Share

cover art for I asked Fable and Codex what to automate. They disagreed.

AI News & Strategy Daily with Nate B. Jones

I asked Fable and Codex what to automate. They disagreed.

•

For deeper playbooks and analysis: https://natesnewsletter.substack.com/p/let-ai-pick-what-to-automate


What's really happening when you stop telling an AI what to automate and ask it to discover the problem itself?

The common story is that AI agents need a tightly specified task — but the reality is that the strongest systems can inspect real work, identify recurring friction, and propose different high-leverage automations.


In this video, I share the inside scoop on giving Fable and Codex the same open brief and getting two very different answers.

  • Why picking the problem is becoming part of the agent's job
  • How Fable found a strategic editorial preflight opportunity
  • What Codex built to validate completed content handoffs
  • Where human judgment still matters
  • How to turn the method into a reusable automation-discovery skill

For operators, builders, and leaders, the shift is from asking which tool to use to asking which recurring problem is worth solving completely.

Subscribe for daily AI strategy and news.


Hosted on Acast. See acast.com/privacy for more information.

More episodes

View all episodes

  • AI Cyber Defense: 8 Audits to Keep Your Business Running

    13:22|
    For deeper playbooks and analysis: https://natesnewsletter.substack.com/What's really happening when attackers can move in minutes but defenders need weeks to put systems back together?The common story is that a week of outages points to one coordinated attack — but the reality is a mix of separate incidents and tightly connected systems that are difficult to repair.In this video, I share the inside scoop on why AI-assisted attacks are accelerating and why cyber defense remains the harder engineering problem.Why the Microsoft outage illustrates the risks of recovery itself.How ARTEX changes the amount of work attackers can automate.Why ConnectWise could revoke access quickly but needed weeks to restore it.What finding every affected Java installation requires.How browser isolation, better models and simulation can help defenders.For builders, operators and business leaders, understanding access, dependencies and recovery matters as much as finding the flaw. The practical challenge is keeping critical systems working while repairing years of accumulated complexity.Hosted on Acast. See acast.com/privacy for more information.
  • Your AI can write a thousand questions when you needed the 50 best answers. How to keep a long run pointed at what you want now, even as the goal shifts

    20:35|
    The full post, with prompts and worksheets for running longer AI work on your own and with a team: https://natesnewsletter.substack.com/p/long-running-ai-agents?utm_source=podcast&utm_medium=podcast&utm_campaign=long-running-ai-agents-launch&utm_content=show-notesWhat's really happening when an AI agent works all night?OpenAI and Anthropic report AI agents working for hundreds of hours and for days. The common story is that longer runs just need a smarter model, but the real question is whether your agent has the current decisions, examples, and instructions behind the work.In this episode, I share the inside scoop on long-running AI agents:Why my research run produced a thousand questions I didn't needHow treating project information like code keeps agents on current decisionsWhat shared skills and permissions change when many agents touch the same workWhere to start alone with one job you keep explainingAn overnight run can save you a day or cost you one, and the difference is the structure you give it.Subscribe for daily AI strategy and news.Hosted on Acast. See acast.com/privacy for more information.
  • AI Robotics and the Bitter Lesson: What to Watch in 2027

    12:19|
    What's really happening when a coding agent starts operating a robot?A general AI coding agent ran a real robot through 24 of 25 trials with no extra robot training. The common story is that better robots need ever more specialized robotics models, but the real question is how much of a robot's intelligence comes from making AI better at everything else.In this episode, I share the inside scoop on robotics and the Bitter Lesson:Why Sutton's Bitter Lesson matters for robots in 2027How Stanford's VIA lets a coding agent aim a real gripperWhat the wipe test shows about an agent checking its own workWhere speed and supervision still limit useful robot workGeneral models may carry a lot of robotics progress with them, but a robot only earns its place once it works fast enough to need little supervision.Subscribe for daily AI strategy and news.Hosted on Acast. See acast.com/privacy for more information.
  • AI Evals for Buying AI: Keep More of the Savings

    20:24|
    For deeper playbooks and analysis: https://natesnewsletter.substack.com/What's really happening when AI keeps improving but customers have less reason to pay for the upgrade?The common story is that more capable models automatically create more revenue — but the reality is that buyers can test what is good enough and move routine work to cheaper models.In this video, I share the inside scoop on who gets paid as AI becomes more useful and the market for intelligence becomes more demanding.Why token use can rise while the bill stays the same.How evals turn model selection into a purchasing decision.What Ramp's spending data reveals about adoption and pricing power.Where productivity gains go when customers expect lower prices.What stays valuable as models improve and technical work becomes easier.For business leaders, the decision is about the full cost of useful work: model spend, review, repairs, and the surrounding systems. Set a baseline on real assignments, then ask what value your business can retain as customers and competitors adapt.Subscribe for daily AI strategy and news.Hosted on Acast. See acast.com/privacy for more information.
  • Jensen Is Out. Here's What Changes for Nvidia and AI

