Token Drop

DaaX.ai
  • 5,0 (1)
  • Technologies
  • Chaque semaine

A weekly, unscripted conversation from the DaaX team on the most interesting developments in AI. Sunil Baliga and Sajjad Khazipura (DaaX Co-Founders), along with Sam Pooni (DaaX Architect) and the occasional guest, explore, discuss, and debate new AI research, news, and real-world use cases. Built for developers and business leaders who want perspectives from experienced AI practitioners. All opinions expressed on this podcast are those of the panelists and do not necessarily reflect the views of their employers.

  1. -4 j

    Algorithms Are Logarithmic, AI Is Quadratic — The Hidden Cost of Long Context Windows

    "Algorithms are logarithmic, AI is quadratic.” That one line — from a friend at a party — is the key to understanding why long context windows get so expensive, so fast. This episode examines the quadratic scaling problem behind standard transformer attention. As the number of tokens grows, every token may need to interact with every other token—so doubling the context length can roughly quadruple the attention computation. The panel discusses what this means for million-token context windows, latency, memory use, cost, and answer quality. They also explore advertised versus effective context, the “lost in the middle” problem, and when retrieval-based approaches may be more practical than placing an entire codebase, contract collection, or document library into a single prompt.

    23 min

  2. 26 juin

    Who Decides What's "Safe" AI? Ethics, Trust & the Global Language Divide

    Dr. Sarah Luger, AI safety and evaluation expert, generative AI research director, University of Edinburgh PhD, and a contributor to ML Commons' AI Luminate benchmark, who earlier in her career worked on IBM Watson's Jeopardy! Challenger, joins Token Drop We trace how the industry's language has shifted from ethics to responsible AI to trust and transparency to safety and security, and why that shift matters. Sarah makes the case that "safety" isn't universal: a product that's safe in one market can quietly fail in another, and the hazard categories most benchmarks use are heavily Western- and English-centric. From there the conversation digs into low-resource languages, Indigenous data sovereignty, the economics of sovereign open-source models for the Global South, the US–China compute duopoly, and why leaderboards both pushed the field forward and got overused. A wide-ranging conversation for developers, founders, and anyone thinking seriously about how AI gets built, evaluated, and trusted across cultures. https://www.linkedin.com/in/sarahluger

    36 min

  3. 20 juin

    Your Employees Are Creating AI IP. Is Your Company Capturing It?

    Werner Goertz, former Gartner analyst, former AWS and IBM Analyst Relations leader, Founder of Anicca AR (Agent Relations), joins Token Drop to discuss a growing challenge in enterprise AI: employees are creating valuable prompts, workflows, and AI-assisted knowledge in individual accounts, but much of that value never becomes institutional knowledge. The conversation explores AI governance, enterprise IP, knowledge capture, productizing AI, and how companies can turn individual AI productivity into long-term strategic advantage. https://www.linkedin.com/in/werner-goertz-849173/ https://anicca-ar.com #artificialintelligence

    38 min

  4. 13 juin

    How AI Is Changing Supply Chains

    Supply chain expert Edwin de Boer, VP Supply Chain Strategy at Philips, joins Token Drop to discuss how AI is being used in supply chains today and the key considerations for scaling AI deployments Edwin’s profile: ⁠https://www.linkedin.com/in/edwindeboer/ #artificialintelligence

    34 min

  5. 6 juin

    Why AI Agents Need an Operating System

    In this episode the DaaX team discusses how everyone is building AI agents, but few are talking about the infrastructure required to make them reliable at scale. #artificialintelligence

    27 min

  6. 30 mai

    What Are Ontologies? Their Role in Knowledge Graphs and AI Domain Super Intelligence

    In this episode the DaaX team discusses Knowledge Graphs & Ontologies, AI's need for them, and their link to domain super intelligence

    29 min

  7. 23 mai

    Forward Deployed Engineers (FDEs) in AI - who, what, why, etc.

    In this episode the DaaX team, joined by special guest Satya Mantha⁠, Principal Software Architect at Fission Labs⁠ (a leader in applied AI engineering), discuss FDEs in AI. Satya’s profile: ⁠https://www.linkedin.com/in/ssmantha/⁠ Fission Labs website: ⁠https://www.fissionlabs.com/⁠

    27 min

  8. 16 mai

    A new era of specialized semiconductors for AI inferencing and training?

    Silicon Valley was abuzz this week with the Cerebras Systems IPO. In this episode the DaaX team, joined by special guest Raminda Madurawe (Founder & CEO of AxPro Semi), discuss whether the Cerebras IPO heralds a new era of specialized semiconductors for AI inferencing and training. https://www.eetimes.com/cerebras-ipo-revives-ai-chip-startup-fever/ https://www.linkedin.com/in/raminda-madurawe-a85748ba/ https://axprosemi.com

    42 min
Tout afficher (14)

À propos

A weekly, unscripted conversation from the DaaX team on the most interesting developments in AI. Sunil Baliga and Sajjad Khazipura (DaaX Co-Founders), along with Sam Pooni (DaaX Architect) and the occasional guest, explore, discuss, and debate new AI research, news, and real-world use cases. Built for developers and business leaders who want perspectives from experienced AI practitioners. All opinions expressed on this podcast are those of the panelists and do not necessarily reflect the views of their employers.

Informations

  • Création
    DaaX.ai
  • Années d’activité
    2 k
  • Épisodes
    14
  • Classification
    Tous publics
  • Copyright
    © ℗2026 DaaX Technologies Inc.
  • Site web de l’émission
    Token Drop

Vous aimeriez peut‑être aussi