Jimmie
@jimmiethongkham.bsky.social
I can't tell if the people inside AI labs *actually* believe AI can end the world. Or they are fear-mongering to encourage regulation to kill competition from smaller players.
The incentive muddies the picture.
The incentive muddies the picture.
September 15, 2026 at 4:09 PM
I can't tell if the people inside AI labs *actually* believe AI can end the world. Or they are fear-mongering to encourage regulation to kill competition from smaller players.
The incentive muddies the picture.
The incentive muddies the picture.
I built a fantasy football start/sit tool:
- import your sleeper roster (just your username and league URL)
- ranks your players based on CBS and Yahoo rankings
- suggests the optimal lineup
https://www.jimmiethongkham.com/ff/start-sit
- import your sleeper roster (just your username and league URL)
- ranks your players based on CBS and Yahoo rankings
- suggests the optimal lineup
https://www.jimmiethongkham.com/ff/start-sit
September 14, 2026 at 4:09 PM
I built a fantasy football start/sit tool:
- import your sleeper roster (just your username and league URL)
- ranks your players based on CBS and Yahoo rankings
- suggests the optimal lineup
https://www.jimmiethongkham.com/ff/start-sit
- import your sleeper roster (just your username and league URL)
- ranks your players based on CBS and Yahoo rankings
- suggests the optimal lineup
https://www.jimmiethongkham.com/ff/start-sit
I find Fable consumes tokens like crazy.
So I always ask it to pass the build to an Opus subagent.
Since then I haven't hit my 5 hour limit.
So I always ask it to pass the build to an Opus subagent.
Since then I haven't hit my 5 hour limit.
September 14, 2026 at 3:05 PM
I find Fable consumes tokens like crazy.
So I always ask it to pass the build to an Opus subagent.
Since then I haven't hit my 5 hour limit.
So I always ask it to pass the build to an Opus subagent.
Since then I haven't hit my 5 hour limit.
Downgraded my $200 Claude sub to $100/mo last week.
And bought the $100/mo ChatGPT Pro
Why? Fable consumes way too much tokens to be usable. And the way it writes (despite updating claude.md) makes it difficult to read.
So far I feel Astra performs better than Fable.
And bought the $100/mo ChatGPT Pro
Why? Fable consumes way too much tokens to be usable. And the way it writes (despite updating claude.md) makes it difficult to read.
So far I feel Astra performs better than Fable.
September 12, 2026 at 4:10 PM
Downgraded my $200 Claude sub to $100/mo last week.
And bought the $100/mo ChatGPT Pro
Why? Fable consumes way too much tokens to be usable. And the way it writes (despite updating claude.md) makes it difficult to read.
So far I feel Astra performs better than Fable.
And bought the $100/mo ChatGPT Pro
Why? Fable consumes way too much tokens to be usable. And the way it writes (despite updating claude.md) makes it difficult to read.
So far I feel Astra performs better than Fable.
So many posts around how to grow, tactics, playbooks, etc
But not enough about managing your own psychology.
How to keep going? How to constantly overcome obstacles?
But not enough about managing your own psychology.
How to keep going? How to constantly overcome obstacles?
September 11, 2026 at 8:19 PM
So many posts around how to grow, tactics, playbooks, etc
But not enough about managing your own psychology.
How to keep going? How to constantly overcome obstacles?
But not enough about managing your own psychology.
How to keep going? How to constantly overcome obstacles?
Moonshot and Fireworks sell Kimi K2.7 at $0.95 per million tokens.
But neither can charge more because it's a commodity.
This is called the Bertrand competition: two sellers selling an identical product so price falls to cost.
But neither can charge more because it's a commodity.
This is called the Bertrand competition: two sellers selling an identical product so price falls to cost.
September 10, 2026 at 4:10 PM
Moonshot and Fireworks sell Kimi K2.7 at $0.95 per million tokens.
But neither can charge more because it's a commodity.
This is called the Bertrand competition: two sellers selling an identical product so price falls to cost.
But neither can charge more because it's a commodity.
This is called the Bertrand competition: two sellers selling an identical product so price falls to cost.
I keep seeing "if everyone can build" or "building is easy".
This is wrong way to think about it.
The vast majority of folks (outside of the X bubble) will not build. They don't care enough to do it.
This is wrong way to think about it.
The vast majority of folks (outside of the X bubble) will not build. They don't care enough to do it.
September 9, 2026 at 4:09 PM
I keep seeing "if everyone can build" or "building is easy".
This is wrong way to think about it.
The vast majority of folks (outside of the X bubble) will not build. They don't care enough to do it.
This is wrong way to think about it.
The vast majority of folks (outside of the X bubble) will not build. They don't care enough to do it.
What is the opposite of AI slop?
We need a word for when AI generates something awesome.
