Context in Enterprise Data17:28 - Why Open Pragmatics Is Open Source22:24 - Services as Software and AI-Native Consulting24:51 - Automating Enterprise Data Migrations27:32 - Sovereign Context and Enterprise AI Models29:20 - Building Trust with Fortune 500 Companies31:16 - Why the Modern Data Stack May Collapse33:28 - The Rise of Context Engineering35:41 - Why Data Gravity Is an Illusion37:43 - The Best Advice Zac Ever Received38:31 - Why Net Revenue Retention Matters More Than ARR41:31 - Why Founders Must Learn to Pivot44:04 - The Decision to Abandon a Growing Platform45:52 - Reinventing Enterprise Software Delivery49:57 - Where to Find Zac ChoiZac explains how Big Context compresses enterprise BI migrations that traditionally take months into days, why it paused a platform growing nearly 30% month over month for 15 to 16 months, and how a Fortune 500 procurement process changed after the company adopted a services-led model. He also shares his father's advice about separating work and play and why net revenue retention reveals more about customer value than ARR alone.The enduring advantage in technology may be less about controlling information than understanding what it means.Pull Quotes"I think data gravity is an illusion.""I think ARR is stupid."Subscribe on Spotify: https://open.spotify.com/show/6Ga6v0YUsHotLhjap67uu5Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/ignite-conversations-on-startups-venture-capital-tech/id1709248824Follow Zac Choi on LinkedIn: https://www.linkedin.com/in/isaacchoi/Follow Zac Choi on Twitter: https://x.com/w3bzacFollow Brian on Linkedin: https://www.linkedin.com/in/bblinkedin/Visit Our Website: https://www.teamignite.vc🎧 Watch, listen, and follow on your favorite platform: https://www.teamignite.vc/podcast💬 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcastPre-order my book LP: The Insider's Guide to Investing in Venture Capital Funds (https://www.thelpbook.com/)

A cybersecurity startup raised over $100 million, reached nearly $30 million in revenue, and was sold for $1 after a cash flow crisis.Mahendra Ramsinghani, founder and managing partner of Secure Octane, joins Brian Bell on the Ignite Podcast to examine the uncomfortable realities of venture capital. He leads a San Francisco seed fund focused on AI and cybersecurity, with exits involving Palo Alto Networks and Microsoft, wrote The Business of Venture Capital (now in its fourth edition), and teaches venture capital at UC Berkeley.Mahendra argues that venture capitalists dramatically overstate their ability to predict winners. Behind polished theses and LP presentations, luck, timing, relationships, and access often decide outcomes. His own record includes a 16x return on a $50,000 Demisto investment, an 8x return that began with a chance encounter at the RSA Conference, and a cybersecurity company that collapsed despite raising substantial capital.The conversation challenges several assumptions. Why do managers insist on concentrated portfolios when outcomes follow power-law distributions? Are megafunds keeping companies private too long? And why do institutional LPs sometimes prioritize protecting their careers over backing emerging managers?In Today’s Episode We Discuss:00:01 Introduction to Mahendra Ramsinghani and Secure Octane00:50 Growing Up in India and Entering Venture Capital03:01 Why Venture Capital Is a Slow Path to Wealth05:36 From NVIDIA Employee to Raising a $20 Million Fund06:42 How a Cold Email Led to a Friendship With Brad Feld08:31 Building Secure Octane’s Cybersecurity Investment Thesis10:39 The Chance Encounter That Produced an 8x Investment12:12 How Demisto Generated a 16x Return14:27 Why Venture Capitalists Avoid Discussing Failure16:11 The $100 Million Startup That Sold for $119:37 How Venture Capital Has Changed Over the Decades21:03 Why Megafunds Keep Companies Private Longer28:04 The Psychology of Venture Capital and Open-Heartedness30:43 The Art of Saying No to Founders35:17 What Separates Professional VCs From Angel Investors38:15 Does Geography Still Matter in Venture Capital?40:00 When Does an Emerging Manager Become Established?41:25 Institutional LPs and Solo GP Succession Risk54:12 Why Institutional LPs Avoid Career Risk59:13 Challenging Traditional Portfolio Diversification01:01:42 The Myth of Venture Capitalists Picking Unicorns01:06:25 The Resilient Founder and Mental Health in Startups01:09:19 How Investors Can Support Struggling Founders01:11:57 AI’s Impact on Venture Capital and Entrepreneurship01:15:40 Access to Y Combinator and Competitive Deal Flow01:19:05 Ethics, Management Fees, and Long-Term Relationships01:21:57 Closing Thoughts and Where to Find MahendraMahendra also shares how a cold email to Brad Feld became a lasting friendship and book collaboration, how a single cash flow disruption destroyed a heavily funded company, and a practical approach to rejecting founders respectfully.Ultimately, venture capital remains an exercise in judgment under uncertainty, where humility matters as much as conviction.Pull Quotes:“Nobody knows. Nobody knows.”