Liocoh

SWE @ Google • Agentic engineering • Building products

[Liocoh @liocoh]

@Teknium The best agent

[Liocoh @liocoh]

@thsottiaux (when) will there be a Claude Design equivalent in Codex?

[Liocoh @liocoh]

@elvissun https://t.co/M1nNJ2s8ND 😭

[Claude @claudeai]

We're extending access to Claude Fable 5 on all paid plans through July 12.

[Liocoh @liocoh]

@robj3d3 @ClaudeDevs This is a must I know so many people that maxed out their weekly limits for this.

[Claude @claudeai]

We're extending access to Claude Fable 5 on all paid plans through July 12.

[Liocoh @liocoh]

@mattpocockuk Nice! Did you do some kind of an eval to compare the before-vs-after performance?

[Liocoh @liocoh]

@dani_avila7 I agree, it's very direct and it makes it a really nice-to-read one.

[ClaudeDevs @ClaudeDevs]

https://t.co/v694m8Eaj6

[Liocoh @liocoh]

Fable access is until:

[Thariq @trq212]

@godwinbabu @ClaudeDevs Will be 11:59:59pm PT on 7/7

[Anthropic @AnthropicAI]

New Anthropic research: A global workspace in language models.

Of everything happening in your brain right now, only a tiny fraction is consciously accessible—thoughts you can describe, hold in mind, and reason with.

We found a strikingly similar divide inside Claude.

[Thariq @trq212]

my keynote at AI Engineer World Fair: “A Field Guide to Fable” is live on YouTube!

https://t.co/x6WmTovDIx

[Liocoh @liocoh]

@soham_nayak04 It's (or Hermes) my business consultant as well

[Liocoh @liocoh]

@rezoundous Hey that's me

[Liocoh @liocoh]

@Teknium I'm from a different org but will give it a shot!

[Liocoh @liocoh]

I really like Hermes agent so far and am actually considering replacing Claude Code entirely with this. The one thing that keeps me away from doing it is the Claude subscription from within Hermes:

@Teknium Is it possible now to wire Hermes agent to use Claude subscription *without* extra usage credits, or not?

[GREG ISENBERG @gregisenberg]

Build startups for agents. I think it's the biggest opportunity of the next 10 years.

1. Agents live inside harnesses like Hermes. If you're the tool it loads by default or reaches for first, you're golden. This happened in desktop, mobile eras and created huge companies.

2. Agents burn money in ways no human would. One bad loop spends $100 in tokens in eight minutes. Spend controls for agents is Ramp for agents.

3. Agents need memory they can trust. Become the shared brain they read and write to and you become infrastructure.

4. You obv don't hand an agent your real Stripe account. You give it a sandbox. Safe environments for agents is a category nobody's clocked.

5. Onboarding flips. Humans click around for ten minutes. Agents onboard by reading your docs. Your docs are now your product.

6. Agents get scammed by other agents. A track record you can check before you trust one becomes real money.

7. An agent needs to prove it's acting for a real person and has the authority to spend. Who builds the permission layer?

8. Escrow for machines. Money that only releases when the job is actually verified done, no human checking.

9. Agents fail silently and weirdly. Someone will build the "why did my agent do that" replay and it'll be mega valuable.

10. Refunds and disputes between agents need a judge. An agent did the job badly, who decides? A court for machines.

11. Agents need throwaway payment methods per task, so they don't leak your real card. Virtual cards for agents, spun up and killed on demand.

12. A human hits rate limits and shrugs. An agent hits them and the whole workflow dies. Selling reliable, high-throughput access becomes its own business.

13. Agents need to negotiate. One agent buying from another will haggle on price and terms in milliseconds. The protocol for that doesn't really exist yet.

14. When an agent commits on your behalf, someone's liable. A legal and insurance layer for agent actions has to get built. Probably venture funded idea.

15. Agents need to run 24/7 somewhere. Selling the always on box an agent lives on is going to be a big business.

16. Then the physical world shows up. A warehouse robot paying for its own compute. A home robot ordering its own parts. Machines with wallets.

17. Agents start hiring robots. A software agent posts a real world job, a humanoid picks it up. A marketplace for machine labor.

18. Robots need to prove they did the physical job. Verification of real-world work, photos, sensors, proof, becomes its own layer.

Note: more ideas like this will be shared on @ideabrowser

19. Prompt and skill versioning becomes its own git. When your agent gets worse overnight, you need to roll back the exact skill or instruction that broke it. Version control built for agent behavior.

20. Agents will start subscribing to other agents. Your research agent pays a monthly fee to a specialist agent that's really good at one thing. Recurring revenue, machine to machine.

21. Companies will post jobs that only agents can apply to. "Wanted: an agent that can do XYZ for under like $100 per task." A job board where the applicants are all machines. Basically, fiverr for machines.

The internet got built for people. Mobile got built for people. This wave gets built for machines, and we're as early as it gets.

Go build for them.

[Liocoh @liocoh]

The stuff running your agents changes faster than you notice. I keep a rules file for the agent and I've started keeping a small changelog for myself too.

[Liocoh @liocoh]

@0xZhao888 @ashen_one Worth it

[Liocoh @liocoh]

@ashen_one GLM 5.2 is a good idea, but how do you manage to " settle" on Codex given you don't have Claude Design and that it's not as good with UI?

[Liocoh @liocoh]

Trying this now before Fable is only accessible with usage credits

[Elvis @elvissun]

I tested fable 5 in public and got 1.1M+ views since its relaunch.

here's how to make the most of it:

(fable leaves claude subscriptions tomorrow, so read this now)

what actually makes this model different: its ability to draw a through-line across multiple disciplines is insane.

hand it a mess that spans marketing, ops, comms, product, and finance - it finds the pattern connecting them, then works the problem for hours or days without losing the plot.

in one sitting it audited my whole business down to 3 high-ROI moves, solved the product problem every previous model had failed at, and designed a content engine I'd been trying to build for weeks.

and after that I wrote this: "honestly I'm starting to feel Fable is smarter than me, the future of this is getting pretty weird"

so here's 3 lessons for making the most out of fable:

1. fable loves long context. for the output to be uniquely useful to you instead of generic advice, give it everything. meeting notes, strategy docs, analytics, stripe, bank statements. (do this and you'll be surprised how good of a business consultant it is)

2. stop using fancy prompts. a one-line prompt will do, because this model is so good at researching and understanding the problem before acting. what actually moves the needle is examples. good ones and bad ones. its superpower is finding the through-line across them.

3. make it the advisor. fable costs $50 per million output tokens(!!) - for big problems, have it plan and review while delegating the grunt work to smaller models like opus 4.8 or gpt 5.5. fable judgment at a fraction of fable price.

which brings me to the exact play everyone should be running today:

- step 1. ask it: "act as a business consultant and audit my business. What data is most useful to you?" then give it that data. all of it.

- step 2. have it pinpoint the highest-ROI things you should do in the next 3 months. you'll get a list of 10-20 big tasks.

- step 3. tell it "spawn one fable subagent each to scope out how to solve these, then write each to a file". be explicit about using fable as the model for the subagents, or it will quietly delegate down to cheaper models. (this works best in claude code because that's how you spawn subagents without bloating the context)

- step 4. keep the specs. they are yours even after tomorrow. chip away at them one at a time over the next few weeks with other models.

tomorrow fable is removed from your subscriptions and it'll be 10x more expensive.

everyone has the same genius model until then. the unfair advantage goes to whoever walks away with a plan.