Keanan Koppenhaver
@kkoppenhaver.bsky.social
Learning everything I can about AI to help agencies spend less time stuck and more time building ➡️ http://floorboardai.com/subscribe
TPMM @retool
TPMM @retool
Pinned
👋 Hey y'all! If you don't know me already, I'm Keanan.
Based in Chicago, music and aviation nerd (and working on my PPL!)
I'll likely spend most of my time on here doing my best to debunk all the AI overhype and talk about how freelancers and agencies can actually use all these new tools.
Based in Chicago, music and aviation nerd (and working on my PPL!)
I'll likely spend most of my time on here doing my best to debunk all the AI overhype and talk about how freelancers and agencies can actually use all these new tools.
Testing, testing. Here is a Retool post from AI Build week.
July 14, 2025 at 5:29 PM
Testing, testing. Here is a Retool post from AI Build week.
Demo'd OpenRouter's auto-model selection at Retool's webinar last week and it blew my mind.
Asked for a joke (2.5s response) then "how many r's in strawberry?"
OpenRouter automatically switched to R1 for reasoning and took 89s to solve it correctly!
Full webinar: www.youtube.com/watch?v=ZJ1C...
Asked for a joke (2.5s response) then "how many r's in strawberry?"
OpenRouter automatically switched to R1 for reasoning and took 89s to solve it correctly!
Full webinar: www.youtube.com/watch?v=ZJ1C...
Model Mix and Match: How to build AI apps with (almost) any LLM
YouTube video by Retool
www.youtube.com
March 19, 2025 at 3:38 PM
Demo'd OpenRouter's auto-model selection at Retool's webinar last week and it blew my mind.
Asked for a joke (2.5s response) then "how many r's in strawberry?"
OpenRouter automatically switched to R1 for reasoning and took 89s to solve it correctly!
Full webinar: www.youtube.com/watch?v=ZJ1C...
Asked for a joke (2.5s response) then "how many r's in strawberry?"
OpenRouter automatically switched to R1 for reasoning and took 89s to solve it correctly!
Full webinar: www.youtube.com/watch?v=ZJ1C...
Wow, Claude got really good at generating complicated SVGs (and taking feedback well) when I wasn't looking...
Well done @anthropic.com 🎉
Well done @anthropic.com 🎉
March 1, 2025 at 9:36 PM
Wow, Claude got really good at generating complicated SVGs (and taking feedback well) when I wasn't looking...
Well done @anthropic.com 🎉
Well done @anthropic.com 🎉
Reposted by Keanan Koppenhaver
Boom Supersonic's experimental XB-1 aircraft has surpassed Mach 1 for the first time in its latest test flight. It's another step toward the possible return of supersonic commercial flights more than two decades after the Concorde's retirement. 🧪
www.newscientist.com/article/2465...
www.newscientist.com/article/2465...
Experimental XB-1 aircraft goes supersonic for the first time
Boom Supersonic’s XB-1 aircraft broke the sound barrier during three test runs, a step toward the possible return of supersonic commercial flights
www.newscientist.com
January 28, 2025 at 6:20 PM
Boom Supersonic's experimental XB-1 aircraft has surpassed Mach 1 for the first time in its latest test flight. It's another step toward the possible return of supersonic commercial flights more than two decades after the Concorde's retirement. 🧪
www.newscientist.com/article/2465...
www.newscientist.com/article/2465...
Reposted by Keanan Koppenhaver
Explainer: What's R1 and Everything Else
This is an attempt to consolidate the dizzying rate of AI developments since Christmas. If you're into AI but not deep enough, this should get you oriented again.
timkellogg.me/blog/2025/01...
This is an attempt to consolidate the dizzying rate of AI developments since Christmas. If you're into AI but not deep enough, this should get you oriented again.
timkellogg.me/blog/2025/01...
January 26, 2025 at 3:17 AM
Explainer: What's R1 and Everything Else
This is an attempt to consolidate the dizzying rate of AI developments since Christmas. If you're into AI but not deep enough, this should get you oriented again.
timkellogg.me/blog/2025/01...