“you have to be invisible but present and it’s very hard, you know.”Subscribe on Spotify: https://open.spotify.com/show/6Ga6v0YUsHotLhjap67uu5Follow Mahendra Ramsinghani on LinkedIn: https://www.linkedin.com/in/mahendraram/Follow Mahendra Ramsinghani on Twitter: https://x.com/mahendra_grFollow Brian on LinkedIn: https://www.linkedin.com/in/bblinkedin/Visit Our Website: https://www.teamignite.vc🎧 Watch, listen, and follow on your favorite platform: https://www.teamignite.vc/podcast💬 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcastPre-order my book LP: The Insider’s Guide to Investing in Venture Capital Funds (https://www.thelpbook.com/)

Most machine learning models never reach deployment, and Gooder AI co-founder Eric Siegel argues that better algorithms won't fix the problem.Eric Siegel is the co-founder of Gooder AI and author of The AI Playbook and Predictive Analytics. With more than 35 years in machine learning, a PhD from Columbia University, and experience teaching AI at Columbia, Siegel has spent his career connecting predictive technology to business outcomes. His first book has been adopted by hundreds of universities, and he has delivered 180 paid keynote speeches. He also founded Machine Learning Week, a conference series running since 2009.In this episode of the Ignite Podcast, host Brian Bell welcomes Eric back to challenge the industry's obsession with generative AI. Siegel argues that most organizations should invest at least as much in predictive AI as generative AI, particularly when improving existing business operations. While generative AI attracts disproportionate investment and attention, predictive models can deliver substantial returns by improving decisions about customer retention, fraud detection, credit scoring, and other operational processes.The deeper problem isn't model sophistication. It's deployment. Siegel explains why predictive AI projects stall when data scientists report precision, recall, and accuracy instead of translating predictions into financial outcomes. Meanwhile, generative AI faces a different obstacle: reliability. Even a system that performs correctly 95% of the time may be commercially unusable when its remaining errors carry serious consequences.His proposed solution combines both approaches. Predictive reliability layers can identify high-risk interactions and route them to humans, potentially making partially autonomous systems viable where fully autonomous systems fail.In Today's Episode We Discuss:00:00 Eric Siegel's 35 Years in Machine Learning03:09 Decision-Making Models and Predictive AI06:41 Why Better Data Beats Complex Models09:24 The AI Playbook and the Future of Predictive AI14:15 Audiobook Production and the New Paperback Edition17:30 Why Enterprise AI Projects Fail20:25 Reliability Layers and Human-in-the-Loop AI24:14 The Six-Step BizML Framework30:56 Can Foundation Models Replace Predictive AI?35:27 Who Should Read The AI Playbook?37:58 Why AI Benchmarks Fail to Measure Business Value41:40 AI Doom, AGI, and the Intelligence Explosion Narrative46:19 Eric Siegel's Predictive Analytics Rap47:10 The AI Playbook's Central MessageSiegel breaks down his six-step BizML framework, explains why technical model scores fail to communicate business value, and describes how routing risky AI decisions to human reviewers can rescue otherwise undeployable systems. He also challenges AI extinction narratives, recalls building a home sound booth to narrate his audiobook, and reveals the provocative original title MIT Press rejected.Ultimately, the value of intelligence lies not in how impressive its capabilities appear, but in whether those capabilities improve decisions in the real world.Pull Quotes:"The thing that tends to pay off more isn't com in increasing the complexity of the analytical method, but rather improving the data.""my P doom is zero."Subscribe on Spotify: https://open.spotify.com/show/6Ga6v0YUsHotLhjap67uu5Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/ignite-conversations-on-startups-venture-capital-tech/id1709248824Follow Eric Siegel on LinkedIn: https://www.linkedin.com/in/predictiveanalytics/Follow Eric Siegel on Twitter: https://x.com/predictanalyticFollow Brian on Linkedin: https://www.linkedin.com/in/bblinkedin/Visit Our Website: https://www.teamignite.vc🎧 Watch, listen, and follow on your favorite platform: https://www.teamignite.vc/podcast💬 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcastPre-order my book LP: The Insider's Guide to Investing in Venture Capital Funds