I vote AI slap 😄
We need a word for when AI generates something awesome.
I vote AI slap 😄
September 8, 2026 at 3:11 PM
What is the opposite of AI slop?
We need a word for when AI generates something awesome.
I vote AI slap 😄
We need a word for when AI generates something awesome.
I vote AI slap 😄
Pricing based on AI usage (tokens) won't work in mainstream apps.
The outcome should be priced at a premium. E.g. book your next trip but charge a 5% premium for the convenience (like Instacart)
The biggest apps will win on small margins on big volume.
The outcome should be priced at a premium. E.g. book your next trip but charge a 5% premium for the convenience (like Instacart)
The biggest apps will win on small margins on big volume.
September 6, 2026 at 4:54 PM
Pricing based on AI usage (tokens) won't work in mainstream apps.
The outcome should be priced at a premium. E.g. book your next trip but charge a 5% premium for the convenience (like Instacart)
The biggest apps will win on small margins on big volume.
The outcome should be priced at a premium. E.g. book your next trip but charge a 5% premium for the convenience (like Instacart)
The biggest apps will win on small margins on big volume.
An AI-driven pullback in the stock market is inevitable.
Expectations have risen faster than reality.
People don't adapt to new technologies quickly. (X is a bubble)
Expectations have risen faster than reality.
People don't adapt to new technologies quickly. (X is a bubble)
September 5, 2026 at 4:54 PM
An AI-driven pullback in the stock market is inevitable.
Expectations have risen faster than reality.
People don't adapt to new technologies quickly. (X is a bubble)
Expectations have risen faster than reality.
People don't adapt to new technologies quickly. (X is a bubble)
Fable 5.1 consumes tokens super quickly. Hard to use it.
September 4, 2026 at 8:09 PM
Fable 5.1 consumes tokens super quickly. Hard to use it.
I tried a couple of text to video models (Kling and Minimax) and they have trouble following the prompt.
Can't get the colors right. Wrong people speaking. Even after multiple edits to the prompt.
I feel like all of the "one shot" posts on here are BS.
Can't get the colors right. Wrong people speaking. Even after multiple edits to the prompt.
I feel like all of the "one shot" posts on here are BS.
September 4, 2026 at 4:10 PM
I tried a couple of text to video models (Kling and Minimax) and they have trouble following the prompt.
Can't get the colors right. Wrong people speaking. Even after multiple edits to the prompt.
I feel like all of the "one shot" posts on here are BS.
Can't get the colors right. Wrong people speaking. Even after multiple edits to the prompt.
I feel like all of the "one shot" posts on here are BS.
2 to 3 experiments out of 10 win at big companies.
That's what I tell myself to keep going.
That's what I tell myself to keep going.
September 3, 2026 at 4:09 PM
2 to 3 experiments out of 10 win at big companies.
That's what I tell myself to keep going.
That's what I tell myself to keep going.
My entire stack as a non-developer:
1) Vercel
2) Github
3) Claude + a bunch of skills
4) Shadcn
Simple is how i like it.
1) Vercel
2) Github
3) Claude + a bunch of skills
4) Shadcn
Simple is how i like it.
September 2, 2026 at 4:09 PM
My entire stack as a non-developer:
1) Vercel
2) Github
3) Claude + a bunch of skills
4) Shadcn
Simple is how i like it.
1) Vercel
2) Github
3) Claude + a bunch of skills
4) Shadcn
Simple is how i like it.
Any indie hackers running AI evals? i don't see it on my feed.
I'm building mine now but curious to see what others have done.
I'm building mine now but curious to see what others have done.
September 1, 2026 at 5:12 PM
Any indie hackers running AI evals? i don't see it on my feed.
I'm building mine now but curious to see what others have done.
I'm building mine now but curious to see what others have done.
The 3 ways you can tell a reddit thread is fake (astroturf for AEO):
1) keyword randomly inserted into the story when its not relevant at all
2) a question in the comment like "what tool do you use?" as the set up
3) post is edited days later to add a link to the product
1) keyword randomly inserted into the story when its not relevant at all
2) a question in the comment like "what tool do you use?" as the set up
3) post is edited days later to add a link to the product
August 31, 2026 at 4:09 PM
The 3 ways you can tell a reddit thread is fake (astroturf for AEO):
1) keyword randomly inserted into the story when its not relevant at all
2) a question in the comment like "what tool do you use?" as the set up
3) post is edited days later to add a link to the product
1) keyword randomly inserted into the story when its not relevant at all
2) a question in the comment like "what tool do you use?" as the set up
3) post is edited days later to add a link to the product
The astroturfing on reddit is getting sophisticated.
Anon accounts will post with an AEO keyword but no product mention to get past mods.
Then days later, it will edit the post with a link to the product.
Crazy.
Anon accounts will post with an AEO keyword but no product mention to get past mods.
