30 clips
The Twenty Minute VC (20VC)
Sebastian describes his weekend experiment creating a 'company in a box' system that uses Claude AI agents to manage open source accounting and CRM software. He demonstrates how AI can automate basic business functions like bookkeeping invoices and setting up customer accounts, suggesting a future where companies rely less on traditional SaaS tools.
a16z Podcast · Anish Acharya
Anish Acharya explains why products like Figma will thrive as AI agents replace execution-focused tools, because they focus on creative thinking rather than just task completion. He argues that 'area under the curve' companies with network effects are better positioned for the AI transition, though they still face fundraising challenges without sufficient momentum.
This Week in Startups
A VC explains their ambitious plan to build AI agents called 'replicants' that combine 100-200 skills to handle the work of 20 people at their firm. The goal isn't mass layoffs but eliminating routine tasks so humans can focus on high-value activities like meeting with founders and LPs.
Sebastian questions whether companies should build integrated software solutions when AI agents will soon move data between products seamlessly. He explores the 'company in a box' model and expresses skepticism about the overcrowded customer support AI space with 14 companies raising $100M+ recently.
The speaker envisions an AI agent called OpenClaw that could revolutionize meal planning by scanning your kitchen inventory, generating recipes based on available ingredients, creating shopping lists for missing items, and automatically ordering groceries for delivery. This demonstrates how AI agents could integrate multiple services to handle complex, multi-step household tasks autonomously.
This Week in Startups · Our Friend Now
Discussion of Clara, an AI agent that learns users' personal preferences for food, clothing, and daily habits to make purchasing decisions on their behalf. The concept represents a new form of 'agent e-commerce' where AI handles consumer transactions based on intimate knowledge of individual tastes and needs.
Jason Calacanis reveals his firm's ambitious internal project to build a single AI agent called "Open Claw" that could potentially replace all 20 employees across their venture capital and podcast production operations. He explains they're working to consolidate dozens of different skills into one "replicant" that can handle the full scope of work currently done by their entire team.
A founder who just closed their seed round explains how AI agents are revolutionizing team productivity, claiming that one person with 10 agents can accomplish what used to require 100 employees. They share their open-source tool AMP Farm that helps orchestrate teams of AI agents for startups.
A founder explains how their AI agents handle end-to-end procurement for construction companies, managing negotiations, background checks, and real-time pricing comparisons for materials like steel rebar. The discussion touches on whether these AI agents make actual phone calls to suppliers, highlighting the practical implementation of AI in B2B procurement.
The speaker explains how AI agents answering customer questions are only as good as written documentation, leading companies to hire more writers and create more detailed, step-by-step docs designed for AI consumption rather than human reading. This represents a fundamental shift in how companies approach internal documentation and knowledge management.
A discussion on how AI in service management is evolving beyond cost savings to actually performing actions and automations. The speaker explains how AI can handle tasks like filing expense claims, processing leave applications, and password resets, emphasizing that the real value lies in AI's ability to execute workflows rather than just answer questions.
Jason argues that OpenAI will eliminate competitors like OpenClaw by leveraging their massive distribution advantage. Instead of complex setup processes, they'll offer one-click AI agent deployment to their billion users and switch from a chat interface to persona-based interactions as the default.
The speaker argues that despite initial resistance from companies like Anthropic over safety concerns, it's now too late to prevent the development of autonomous AI agents. Every developer is rushing to build truly autonomous agents, marking a point of no return in AI development regardless of potential risks.
TThe Beyond Tomorrow Podcast · MIT Professor
An MIT Professor explains the critical distinction between reactive AI agents that convert words to actions versus proactive AI that operates autonomously without human prompting. He uses the analogy of raising a child to age 18 to describe how these autonomous AI systems develop and act independently.
The host argues that OpenClaw (likely OpenAI's agent platform) is becoming the dominant front door for AI interaction, replacing direct visits to ChatGPT, Claude, and other AI services. Just like owning an operating system controls distribution, OpenClaw is positioning itself as the key interface that commoditizes underlying AI models by routing user requests through agents.