VoiceBit discovered that customers would order pizza from an AI agent even when its first attempt failed to capture the toppings.Jay Patel is the co-founder and CTO of VoiceBit, an AI-powered ordering platform helping restaurants reclaim direct orders from third-party delivery apps. Before founding VoiceBit, Jay worked on Azure support AI at Microsoft, where his team used generative AI to deflect thousands of support cases. He previously worked at Visa on loyalty systems and large-scale data processing. VoiceBit emerged from Berkeley SkyDeck and closed an oversubscribed pre-seed round led by 1984 VC.In this episode of the Ignite Podcast, Brian Bell and Jay examine why building a sophisticated voice agent is only a small part of solving restaurant ordering. Jay explains how VoiceBit evolved from answering phone calls into an integrated ordering platform covering payments, delivery dispatch, websites, and ChatGPT ordering.The counterintuitive insight is that restaurant owners do not actually want AI. They want missed orders recovered, customer relationships restored, and commissions eliminated. Jay explains why phone orders remain surprisingly resilient, how AI-powered upselling generates a reported 5% to 8% revenue lift, and why the company's most important engagement metric became misleading after introducing conversational memory.In Today's Episode We Discuss:00:01 Introducing Jay Patel and VoiceBit00:49 From self-taught programmer to Microsoft AI engineer05:40 The idea behind VoiceBit08:33 The first AI pizza order and its unexpected lesson09:46 How VoiceBit automates restaurant phone orders12:10 Reclaiming customers from third-party delivery apps14:28 Why restaurants must own their customer relationships16:00 VoiceBit's pricing and commission-free model16:58 Measuring AI adoption and customer engagement19:14 How conversational memory changes ordering behavior20:41 Convincing restaurants through real-world pilots22:34 AI-powered upselling and revenue growth24:21 Turning restaurant websites into interactive AI experiences25:22 Why phone ordering remains a major revenue channel26:19 Competing in the restaurant voice AI market29:28 Building the infrastructure for direct restaurant ordering31:03 Why VoiceBit launched ChatGPT ordering32:20 The future of restaurant margins and AI productivity34:59 Moving from Microsoft to startup ownership36:21 Why startups should ship before achieving perfection37:24 What changed Jay's mind about customer trust in AI38:31 Why restaurant owners don't want AI for its own sake39:00 Staying curious and being willing to be wrong41:03 Multilingual AI and restaurant accessibility43:27 Why computer science fundamentals still matter44:34 Jay's interest in building an AI-powered hedge fund46:00 Where to find Jay and VoiceBit46:40 AI existential risk and Jay's 10% P(doom) estimate50:39 P versus NP and the future of computationThe most useful technology often succeeds by making itself invisible, leaving customers with a better service rather than another system to manage.Pull Quotes"What they want is a solution that just works, right?""You have to be owner of everything."Subscribe on Spotify: https://open.spotify.com/show/6Ga6v0YUsHotLhjap67uu5Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/ignite-conversations-on-startups-venture-capital-tech/id1709248824Follow Jay Patel on LinkedIn: https://www.linkedin.com/in/imjaypatel12/Follow Brian on Linkedin: https://www.linkedin.com/in/bblinkedin/Visit Our Website: https://www.teamignite.vc🎧 Watch, listen, and follow on your favorite platform: https://www.teamignite.vc/podcast💬 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast📝 Read the full episode breakdown on our blog: https://www.teamignite.vc/blogPre-order my book LP: The Insider's Guide to Investing in Venture Capital Funds

A Los Angeles dentist increased new patients by 33% and revenue by 23% after turning his skateboarding obsession into a marketing strategy.David Meerman Scott joins Brian Bell for this episode of the Ignite Podcast to discuss The Fandom Playbook, his fourteenth book. David is a bestselling author whose books have sold approximately one million copies across 30 languages. A longtime HubSpot advisor, former technology CMO, and creator of the concept of newsjacking, he has spent decades challenging conventional marketing practices. His latest book translates those ideas into nine actionable plays for building passionate customer communities.David argues that founders make a fundamental mistake when they lead with their products instead of giving people reasons to become fans. Customers can be acquired through transactions, but fandom grows through shared identity, personal connection, and experiences people want to repeat. His