This is an attempt to consolidate the dizzying rate of AI developments since Christmas. If you're into AI but not deep enough, this should get you oriented again.
timkellogg.me/blog/2025/01...
🤯 MacWhisper just transcribed ~20 mins of video in about 3 minutes completely locally.
Guess I need to get one of these new M4 chips for my personal machine.
Guess I need to get one of these new M4 chips for my personal machine.
January 22, 2025 at 12:16 AM
🤯 MacWhisper just transcribed ~20 mins of video in about 3 minutes completely locally.
Guess I need to get one of these new M4 chips for my personal machine.
Guess I need to get one of these new M4 chips for my personal machine.
The creative process is a combination of the creative tasks (initial idea, writing, etc) and the tedious tasks (outlining, distribution, etc).
AI helps make the tedious parts of the process easier so you can focus on your creativity.
For more on this, take a listen www.youtube.com/watch?v=j9o7...
AI helps make the tedious parts of the process easier so you can focus on your creativity.
For more on this, take a listen www.youtube.com/watch?v=j9o7...
January 21, 2025 at 8:38 PM
The creative process is a combination of the creative tasks (initial idea, writing, etc) and the tedious tasks (outlining, distribution, etc).
AI helps make the tedious parts of the process easier so you can focus on your creativity.
For more on this, take a listen www.youtube.com/watch?v=j9o7...
AI helps make the tedious parts of the process easier so you can focus on your creativity.
For more on this, take a listen www.youtube.com/watch?v=j9o7...
Reposted by Keanan Koppenhaver
the absolutely most absurd thing about AI rn is that distillation can **improve** the model
1. start with a BAM (Big Ass Model)
2. generate training data
3. train a TAM (Tiny Ass Model) on the synthetic data
seems like it breaks information theory until you view it as a gradient
1. start with a BAM (Big Ass Model)
2. generate training data
3. train a TAM (Tiny Ass Model) on the synthetic data
seems like it breaks information theory until you view it as a gradient
January 17, 2025 at 1:46 PM
the absolutely most absurd thing about AI rn is that distillation can **improve** the model
1. start with a BAM (Big Ass Model)
2. generate training data
3. train a TAM (Tiny Ass Model) on the synthetic data
seems like it breaks information theory until you view it as a gradient
1. start with a BAM (Big Ass Model)
2. generate training data
3. train a TAM (Tiny Ass Model) on the synthetic data
seems like it breaks information theory until you view it as a gradient
The "Crawl → Walk → Run → Fly" framework for agencies starting with AI:
Crawl: AI meeting notes and off-the-shelf tools
Walk: ChatGPT/Claude basics
Run: Custom prompts and instructions
Fly: Build custom AI-powered tools
Start small and see what works!
Full episode: www.youtube.com/watch?v=j9o7...
Crawl: AI meeting notes and off-the-shelf tools
Walk: ChatGPT/Claude basics
Run: Custom prompts and instructions
Fly: Build custom AI-powered tools
Start small and see what works!
Full episode: www.youtube.com/watch?v=j9o7...
An AI Crash Course for WordPress Users and Developers ft. Keanan Koppenhaver
YouTube video by Webmasters with Brian Coords
www.youtube.com
January 16, 2025 at 3:10 PM
The "Crawl → Walk → Run → Fly" framework for agencies starting with AI:
Crawl: AI meeting notes and off-the-shelf tools
Walk: ChatGPT/Claude basics
Run: Custom prompts and instructions
Fly: Build custom AI-powered tools
Start small and see what works!
Full episode: www.youtube.com/watch?v=j9o7...
Crawl: AI meeting notes and off-the-shelf tools
Walk: ChatGPT/Claude basics
Run: Custom prompts and instructions
Fly: Build custom AI-powered tools
Start small and see what works!
Full episode: www.youtube.com/watch?v=j9o7...
Excited to see this live! We covered a lot, so give it a listen!
New Webmasters episode: I talk with @kkoppenhaver.bsky.social all about AI- what its good for, bad for, and how WordPress builders can start using it in their workflow.
webmasters.fm/an-ai-crash-...
webmasters.fm/an-ai-crash-...
