The Neutral Compute Layer: Breaking the Closed AI Financial Bottleneck with Eugene Cheah

The Neutral Compute Layer: Breaking the Closed AI Financial Bottleneck with Eugene Cheah

Join Eugene Cheah, CEO and Co-founder of Featherless AI, for an eye-opening exploration into the hidden infrastructure crisis threatening the scalability of enterprise artificial intelligence. For the past few years, the corporate tech playbook has been completely dependent on a handful of closed-source foundational models running on single clouds. Yet, as companies attempt to shift from superficial prototypes to full-scale production, the financial reality of token fees has turned into an unsustainable tax—inherently constraining engineering speed and wiping out margins. Drawing from his elite computer science background at the National University of Singapore and his leadership in the Linux Foundation's attention-free RWKV project, Eugene explains why the multi-billion-dollar enterprise landscape must migrate to a neutral open-source framework, and how Featherless AI is making that shift an instant operational reality. 🎯 Insights You’ll Learn: The Token Tax Crisis: Why spending over a million dollars annually on closed-source tokens creates an artificial ceiling on enterprise software development and product margins. The "Elephant in the Room": How high-performing, cheaper, and faster open-weight models are systematically dismantling the idea that enterprises need to pay a premium for massive, general-purpose models. Taming "Token Anxiety": How moving to flat capacity pricing and dynamic scaling allows enterprise dev teams to size their server fleets to throughput needs rather than model count. Replacing the Transformer: The technical architecture behind RWKV—the attention-free open model project under the Linux Foundation designed to completely bypass the memory limits of traditional transformers. The $20M Series A Blueprint: Strategic takeaways from closing their latest institutional round co-led by AMD Ventures and Airbus Ventures to scale their global model marketplace. 🌍 Why This Matters: The ongoing pre-training wars between massive tech labs have obscured the actual bottleneck of enterprise deployment: flexibility and predictability. Forcing a company to hardcode its entire architecture to a single closed vendor leaves them exposed to abrupt API pricing shocks, unexpected model deprecations, and severe data sovereignty vulnerabilities. Eugene Cheah is building the definitive neutral alternative. 👤 Expert Background: CEO & Co-founder of Featherless AI, the San Francisco-based serverless AI platform providing instant API access to the largest library of specialized open models. Co-leader of the RWKV Project under the Linux Foundation Data & AI, pioneering linear-attention architectures that scale to massive contexts with zero quadratic memory penalties. Former CTO & Co-founder of Uilicious, an automated, low-code user interface testing platform utilized by international enterprise dev teams. 🎙️ Hosted by Gary Fowler, CEO of GSD Venture Studios — global AI entrepreneur, investor, and innovation leader spotlighting founders navigating the realities of building in the AI era. 💡 Perfect For: Chief Information Officers, Principal AI Engineers, Enterprise Architects, Venture Capitalists investing in deep-tech infrastructure, and software founders looking to maximize model performance without surrendering control of their unit economics. 🚀 Timely Topic: As enterprise buyers reevaluate their cloud spending in 2026, the ultimate competitive advantage belongs to platforms that can replace volatile, usage-based fees with predictable hardware execution. Subscribe for more global founder conversations from GSD Venture Studios: https://gsdvs.com [https://gsdvs.com] #OpenSourceAI #AIInference #FeatherlessAI #RWKV #CloudInfrastructure #DeepTech #EugeneCheah #GaryFowler #GSDVentureStudios

The Tribal Bottleneck: Why AI Agents Fail Without Structured IT Data with Pinar Ormeci