Then days later, it will edit the post with a link to the product.
Crazy.
August 30, 2026 at 8:39 PM
The astroturfing on reddit is getting sophisticated.
Anon accounts will post with an AEO keyword but no product mention to get past mods.
Then days later, it will edit the post with a link to the product.
Crazy.
Anon accounts will post with an AEO keyword but no product mention to get past mods.
Then days later, it will edit the post with a link to the product.
Crazy.
I got a dev board. What should I build first?
https://www.waveshare.com/esp32-s3-touch-amoled-1.8.htm?sku=29957
https://www.waveshare.com/esp32-s3-touch-amoled-1.8.htm?sku=29957
August 29, 2026 at 4:09 PM
I got a dev board. What should I build first?
https://www.waveshare.com/esp32-s3-touch-amoled-1.8.htm?sku=29957
https://www.waveshare.com/esp32-s3-touch-amoled-1.8.htm?sku=29957
I'm not a developer. Here's what my AI stack costs:
- Claude: $200/mo
- ChatGPT: $20/mo
- https://Fal.ai: $25 (credits)
- xAI: $5 (pay per use)
Call it ~$250/mo but plenty for what I need.
- Claude: $200/mo
- ChatGPT: $20/mo
- https://Fal.ai: $25 (credits)
- xAI: $5 (pay per use)
Call it ~$250/mo but plenty for what I need.
August 28, 2026 at 4:09 PM
I'm not a developer. Here's what my AI stack costs:
- Claude: $200/mo
- ChatGPT: $20/mo
- https://Fal.ai: $25 (credits)
- xAI: $5 (pay per use)
Call it ~$250/mo but plenty for what I need.
- Claude: $200/mo
- ChatGPT: $20/mo
- https://Fal.ai: $25 (credits)
- xAI: $5 (pay per use)
Call it ~$250/mo but plenty for what I need.
AI is increasing the pace of business.
https://Outbid.lol generated $200k in 5 days.
Dozens of copycats shortly after.
Speed is no longer a differentiator. It's table stakes.
https://Outbid.lol generated $200k in 5 days.
Dozens of copycats shortly after.
Speed is no longer a differentiator. It's table stakes.
August 27, 2026 at 4:09 PM
AI is increasing the pace of business.
https://Outbid.lol generated $200k in 5 days.
Dozens of copycats shortly after.
Speed is no longer a differentiator. It's table stakes.
https://Outbid.lol generated $200k in 5 days.
Dozens of copycats shortly after.
Speed is no longer a differentiator. It's table stakes.
What does the success of https://outbid.lol and the spawn of copycats say about todays industry?
August 26, 2026 at 4:09 PM
What does the success of https://outbid.lol and the spawn of copycats say about todays industry?
week 7 on x. the data:
21 → 22 followers
10 posts: 185 impressions
21 replies: 10,388 impressions
In an effort to reply more, I ask myself three questions: 1) what do i want to know 2) how can i relate? 3) what is the "yes and"? I tripled my replies this week as a result.
21 → 22 followers
10 posts: 185 impressions
21 replies: 10,388 impressions
In an effort to reply more, I ask myself three questions: 1) what do i want to know 2) how can i relate? 3) what is the "yes and"? I tripled my replies this week as a result.
August 25, 2026 at 4:09 PM
week 7 on x. the data:
21 → 22 followers
10 posts: 185 impressions
21 replies: 10,388 impressions
In an effort to reply more, I ask myself three questions: 1) what do i want to know 2) how can i relate? 3) what is the "yes and"? I tripled my replies this week as a result.
21 → 22 followers
10 posts: 185 impressions
21 replies: 10,388 impressions
In an effort to reply more, I ask myself three questions: 1) what do i want to know 2) how can i relate? 3) what is the "yes and"? I tripled my replies this week as a result.
I built a fantasy football draft board:
August 24, 2026 at 3:18 PM
I built a fantasy football draft board:
If you added a writing style to your Agents.MD file then you may have noticed it drift over time in long conversations. Mine was only being followed 60% of the time.
That's because of context rot.
That's because of context rot.
August 22, 2026 at 4:32 PM
If you added a writing style to your Agents.MD file then you may have noticed it drift over time in long conversations. Mine was only being followed 60% of the time.
That's because of context rot.
That's because of context rot.
AI has a positioning problem.
The vast majority of folks are fearful and doubtful of AI. e.g. data centers, taking jobs, etc
The vast majority of folks are fearful and doubtful of AI. e.g. data centers, taking jobs, etc
August 21, 2026 at 4:37 PM
AI has a positioning problem.
The vast majority of folks are fearful and doubtful of AI. e.g. data centers, taking jobs, etc
The vast majority of folks are fearful and doubtful of AI. e.g. data centers, taking jobs, etc