AI agents move from answering questions to doing the work

The Most Useful New AI Features and Tools to Try

The episode maps a shift from headline-making model launches toward practical agents, enterprise integrations, infrastructure bets, and coordinated defenses.

3 key takeaways
  1. 1Browser agents, customizable harnesses, and temporary-chat controls make everyday AI more capable and manageable.
  2. 2NVIDIA’s reported Hugging Face deal and Salesforce’s AI growth point toward consolidation and partnership across the AI stack.
  3. 3New speech and video systems push AI closer to real-time service interactions and controllable visual environments.

Don't miss

The Hermes Agent segment shows how specialist bots, managed browsing, remote computers, and background work could make AI systems more configurable and operational.

The brief

The episode opens with reported NVIDIA plans to acquire Hugging Face for $12.9 billion, framing the deal as a bid for a fuller open AI ecosystem—and raising questions about Hugging Face’s neutrality.

NVIDIA’s infrastructure expansion and Salesforce’s AgentForce growth complicate the idea that AI simply replaces existing software: capital, chips, data, and enterprise distribution remain decisive.

The governance picture is less settled. A proposed FINRA-like AI self-regulator has stalled, while major companies call for coordinated defenses against AI-enabled cyberattacks.

The practical centerpiece is a new class of agents: Claude Cowork can browse and complete web tasks, while Hermes Agent adds specialist bots, remote computers, model flexibility, and background work.

The roundup closes with ClaudeForce’s Salesforce integration and new speech and video systems, suggesting frontier models may spread through established software as much as replace it.

What was said on this episode

24 statements · 15 positive · 4 negative · 5 neutral

  1. Nathaniel Whittemoreon NVIDIA acquisition of Hugging FaceNeutral1:45

    NVIDIA is paying approximately 80 times Hugging Face’s annualized revenue

    “Hugging Face only has $150 million in annualized revenue, meaning that Nvidia is technically paying around 80 times revenue as a price.”

    Listen at 1:45

  2. Nathaniel Whittemoreon NVIDIA acquisition of Hugging FaceNeutral2:00

    NVIDIA is acquiring Hugging Face for strategic reasons rather than revenue.

    “NVIDIA is very much not buying Hugging Face for their revenue.”

    Listen at 2:00

  3. Nathaniel Whittemoreon NVIDIA acquisition of Hugging FacePositive2:05

    NVIDIA’s Hugging Face deal advances a full-stack open AI business model

    “the deal seems to confirm that NVIDIA is acquiring their way into a full-stack open business model.”

    Listen at 2:05

  4. Hugging Face is the leading distribution channel for open models.

    “Hugging Face is the premier distribution channel for open models.”

    Listen at 2:09

  5. Nathaniel Whittemoreon Open AI modelsPositive3:08

    Many routine AI processes will use open, lightweight, customizable models

    “many of their common and core day-to-day processes are run not on the state-of-the-art frontier models, but on something open, lighter, more nimble, more customizable.”

    Listen at 3:08

  6. Nathaniel Whittemoreon NVIDIA acquisition of Hugging FaceNegative4:06

    NVIDIA ownership could threaten trust in Hugging Face

    “An NVIDIA acquisition could threaten that trust”

    Listen at 4:06

  7. Nathaniel Whittemoreon NVIDIA acquisition of Hugging FacePositive4:42

    The Hugging Face acquisition is receiving unusually high optimism

    “there is a lot more optimism around this particular acquisition than you normally see.”

    Listen at 4:42

  8. Nathaniel Whittemoreon Plural AIPositive5:16

    Plural AI produces a better world than concentrated AI

    “The world is better with plural AI.”

    Listen at 5:16

  9. NVIDIA’s free cash flow fell 50% year over year

    “Free cash flow fell by 50% compared to a year ago”

    Listen at 5:56

  10. Nathaniel Whittemoreon Amazon purchase of NVIDIA chipsPositive6:54

    Amazon’s chip order will keep NVIDIA demand solid for years

    “this is a huge additional order that will keep demand solid for years into the future.”

    Listen at 6:54

  11. NVIDIA stock rose 4.8% after hours

    “the stock up by 4.8% in after-hours trading.”

    Listen at 7:02

  12. NVIDIA’s revenue-sharing financing arrangement could attract antitrust scrutiny

    “There is a belief that the arrangement could attract scrutiny for antitrust”

    Listen at 7:38

  13. Nathaniel Whittemoreon AI tools in traditional softwarePositive8:27

    AI adoption within traditional software products increases revenue

    “adoption of AI tools within traditional software products is leading to a revenue boost.”

    Listen at 8:27

  14. Nathaniel Whittemoreon Salesforce AI revenueNeutral8:47

    AI revenue remains a modest share of Salesforce revenue.

    “AI revenue is still a relatively modest part of the picture.”

    Listen at 8:47

  15. Salesforce stock rose 22% on Thursday

    “The stock was up a staggering 22% on Thursday.”

    Listen at 8:54

  16. Nathaniel Whittemoreon AI industry self-regulatory bodyNeutral11:36

    The proposed AI self-regulator would be modeled on FINRA.

    “The new body would be modeled on FINRA”

    Listen at 11:36

  17. Nathaniel Whittemoreon AI industry self-regulatory bodyNeutral11:41

    A proposed AI self-regulator would require labs to share security information and test models

    “this approach would require AI labs to share notes on security issues, conduct safety testing for new models, and create regulations that govern the industry.”

    Listen at 11:41

  18. Hermes supports agents that learn to execute tasks more cheaply and effectively

    “self-improving agents that learn to execute tasks cheaper and more effectively”

    Listen at 24:44

  19. Claude agents can access Salesforce CRM data and workflows

    “Salesforce users to access their CRM using Claude agents.”

    Listen at 25:39

  20. Nathaniel Whittemoreon Enterprise agent infrastructureNegative27:49

    Enterprise agent infrastructure and governance will take another one to two years

    “it's hard to display software outright when the agentic guardrails, governance, and infra doesn't exist in the enterprise and won't for another one to two years.”

    Listen at 27:49

  21. Nathaniel Whittemoreon Native AI integrationsPositive28:10

    Native AI integrations should deliver value with less implementation effort

    “these native integrations should deliver customers the value they want from AI without as much of a lift.”

    Listen at 28:10

  22. Nathaniel Whittemoreon Enterprise softwarePositive28:15

    The current narrative makes enterprise software difficult to view bearishly

    “Hard to be bearish on enterprise software at present given the magnitude of this narrative development.”

    Listen at 28:15

  23. Nathaniel Whittemoreon Enterprise softwarePositive28:19

    Embedded data, workflows, expertise, and incumbency remain software moats.

    “Data, operationally embedded workflows, expertise, incumbency are still the moats for software”

    Listen at 28:19

  24. Venice surpassed $100 million in annual recurring revenue

    “Venice is a privacy-focused AI alternative that recently crossed the threshold of 100 million in ARR”

    Listen at 33:41

Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.

Weekly email

State of the Pod, every Monday

The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.

Listen to the full episode and explore every guest, topic, and moment on PodLume.