on YouTube 14 nov 2026
field notes · new episode
i ran my viral Fable 5.2 conductor for 40 hours - here's what it actually built
from the episode: i ran my viral Fable 5.2 conductor for 40 hours - here's what it actually built | LAB 0041
Lenny's AI Builders - LAB 0041
the conductor paragraph, the four steps to run it, and the honest numbers from my 40-hour run.
i ran my viral Fable 5.2 conductor for 40 hours on my real app.
one conductor, 73 worker threads, 25 pull requests merged. and my weekly Fable limit was gone in 13 hours.
so here's what it actually built, and the one paragraph that runs yours.
why it matters
this is the new meta. you use Fable 5.2 as the conductor and Opus 5.5 as the worker threads.
Fable 5.2 is twice as costly as Opus 5.5. so Fable does the thinking and Opus does all of the work.
it's not perfect and it's not god level. but it makes a huge difference to how much you have to hold in your head while you run lots of agents at once.
Fable directs the show now. Opus does the work.
what you get
- the paragraph. the same one i read out in the episode. you can copy it below.
- the four steps. how to set it up in Claude Code.
- the honest numbers. what my run built, and the mess most people won't show you.
the real example: my run
i ran one conductor on Eddy Studio, my Mac app, for 40 hours. this was in T3 Code, with my full conductor skill. the paragraph below is the short version of it.
Fable 5.2 plans, then gives each task to a worker thread. when the worker is done, Fable checks the work.
25 pull requests got merged. three of them: an update button, click a scene and the video jumps to it, and a feedback card that opens whole.
the stats: Fable sent 259 messages to 73 threads. 150 of those went to Opus 5.5. that's 58%.
and here's the stuff that wasn't so clean:
- 19 turns failed. 17 of them were usage limit errors.
- it stopped to ask me 20 things.
- one fix got parked after five outside checks failed.
- it opened one duplicate thread by mistake and caught it before it wrote anything.
i hit my weekly Fable limit 12 hours and 55 minutes after my first Fable 5.2 prompt. about 158 of the 268 turns came after that notice.
how i counted: a pull request counts when it links to a thread the conductor ran. i didn't check who wrote every line. 8 of the 25 were opened before the conductor started. the 13 hours covers all my Fable use, not only the conductor.
try it
- open Claude Code.
-
type
/modeland pick Fable 5.2. that's your conductor. - set your sub-agents to Opus 5.5. those are your workers.
- give it a massive job and add the paragraph below.
start with three workers, not 73. 73 threads is where mine ended up after 40 hours. it won't spin up 73 all at once.
copy this prompt
add this paragraph to your job:
you're the conductor. you do the thinking: the plan, the architecture, the strategy.
split the job into tasks. give each one to its own Opus 5.5 worker.
nudge every worker until its task is done.
while they work, sit idle. don't do their work for them.
when a worker reports back, check it, then give out the next task.
handle anything that doesn't need me. only stop when you need a decision from me.that paragraph is the short version. the cleaned-up skill that ran my 40-hour build is in my skills repo, with the rest of my skills.
check it against the paragraph
my 40-hour run used my full conductor skill, not this paragraph. so check yours against the paragraph, line by line:
- does it give each task to its own worker?
- does it sit idle while they work?
- does it check each report before it gives out the next task?
- does it only stop when it needs a decision from you?
limits and dates
- one run on one app. this is my run on Eddy Studio, not a benchmark.
- merged, not shipped. the 25 are merged pull requests. i'm not claiming any of it had reached testers when i filmed.
- not every thread was Opus. 58% of the conductor's messages went to Opus threads.
- Fable has its own weekly limit. mine was gone in 13 hours, and that covers all my Fable use.