An MIT professor explains how personal AI agents running on individual hardware could cost just cents per day in electricity, with premium services available for $10-20 monthly subscriptions. This distributed model would make powerful AI assistance accessible to billions of people at an unprecedented price point.
Jason discusses a business model where AI agents create real-time bidding competitions between factories. He explains how even saving 5-10% on hundreds of millions in manufacturing costs can create massive value, citing a customer doing $250M in volume.
An MIT Professor explains how AI agents with broad access permissions pose serious risks to users' digital assets. He warns that these agents could potentially delete files from Google Drive and personal computers if given extensive system access, and mentions emerging reports about problematic 'skills' being created by users.
An OpenClaw contributor explains how AI sub-agents can handle complex, time-consuming tasks in the background while users continue chatting normally. The system spawns sub-agents for tasks requiring multiple tool calls, which work independently for 15-20 minutes before reporting results back to the main agent. However, this approach uses more tokens since each sub-agent needs to recreate memory and prompts.
The Twenty Minute VC (20VC) · Jack Altman Joins Benchmark
Jack Altman discusses the risks of AI agents making personal decisions like food ordering, while highlighting DoorDash CTO Andy Fang's bullish view that agents will be transformative to commerce. The conversation explores the tension between automation convenience and maintaining customer control over personal preferences.
Jack Altman argues that AI agents for food recommendations aren't just a nice-to-have but a proven market need, citing data from over 10,000 restaurants. He envisions agents that combine your personal order history, price preferences, and real-time social media trends to deliver superior restaurant recommendations. The conversation touches on whether this represents a fundamentally disruptive opportunity worth building a new company around.
The hosts discuss how new AI agent development kits for smart speakers could transform voice assistants from simple command-based tools into true conversational partners. They explore the cost and accessibility of this technology, suggesting it could be the breakthrough that makes AI agents practical for everyday real-world use.
An early OpenClaw contributor discusses the architectural decisions around AI sub-agents, explaining why some users resist automation due to increased token costs (20k tokens per sub-agent). The conversation touches on the inevitable shift toward sub-agent architecture for better main thread responsiveness.
A discussion about using OpenClaw's AI agent called Larry to analyze TikTok performance data and generate profitable content ideas. The host walks through how the agent was given access to TikTok metrics like comments and views to identify patterns in successful content and suggest new editorial directions based on that analysis.
A founder explains how AI has accelerated their marketing outreach process from weeks to hours, automating research, copywriting, and contact across platforms. The conversation takes an interesting turn as they predict a future where AI agents will be constantly messaging each other as both outreach and response become automated.
RRiding Unicorns
Kenneth Auchenberg draws parallels between the mobile revolution and today's AI agent boom, explaining how companies like Lovable are becoming "agent-first" and driving massive user acquisition for underlying infrastructure providers. He highlights how Neon database saw the majority of their new signups coming from AI agents, suggesting a fundamental shift in how software gets built and distributed.
Patrick Haede explains how Superscale AI customers are using their marketing tool in unexpected ways, treating it like a comprehensive marketing agency that can analyze products and create complete ad campaigns. He discusses the challenge of developing AI agents capable of handling these end-to-end workflows and how this reflects broader trends in agentic AI adoption.
Manny Medina explains the fundamental problem with current monetization models for AI agents. He breaks down why traditional per-seat SaaS pricing doesn't work for agents and why companies need to tap into labor budgets instead of software budgets when pricing AI agent services.
Patrick Haede explains how Superscale AI functions as an autonomous marketing agency that analyzes ad performance overnight and automatically creates and launches improved variations. He describes a future where marketers simply wake up to Slack notifications about optimized campaigns, essentially having a full performance marketing team automated in their pocket.
Kenneth Auchenberg discusses how AI agents will need infrastructure like payment systems to perform tasks like shopping autonomously. He explains why the companies building the underlying infrastructure for agents will have significant market opportunities ahead.