central argument is that companies should stop trying to control every interaction and start creating conditions for customers to participate, contribute, and advocate voluntarily.That argument becomes more relevant as AI-generated content floods digital channels. David, who helped popularize digital marketing, now believes genuine human connection offers a powerful alternative to algorithm-driven distribution. Rather than chasing reach or paying influencers who may never use a product, founders can create experiences and resources that turn customers into advocates.In Today's Episode We Discuss:00:00 Introduction with David Meerman Scott00:33 Introducing The Fandom Playbook01:23 Why founders struggle to build fandom02:32 The neuroscience behind fandom and belonging04:30 Writing Fanocracy with his daughter07:08 Brian Bell's book-writing experience08:00 Turning books into a global speaking business09:27 How HubSpot builds community through live events13:03 The nine plays for growing fans13:18 Gamification and the Green Shorts Challenge17:17 HubSpot Academy's certification strategy18:21 Letting go of control over your brand21:05 The Grateful Dead's music-sharing strategy22:00 Turning proprietary data into free resources26:25 Why founders should share personal passions28:00 How a skateboarding dentist grew his practice31:03 Human connection in an era of AI-generated content34:46 Using AI and adult-learning expertise to write a playbook38:24 AI training data, copyright, and author compensation41:28 Why David delayed his AI companion chatbot44:58 How AI is changing marketing and fandom46:58 The dangers of chasing social media algorithms50:01 Why genuine fans outperform paid influencers51:40 HubSpot, inbound marketing, and category creation53:05 How traditional PR can influence AI search results56:34 Where to find David and The Fandom PlaybookThe enduring principle is that people have always gathered around shared interests, and the strongest businesses learn how to earn a place in those communities.Pull Quotes:"The biggest mistake I see is founders focus way too much on their product and service.""When the whole world is making artificial stuff, you can stand out if you have a genuine human connection."Subscribe on Spotify: https://open.spotify.com/show/6Ga6v0YUsHotLhjap67uu5Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/ignite-conversations-on-startups-venture-capital-tech/id1709248824Follow David Meerman Scott on LinkedIn: https://www.linkedin.com/in/davidmeermanscott/Website: https://www.davidmeermanscott.com/fandom-marketingPre-order the book: https://www.amazon.com/exec/obidos/ASIN/1613084978/freshspotpubl-20Visit Our Website: https://www.teamignite.vc🎧 Watch, listen, and follow on your favorite platform: https://www.teamignite.vc/podcast💬 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast📝 Read the full episode breakdown on our blog: https://www.teamignite.vc/blog

A $200 Claude subscription can deliver almost $10,000 worth of usage, but Jonathan Archer thinks those economics will not last.Jonathan Archer is co-founder and CEO of Open LLM, a gateway designed to put multiple AI models, subscriptions, and API keys behind one layer. Before Open LLM, he spent roughly six years advising crypto projects across business development, growth, and community. He met his co-founder at the Avalanche Summit in Buenos Aires before the pair began building side projects together.Open LLM started with a simple frustration: hitting AI subscription rate limits and then wasting time maintaining open-source alternatives. Jonathan’s thesis is that the future will be multi-model, not dominated by one model or one interface. Different models already have different strengths, and users should be able to combine them without constantly moving between products.That leads to his more provocative argument: AI companies should not trap users inside proprietary harnesses. “It’s their tokens,” Jonathan says. Open LLM is being built so users can bring subscriptions and API keys, maintain a fallback chain across providers, share inference between devices, and eventually treat multiple machines as part of one AI infrastructure layer.The conversation also moves beyond routing. Jonathan describes a future where agents spin up machines and subagents, pull inference from another authenticated device, and use lower-cost models for smaller jobs rather than burning the most capable model on everything. Open LLM is also exploring unified memory, one-click VPS deployment, multi-device sessions, and enterprise routing.In Today's Episode We Discuss:00:01 Jonathan Archer and Open LLM03:34 The Origin of Open LLM05:30 AI Subscription Economics07:28 Subscription Routing and Compliance09:01 Unified AI Memory11:04 Multi-Device AI Infrastructure13:19 