- the price when i filmed: Fable 5.2 is $8 in and $40 out per million tokens. Opus 5.5 is $4 in and $20 out.
the rest of this LAB
- Gemini 4 Carbon is out. it's Google's strongest model. Google's own line is that Argon is still the one for most software and knowledge work, and Carbon steps in for your hardest tasks. Cursor has it at the top of CursorBench at 60.4%, with Fable 5.2 at 60% on the same board. Artificial Analysis still has Fable ahead of Carbon, 65 to 60. in LAB 0032 i said i'd make a short admitting i was wrong if Carbon was actually better than Opus 5.5. there is no short yet. my take: it puts Google on the board, but i seriously doubt it gets you to swap over from Opus 5.5 and Fable 5.2.
- Claude gets Sheets, Data and Whiteboards. Whiteboards and Sheets are in beta on paid plans and Data is fully out. Sheets is a spreadsheet inside Claude with 144 functions and up to 100,000 cells. no pivot tables, no macros and no live data from your connected apps yet. Data is the one that used to be called dashboards: ask a question about your numbers and it builds a dashboard, a written story or a short narrated video. Whiteboards is the one i'm most excited about. you talk, type or sketch, and Claude groups it into a diagram or a flow, finds the gaps and turns the board into a project plan. Miro is $8 a member a month on its starter plan, billed yearly, and this comes with the Claude plan you already pay for. Claude's whiteboard has a send to Miro button, so the official story is "works with Miro". tonight: turn on dictation in Claude and brain dump a project you've been putting off. my take: i'd hate to be Miro right now.
- MCP Apps. in public preview for developers, with Canva, Notion, Asana, HubSpot and Vercel among the launch partners. it's a shared open standard, so ChatGPT supports it as well.
- the free full reset. Anthropic gave every Pro, Max, Ultra and Team user one free full reset across all models and all limits. it's in settings, then usage, then reset for free. you can't undo it, and it runs out on December 10 at 2 pm Pacific. that's 6 am Friday 11 December in Perth. Anthropic's reason, in their words: they fixed a few small issues that used more limits than intended. my take: don't press it today because you don't need to. save it for the day your agents eat your whole week. i pressed mine on camera. i had a beta to ship.
- Businesses in ChatGPT. it's in beta. you claim your business, verify it and get a blue check in front of what OpenAI says is 1.2 billion weekly users. business agents are for online stores and services in the US only for now. OpenAI says claiming doesn't guarantee higher placement or sales. i read the terms, and there are three things you should know. one: your data is used for developing and improving their product. two: you agree, irrevocably and perpetually, not to bring claims over that data, or over what's published on websites you own, unless they've broken the terms. three: they can list your business whether you claim it or not, and claiming doesn't give you exclusive control. my take: claim it if the listing is worth more to you than those rights.
- Anthropic's IPO. the roadshow has started. the filing says 825 million shares at $70 to $78 each, and the ticker is ANT. that's a $58 to $64 billion sale, and it values the company at roughly $916 billion to just over a trillion. CNBC reports it prices on November 18. look at the week they picked: Fable 5.2, Sheets, Whiteboards, a free reset for every paid plan and this roadshow. my take: the aura around Anthropic right now is not a coincidence. it's a launch calendar built for Wall Street, and builders got the best week of shipping we've had out of it.
- the Astra rumour. the rumour is OpenAI is holding GPT-6.2 Astra until just before the IPO. nothing ties Astra to the IPO week except one account's post. it's a rumour.
sources
- my conductor post on X
- the same post on Threads
- my skills repo
- Google AI on Gemini 4 Carbon
- Cursor on Carbon and CursorBench
- Claude on Whiteboards, Sheets and Data
- Claude Sheets help page
- ClaudeDevs on MCP Apps
- ClaudeDevs on the free full reset
- OpenAI on Businesses in ChatGPT
- the Businesses in ChatGPT terms
- Anthropic's IPO roadshow post
- the S-1/A filing
next step
watch LAB 0041 on YouTube - the practical starts at 12:59.
want the next one? get the next field notes. optional - these notes stay free to read.