An AI Crash Course for WordPress Users and Developers ft. Keanan Koppenhaver - Webmasters.FM
Brian chats with AI expert and WordPress developer Keanan Koppenhaver about how AI tools are transforming workflows. They explore AI’s strengths, limitations, and practical uses in coding, brainstormi...
webmasters.fm
January 14, 2025 at 2:50 PM
Excited to see this live! We covered a lot, so give it a listen!
🔥 Just dropped: Watch me build an AI tool that turns academic papers into engaging blog posts! Using Retool and Anthropic's Claude, we'll go from PDF upload to polished blog post faster than you can say "publish."
Full video: youtu.be/mMB4vhoNUSo
Full video: youtu.be/mMB4vhoNUSo
I Built an AI Paper-to-Blog Tool (That Actually Works!) | HHCIB #1
YouTube video by How Hard Could It Be?
youtu.be
January 8, 2025 at 3:51 PM
🔥 Just dropped: Watch me build an AI tool that turns academic papers into engaging blog posts! Using Retool and Anthropic's Claude, we'll go from PDF upload to polished blog post faster than you can say "publish."
Full video: youtu.be/mMB4vhoNUSo
Full video: youtu.be/mMB4vhoNUSo
Watching yourself back on a video recording truly never gets less painful. I already have so many ideas about how to make the next one better, but gotta ship the first one first!
January 8, 2025 at 3:29 AM
Watching yourself back on a video recording truly never gets less painful. I already have so many ideas about how to make the next one better, but gotta ship the first one first!
This was a fun one to write!
Apps of the future 🤝 Auth from the future
Here's @kkoppenhaver.bsky.social with a tutorial on how to add #authentication to a @unitygames.bsky.social VR app:
✔️ Google and Discord social logins
✔️ OTP #MFA
✔️ #OpenIDConnect compatible
Get started: descope.com/blog/post/un...
#developers
Here's @kkoppenhaver.bsky.social with a tutorial on how to add #authentication to a @unitygames.bsky.social VR app:
✔️ Google and Discord social logins
✔️ OTP #MFA
✔️ #OpenIDConnect compatible
Get started: descope.com/blog/post/un...
#developers
Add Authentication and MFA to Unity VR With OIDC
Learn how to add seamless social login and MFA authentication to your Unity VR application using Descope and OIDC, simplifying user login with minimal coding.
descope.com
January 6, 2025 at 7:03 PM
This was a fun one to write!
I spend so much of my time seeing a cool idea online, asking myself "How Hard Could It Be?" and inevitably diving in and trying to find out.
Now, you can look over my shoulder as I do it!
The first episode of "How Hard Can It Be?" is coming very soon to YouTube!
www.youtube.com/watch?v=0CQ4...
Now, you can look over my shoulder as I do it!
The first episode of "How Hard Can It Be?" is coming very soon to YouTube!
www.youtube.com/watch?v=0CQ4...
How Hard Could It Be? (Trailer)
YouTube video by How Hard Could It Be?
www.youtube.com
January 3, 2025 at 3:14 PM
I spend so much of my time seeing a cool idea online, asking myself "How Hard Could It Be?" and inevitably diving in and trying to find out.
Now, you can look over my shoulder as I do it!
The first episode of "How Hard Can It Be?" is coming very soon to YouTube!
www.youtube.com/watch?v=0CQ4...
Now, you can look over my shoulder as I do it!
The first episode of "How Hard Can It Be?" is coming very soon to YouTube!
www.youtube.com/watch?v=0CQ4...
Reposted by Keanan Koppenhaver
It's been 2 years since ChatGPT and *still* people think prompt engineering isn't necessary to get good results.
I'd argue it's more important now as LLMs get better since they will converge on "average" output better, not "good" output.
I'd argue it's more important now as LLMs get better since they will converge on "average" output better, not "good" output.
General consensus in the replies and quotes of this seems to be that the entire concept of "AI skills" is a joke - how hard is typing text into a chatbot, really?