Self-Hosted Models and VPS Deployment14:43 Open LLM Pricing and Free Tier16:19 Cross-Model Workflows17:03 The Technical Challenge of AI Routing19:10 The Multi-Model AI Future20:24 Free vs Paid Plans21:37 Fallback Chains and Model Selection23:38 AI Subagents and Token Efficiency24:35 Open Source vs Closed Infrastructure26:12 Multi-Device Sessions and Enterprise AI28:15 Breaking AI Platform Lock-In29:53 OpenRouter and the Gateway Model32:06 AI Gateway Acquisition Value34:09 Open LLM and Vercel36:18 Dynamic Model Discovery37:38 Beyond the AI Gateway38:29 Rapid-Fire Questions39:00 Crypto Adoption and Real-World Utility40:56 Crypto, Payments, and StablecoinsOne of the most concrete techniques is Open LLM’s configurable fallback chain, which moves to another model when usage runs out or a provider goes down. Jonathan also explains why subagents should default to less capable models for smaller tasks, comparing the alternative to sending Steve Jobs to retrieve a file. And he describes dynamic discovery that continuously queries providers for newly available models.The bigger idea is older than AI: durable infrastructure often wins by preserving choice rather than forcing everything into one closed system.Pull Quotes“It’s their tokens, they should be able to go.”“We’re gonna live in a multi-model future.”Subscribe on Spotify: https://open.spotify.com/show/6Ga6v0YUsHotLhjap67uu5Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/ignite-conversations-on-startups-venture-capital-tech/id1709248824Follow Jonathan Archer on LinkedIn: https://www.linkedin.com/in/jonathan-archer1/Follow Brian on Linkedin: https://www.linkedin.com/in/bblinkedin/Visit Our Website: https://www.teamignite.vc🎧 Watch, listen, and follow on your favorite platform: https://www.teamignite.vc/podcast💬 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast📝 Read the full episode breakdown on our blog: https://www.teamignite.vc/blog

Autonomize AI now powers four of the five largest U.S. health enterprises, targeting the administrative work consuming clinicians’ time.Ganesh Padmanabhan is the founder and CEO of Autonomize AI. Before founding the company, he spent 11 years at Dell, where he worked across engineering, product management, and general management and helped build what he describes as Dell’s youngest billion-dollar business. He later worked on enterprise AI before starting Autonomize roughly four and a half years ago.Ganesh argues that healthcare’s core technology problem is not a shortage of innovation. It is fragmentation. Decades of point solutions have optimized individual workflows without improving the system as a whole. Autonomize is taking the opposite approach by building infrastructure for AI agents that can work across healthcare administration, including utilization management, care management, claims, payment integrity, and revenue cycle management.His more provocative argument is that deploying agents is only the beginning. Enterprises already have systems for managing human workers, but almost nothing comparable for governing, evaluating, routing, and improving AI workers. Ganesh believes that infrastructure will eventually resemble the ERP, HR, recruiting, and learning systems companies built around human organizations.In Today's Episode We Discuss:00:00 Ganesh Padmanabhan and Autonomize AI00:27 From mechanical engineering to software01:59 Taking apart and rebuilding his grandfather’s compressor03:41 Playing cricket at the national level in India06:35 What Dell taught Ganesh about operational discipline11:15 Why COVID pushed him toward healthcare14:17 Why healthcare’s point solutions fail to compound17:47 Losing a friend to breast cancer and the clinical trial problem21:51 Biotech innovation, regulation, and faster clinical trials24:34 What Autonomize’s AI agents actually do30:32 Genie AI and letting clinicians build workflows33:10 RLM Opt and reducing token usage36:48 Why Autonomize uses a library of roughly 55 models38:13 Why enterprises need infrastructure to manage AI agents40:21 Why today’s AI adoption resembles early cloud migration42:32 Growing from roughly 40 to 140-150 employees43:22 Building company culture when AI can give you easy answers44:59 Building an AI-native health enterprise47:23 Moving healthcare from sick care toward prevention and longevity47:46 Where to find Ganesh and Autonomize AIOne of Autonomize’s earliest projects came directly from Ganesh’s experience watching clinical-trial teams manually search medical records that could run thousands of pages. He also explains how a nurse’s 30-minute chart review might become a two-minute review, why Genie AI puts workflow creation directly into clinicians’ hands, and why Autonomize converts repeatedly proven facts into deterministic code instead of forcing an LLM to reason from scratch every time.The broader lesson is an old one applied to a new technology: durable systems are built by redesigning the work, not merely adding faster tools to the old process.Pull Quotes“Everybody has it, everybody's using it, but we haven't really figured out what it takes to make it work.”