    11:02|
    For deeper playbooks and analysis: https://natesnewsletter.substack.com/What's really happening to NVIDIA's position after Jensen Huang's resignation?The common story is that the product roadmap tells us everything — but the reality is that customer relationships, financing, and political influence also shape how AI infrastructure gets built.In this video, I share the inside scoop on NVIDIA's next deal, the cost of switching chip platforms, and the implications for your AI bill.Why replacing coordination and deal-making authority matters.How CUDA supports NVIDIA's competitive position.What Anthropic's AMD example means for software portability.How infrastructure financing and export rules affect customers.Why competition could lead to cheaper inference.For operators and builders, the stakes are practical: deployment schedules, access to compute, and the price of using AI.Subscribe for daily AI strategy and news.Hosted on Acast. See acast.com/privacy for more information.
  • Is Claude Fable 5.2 Worth the Cost? What to Use It For

    14:06|
    For deeper playbooks and analysis: https://natesnewsletter.substack.com/What's really happening when a stronger AI model still feels like an assistant?The common story is that more capability makes every task better — but the reality is that people still have to find problems where the improvement is worth the cost.In this video, I share the inside scoop on my early access to Fable 5.2 and the e-ink display project it helped finish after earlier attempts fell short.How the model inferred my design intent from failed attempts.Why a proactive handoff made the build easier.What task-level saturation means for everyday work.How to identify projects with enough value to justify expensive AI.Why judging solutions and communicating intent remain essential skills.For operators and builders, the useful question is where better AI can deliver a result worth paying for. Meanwhile, labs have a clear use for that intelligence: building the next model.Subscribe for daily AI strategy and news.Hosted on Acast. See acast.com/privacy for more information.
  • Jensen Huang Scandal Explained: Who Gets Nvidia's AI Chips

    20:58|
    For deeper playbooks and analysis: https://natesnewsletter.substack.com/What's really happening with Jensen Huang, chip access, and the pace of AI?The common story is that this is a CEO scandal — but the reality is that scarce computing power connects the NVIDIA investigation, AI-assisted research, and businesses' growing cyber-defense needs.In this video, I share the inside scoop on:Why Chinese buyers are waiting for NVIDIA chips.How overseas computing access complicates the story.What AI-assisted research could mean for the next generation of models.How an internal AI tool figured in the Binance attack.For operators and builders, chip availability shapes both what AI can do and who can use it. Nate distinguishes unresolved allegations from the reporting and considers what the next year could mean for business planning.Subscribe for daily AI strategy and news.Hosted on Acast. See acast.com/privacy for more information.
  • AI Data Center Backlash: Why It Helps Big Tech

    19:31|
    For deeper playbooks and analysis: https://natesnewsletter.substack.com/What's really happening when Americans use more AI but reject new data centers?The common story is that the backlash will stop the AI buildout — but the reality is that demand keeps growing, and constraints change who can serve it and where investment goes.In this video, I share the inside scoop on the consequences beyond the ballot box:Why delays hit smaller developers hardest.How existing capacity can strengthen large technology companies.Where investment, jobs, and computing power may move next.What makes orbital compute a speculative alternative.Why the industry needs to connect AI with empowerment and opportunity.For operators, builders, and people using AI at work, construction headlines are only one signal. Watch what companies do with the next project and how they secure the capacity they need.Subscribe for daily AI strategy and news.Hosted on Acast. See acast.com/privacy for more information.
  • OpenAI's Git Backend for AI Agents: Vijaye Raji Interview

    46:27|
    For deeper playbooks and analysis: https://natesnewsletter.substack.com/What changes when agents can write code faster than teams can review, test, and ship it?Nate Jones talks with Vijaye Raji, CTO of Applications at OpenAI, about how agents are changing work inside the company—from engineering infrastructure to finance and product development.Why code review, testing, and repositories become the next bottlenecks.How to decide when a human should review an agent's response.Why finance teams are sharing purpose-built sites instead of spreadsheets.What founders should ask when a more capable model arrives.How small teams use models to build and iterate faster.For builders and leaders, the conversation offers concrete ways to rethink workflows around improving models while keeping human judgment in the right places.Subscribe for daily AI strategy and news.Hosted on Acast. See acast.com/privacy for more information.