Join Pinar Ormeci, CEO of Lexful, for an unvarnished examination of the execution barrier stalling the next era of enterprise automation. While the tech industry rushes to build autonomous AI agents capable of handling complex work, they are systematically ignoring a foundational crisis: AI is only as reliable as the data it has access to. Nowhere is this "knowledge debt" more visible than in the Managed Service Provider (MSP) sector—the massive industry that quietly operates the digital infrastructure and cybersecurity for millions of global businesses. Drawing from over two decades of experience scaling global tech giants and leading high-growth startups through major acquisitions, Pinar breaks down why "tribal knowledge" is the ultimate bottleneck for AI, and how Lexful is reengineering how technical knowledge is captured, updated, and used. 🎯 Insights You’ll Learn: The Tribal Bottleneck: Why layering advanced AI agents onto outdated wikis, chaotic Slack channels, and undocumented "engineer memory" results in complete operational failure. The MSP Reality Check: How the teams running the world's commercial IT lines are trapped in an "operational mess," working off documentation that goes stale the moment it is saved. The Multi-Tab Scavenger Hunt: Why manual documentation workflows fail, forcing junior technicians to waste valuable minutes hunting for credentials, passwords, and network layouts during high-stakes outages. Defining AI-Ready Documentation: Moving beyond basic text paragraphs to construct machine-readable, structured relationship layers that AI models can safely reason over. Inside the "Knowledge Ops" Framework: How Lexful’s platform automatically captures, organizes, and maintains IT documentation as engineers work, bypassing the need for manual record-keeping. Deploying Ask Lex: Utilizing an intentional, context-aware AI data assistant to answer complex deployment queries in natural language, dropping technician onboarding timelines from months to days. The $150M+ Executive Playbook: Applying elite operational discipline from Qualcomm, Ericsson, and Timus Networks to build deep, venture-backed enterprise category leaders. 🌍 Why This Matters: In 2026, over 90% of IT organizations acknowledge that artificial intelligence is fundamental to their survival, yet the vast majority state that poor data quality blocks their implementation. When critical infrastructure configurations exist only in a senior engineer's head, the organization is incredibly vulnerable to human churn and tech burnout. Pinar Ormeci is solving this structural challenge. Backed by a $7 million seed round led by Top Down Ventures and York IE, Lexful treats technical documentation not as a passive chore, but as active operational intelligence. By natively mapping relationships between client assets, credentials, and live runbooks from day one, Lexful provides the essential, secure knowledge engine required to transition IT service delivery from chaotic manual response to high-velocity, autonomous execution. 👤 Expert Background: CEO of Lexful, the premier AI-native IT documentation platform transforming knowledge infrastructure for the global MSP community. Former CEO of Timus Networks, a high-growth cybersecurity innovator recently acquired by CyberFOX. Distinguished Global Executive who managed a $150M+ P&L at TNS (Nasdaq: TNSI) through its successful strategic acquisition by Siris Capital. R&D Systems Engineer & Consultant, who launched her career directing systems development at Qualcomm and leading large-scale digital transformation initiatives across Europe for Ericsson. Recognized Industry Leader, named among the Top 50 Women CEOs, holding corporate and advisory board seats across the global cybersecurity and AI ecosystems.

The Identity Blueprint: Fusing Premium Brand Strategy with Frontier AI Execution with Beau Catley

Join Beau Catley, CEO and Founder of Nardo, for a deep dive into why pure software engineering is no longer a sustainable enterprise moat without an intentional brand identity. For years, the traditional tech playbook focused entirely on building efficient features while treating customer experience and branding as an afterthought. Conversely, consumer companies prioritized aesthetics while utilizing manual, fragmented operational backends. Beau argue that the defining enterprises of the next generation will be those that erase this divide entirely—fusing elite cultural brand relevance, high-end user interfaces, and generative AI execution into a single, cohesive product ecosystem. Drawing from a career building one of Australia's most prominent independent streetwear labels (Geedup) and executing premium apparel production lines for digital giants like TikTok and Google.🎯 Insights You’ll Learn:The Brand-Tech Fusion Thesis: Why the multi-billion-dollar enterprise winners of the coming decade will be built on the intersection of highly emotional brand loyalty and deep technical execution.Dismantling the 130-Touchpoint Chain: How Nardo condenses the archaic, manual sportswear design and manufacturing pipeline down to under 30 friction points using integrated software workflows.The AI-Native Design Studio: Leveraging advanced neural networks to instantly convert basic club parameters, sponsor dimensions, and colors into completely custom, production-ready vector teamwear catalogs.Disrupting the Middleman Premium: The strategic pricing math behind undercutting predatory teamwear markups by 20% to 30% while retaining structural gross margins. From Grassroots to Global Scale: Lessons from securing an initial $1.1 million pre-seed round backed by Australian soccer icon Tim Cahill to accelerate expansion across the US, UK, and Middle East. The Streetwear Playbook in B2B Tech: Importing consumer scarcity, high-end fit dynamics, and localized community building directly into a rigid corporate distribution sector.🌍 Why This Matters:Grassroots and semi-professional athletic clubs represent the absolute emotional epicenter of global community sports, yet their procurement backend remains trapped in an offline, spreadsheet-reliant system. Volunteers and club administrators spend countless uncompensated hours chasing fragmented manufacturers, dealing with inconsistent sizing templates, and navigating opaque pricing schemes. Beau Catley is rewriting this broken industry playbook. By positioning Nardo as the definitive operating infrastructure layer for local sports apparel, the platform frees local clubs from severe administrative headaches and returns valuable capital directly to community player fields. For Beau, the ultimate goal is clear: to ensure that local teams at the neighborhood park can access the same elite design toolchains, custom fit profiles, and seamless manufacturing speeds as professional franchises. 👤 Expert Background:CEO & Founder of Nardo, an innovative AI-driven sports technology enterprise digitizing the end-to-end design, sourcing, and tracking lifecycle for global team apparel. Co-Founder of Geedup, scaling the foundational streetwear collective into one of Australia's most recognized independent fashion and lifestyle brands.Founder of Nowear Supply, a premier custom manufacturing agency overseeing high-volume specialized apparel lines for Tier-1 corporate ecosystems including Disney, Google, eBay, and TikTok.B2B Sports Tech Pioneer, recently closing an oversubscribed international capital raise alongside strategic sports icon Tim Cahill to spearhead cross-border market expansion. Subscribe for more global founder conversations from GSD Venture Studios: ⁠https://gsdvs.com⁠⁠#SportsTech⁠ [https://gsdvs.com⁠⁠#SportsTech⁠] ⁠#NardoExperience⁠ ⁠#BrandXTech⁠ ⁠#GrassrootsSports⁠ ⁠#SupplyChainAutomation⁠ ⁠#Teamwear⁠ ⁠#BeauCatley⁠ ⁠#GaryFowler⁠ ⁠