“Healthcare is more about care than health.”Subscribe on Spotify: https://open.spotify.com/show/6Ga6v0YUsHotLhjap67uu5Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/ignite-conversations-on-startups-venture-capital-tech/id1709248824Follow Ganesh Padmanabhan on LinkedIn: https://www.linkedin.com/in/padmanabhan/Follow Brian on Linkedin: https://www.linkedin.com/in/bblinkedin/Visit Our Website: https://www.teamignite.vc🎧 Watch, listen, and follow on your favorite platform: https://www.teamignite.vc/podcast💬 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast📝 Read the full episode breakdown on our blog: https://www.teamignite.vc/blog

After 23 years at Google, employee #319 Alana Karen says tech careers are getting harder as AI reshapes jobs and corporate priorities.Alana Karen spent 23 years at Google across search, ads, fiber, and other teams, joining as employee #319. She is the author of the best-selling Adventures of Women in Tech, now in its second edition, which grew out of years of writing about her own performance feedback and interviewing 80 women about their careers in technology.The first edition carried a relatively straightforward message: if you wanted a career in tech, the opportunity was worth pursuing despite the problems. Five years later, Karen is less certain. Layoffs, post-COVID scarcity, AI investment, and changing incentives have created what she describes as a much harder environment where even engineers no longer have the security they once assumed.Her most counterintuitive finding is that many of the problems driving people away from tech are less dramatic than the headlines suggest. The women she interviewed often talked about bad management, shrinking opportunity, constant manager changes, and whether their work still mattered. She also challenges the belief that the best technology must come from five to ten people working around the clock. That model can produce results, but burnout, failure, and organizational damage are real costs.In Today's Episode We Discuss:00:00 Alana Karen's 23 years at Google00:28 Joining Google as employee #31902:11 Growing up with scarcity and choosing a career path07:01 Why staying at Google for 20 years mattered08:45 When is enough actually enough?10:26 Google, AI, and making veterans reprove themselves11:55 How post-COVID scarcity changed Google's culture14:23 Publishing real Google performance feedback16:26 How that feedback became Adventures of Women in Tech19:12 What 80 women actually said about careers in tech23:19 Bad managers and the hidden retention problem25:14 Why the second edition needed to be different28:26 Engineers, AI investment, and the hard tech era31:32 The myth of the five-person team working 24/736:28 Principles, Facebook, capitalism, and business models44:06 What AI fluency now means for your career48:15 How fast AI is improving51:16 Why tech may no longer be the obvious career choice53:34 Trust, ambition, feedback, and career decisions57:50 What Alana hopes people remember59:00 Where to find Alana KarenKaren also explains how publishing years of Google performance reviews became the seed for her book, why she declined to jump to Facebook despite knowing people there, and how Gemini marked up a photo of her car to show where to place Honda's capsicum-infused rat-deterrent tape. She describes sitting through Google leadership presentations that framed small-team startup mode as AI best practice, while questioning whether burning people out should be treated as the only path to breakthrough products.The larger question is whether careers should be built around loyalty to an industry or around the ability to keep adapting as industries change.Pull Quotes“They are willing to fire the engineers to buy the machines to hopefully replace them.”“Don't marry tech and never look around.”Subscribe on Spotify: https://open.spotify.com/show/6Ga6v0YUsHotLhjap67uu5Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/ignite-conversations-on-startups-venture-capital-tech/id1709248824Follow Alana Karen on LinkedIn: https://www.linkedin.com/in/alanakaren/Follow Alana Karen on X: https://x.com/alanatkarenFollow Brian on Linkedin: https://www.linkedin.com/in/bblinkedin/Visit Our Website: https://www.teamignite.vc🎧 Watch, listen, and follow on your favorite platform: https://www.teamignite.vc/podcast💬 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast📝 Read the full episode breakdown on our blog: https://www.teamignite.vc/blog