The Org Chart Is the Runtime: AI Shifts to Multi-Agent Workflows with Ege Celik

Join Ege Celik, Co-Founder and CEO of Atlantic AI Inc., for an unvarnished examination of why the current enterprise fascination with text-based chatbots is fundamentally hitting a wall. While the initial wave of corporate AI focused on superficial text summaries and basic Q&A interfaces, it completely ignored the execution bottleneck: work doesn't get done by talking to a blank prompt; it gets done by executing multi-step workflows across an organization's tools, permissions, and reporting hierarchies. Drawing from an exceptional trajectory that spans launching a multi-city digital marketing agency at age 16, serving as an EdTech CMO at 17, and co-developing EEG-guided neurotechnology protocols, Ege is treating the company organization chart not as a static visual graphic, but as the active software runtime layer for enterprise AI. In this episode—following Atlantic AI's recent $5M seed valuation—we explore how the team is shifting the paradigm from generalized copilots to dedicated, role-specific autonomous agents that execute workflows end-to-end. 🎯 Insights You’ll Learn: The Chatbot Fallacy: Why providing employees with a generic chat prompt introduces operational context fragmentation rather than solving it. The Org Chart as Runtime: How Atlantic AI programmatically maps autonomous agents directly to a company's internal reporting lines, tool permissions. From Prompt to Production Workflow: How multi-agent orchestration engines parse a single natural language instruction to simultaneously query Jira, execute record transformations in Salesforce, and log audit trails in Slack. The Neurotech Alignment Pattern: Applying foundational data-isolation frameworks from EEG signal analysis to model-agnostic enterprise retrieval architectures. Row-Level RBAC & Airgapped Security: Designing enterprise data pipelines where customer training data leaks are completely blocked by strict client-managed KMS keys. The Velocity of High-School Sprints: Inside the unconventional operational reality of scaling an active enterprise SaaS platform to Silicon Valley adoption while bypassing traditional corporate gatekeepers. 🌍 Why This Matters: In 2026, the biggest obstacle companies face when adopting AI automation is not baseline model intelligence; it is the extreme fragmentation of institutional knowledge across hidden Slack threads, isolated Notion pages, and disconnected legacy databases. When an enterprise forces an employee to manually copy-paste data between tools just to keep an LLM informed, the productivity gain is completely wiped out. Ege Celik is fixing this broken integration layer. By engineering Atlantic AI to hook natively into corporate tech stacks via one-click OAuth authentication, the platform forms a persistent, secure organizational brain. 👤 Expert Background: Co-Founder & CEO of Atlantic AI Inc., a Delaware-incorporated enterprise agent platform scaled to a $5M valuation alongside co-founder and CTO Ruzgar Imren. Neurotechnology Researcher, who co-developed Auralpha, an advanced engineering initiative investigating EEG-guided acoustic loops to systematically compress cognitive distraction recovery times. Serial Growth Operator, who scaled an independent digital marketing agency serving international hubs from Hong Kong to London at age 16 before stepping in as an EdTech CMO at 17. 🎙️ Hosted by Gary Fowler, CEO of GSD Venture Studios — global AI entrepreneur, investor, and innovation leader spotlighting founders navigating the realities of building in the AI era. Subscribe for more global founder conversations from GSD Venture Studios: https://gsdvs.com [https://gsdvs.com] #EnterpriseAI #AIAgents #AtlanticAI #MultiAgentOrchestration #TechFounders #SiliconValleyTech #EgeCelik #GaryFowler #GSDVentureStudios

The Deployable Core: Nuclear Fusion Shifts to Standard Hardware with Joe Finberg

Join Joe Finberg, Founder and CEO of Laurelin Technologies, for a radical evaluation of why nuclear fusion must transition away from multi-decade, national-lab-scale civil engineering projects. For seventy years, the global race for fusion has been synonymous with massive, capital-intensive facilities housing giant tokamaks and stellarators. Joe argues that if fusion is to solve the clean energy crisis in our lifetimes, it cannot remain a perpetual science experiment; it must become a modular, mass-manufacturable piece of industrial infrastructure. In this episode, we unpack how Laurelin Technologies is bypassing the traditional Deuterium-Tritium roadmap to engineer a containerized, pulsed Deuterium-Deuterium (²H – ²H) reactor using advanced AI-driven plasma simulation. 🎯 Insights You’ll Learn: The Infrastructure Metric: Why true commercial fusion will be won by deployable, containerized hardware footprints rather than centralized, single-site utility megaprojects. The Power of High Beta (β): How Field-Reversed Configurations operate without a restrictive central magnetic column, admitting an order of magnitude higher plasma pressure relative to traditional magnetic confinement. Overcoming the ²H – ²H Penalty: Shifting the harsh Lawson-criterion triple-product penalty away from static thermal design by deploying rapid, repetitive pulsed operations. Direct Electromagnetic Recovery: Moving past inefficient, Carnot-limited working-fluid steam cycles to extract electrical energy inductively right at the plasma boundary. AI-Assisted Kinetic Search: How Laurelin leverages high-fidelity machine learning models to accelerate design iterations and map complex electromagnetic gyrokinetic turbulence waveforms in real time. The 40-Foot ISO Envelope: Why sizing reactor-core hardware to fit standard intermodal transportation lines transforms the logistical, regulatory, and supply-chain realities of nuclear deployment. The Physics-Philosophy Synthesis: How combining deep academic tracks at Columbia and Oxford shapes Joe's execution strategy—balancing absolute scientific rigor with venture velocity. 🌍 Why This Matters: The historic achievement of fusion ignition proved that the core physics work. However, the subsequent challenge is entirely an engineering and deployment problem. Traditional designs demand complex, multi-billion dollar shielding and infrastructure that limit their utility to a handful of global grid positions. Laurelin Technologies is building a different reality. 👤 Expert Background: Founder & CEO of Laurelin Technologies, a San Francisco-based deep-tech startup pioneering containerized fusion reactor architecture. Plasma Physics Alumnus & Researcher, who pursued deep foundational work across electromagnetic gyrokinetic turbulence, tokamak confinement boundaries, and advanced plasma mechanics. 🎙️ Hosted by Gary Fowler, CEO of GSD Venture Studios — global AI entrepreneur, investor, and innovation leader spotlighting founders navigating the realities of building in the AI era. 💡 Perfect For: Nuclear engineers, clean-tech venture capitalists, defense procurement officers, power grid architects, and deep-tech founders exploring how multi-agent AI and advanced simulation compress physical hardware R&D cycles. 🚀 Timely Topic: The next decade of global power generation belongs to clean, autonomous, decentralized baseload systems. Joe Finberg is delivering the precise technology stack and operational philosophy required to transform nuclear fusion into a standardized, transportable asset. Subscribe for more global founder conversations from GSD Venture Studios: https://gsdvs.com [https://gsdvs.com] #FusionEnergy #CleanTech #LaurelinTechnologies #FieldReversedConfiguration #NuclearInnovation