I will continue to argue that it's genuinely difficult, and that the challenge in using these tools is widely underestimated
I will continue to argue that it's genuinely difficult, and that the challenge in using these tools is widely underestimated
No, they're not caught between anything. Stop it. The fact that AI may be used or useful in future jobs does not create a single dilemma at all for college students in how to actually complete their coursework. This is all nonsense & excuses for not doing the reading & learning how to think & write.
December 26, 2024 at 8:06 PM
It's been 2 years since ChatGPT and *still* people think prompt engineering isn't necessary to get good results.
I'd argue it's more important now as LLMs get better since they will converge on "average" output better, not "good" output.
I'd argue it's more important now as LLMs get better since they will converge on "average" output better, not "good" output.
Reposted by Keanan Koppenhaver
Getting the most out of an AI is no different than having an employee or a consultant. If you can clearly articulate what you need done, then you've got fuck all chance of getting a good outcome. I don't know why anyone would think you can throw a poorly worded demand at an LLM and it will work.
December 27, 2024 at 9:47 AM
Getting the most out of an AI is no different than having an employee or a consultant. If you can clearly articulate what you need done, then you've got fuck all chance of getting a good outcome. I don't know why anyone would think you can throw a poorly worded demand at an LLM and it will work.
Reposted by Keanan Koppenhaver
OpenAI announced their newest o3 model techcrunch.com/2024/12/20/o...
OpenAI announces new o3 models | TechCrunch
OpenAI saved its biggest announcement for the last day of its 12-day "shipmas" event. On Friday, the company unveiled o3, the successor to the o1
techcrunch.com
December 21, 2024 at 12:16 PM
OpenAI announced their newest o3 model techcrunch.com/2024/12/20/o...
More news coming soon, but I'm very excited about this. The early feedback from some folks who have seen it has been VERY positive. Stay tuned 👀
December 20, 2024 at 10:03 PM
More news coming soon, but I'm very excited about this. The early feedback from some folks who have seen it has been VERY positive. Stay tuned 👀
Reposted by Keanan Koppenhaver
It's wild to me how much expertise you need to engage successfully with these new AI systems
Quick: what sort of prompts should you run against GPT-4o vs Gemini 1.5 Flash vs o1 vs o1-pro vs gemini-2.0-flash-thinking-exp?
Quick: what sort of prompts should you run against GPT-4o vs Gemini 1.5 Flash vs o1 vs o1-pro vs gemini-2.0-flash-thinking-exp?
December 20, 2024 at 1:58 AM
It's wild to me how much expertise you need to engage successfully with these new AI systems
Quick: what sort of prompts should you run against GPT-4o vs Gemini 1.5 Flash vs o1 vs o1-pro vs gemini-2.0-flash-thinking-exp?
Quick: what sort of prompts should you run against GPT-4o vs Gemini 1.5 Flash vs o1 vs o1-pro vs gemini-2.0-flash-thinking-exp?
Reposted by Keanan Koppenhaver
🔥 NEW SERIES ALERT — introducing a game show for developers: Leet Heat 🌶️🌶️
contestants face the worst tech interview of their careers: bring the web dev knowledge... or face the heat 🌶️
dev trivia & spicy bites, hosted by @marktechson.com and sponsored by @jetbrains.com WATCH: lwj.dev/leet-heat/s1e1
contestants face the worst tech interview of their careers: bring the web dev knowledge... or face the heat 🌶️
dev trivia & spicy bites, hosted by @marktechson.com and sponsored by @jetbrains.com WATCH: lwj.dev/leet-heat/s1e1
Pilot: A Gameshow for Developers — Leet Heat S1E1
Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.
lwj.dev
December 18, 2024 at 2:46 PM
🔥 NEW SERIES ALERT — introducing a game show for developers: Leet Heat 🌶️🌶️
contestants face the worst tech interview of their careers: bring the web dev knowledge... or face the heat 🌶️
dev trivia & spicy bites, hosted by @marktechson.com and sponsored by @jetbrains.com WATCH: lwj.dev/leet-heat/s1e1
contestants face the worst tech interview of their careers: bring the web dev knowledge... or face the heat 🌶️
dev trivia & spicy bites, hosted by @marktechson.com and sponsored by @jetbrains.com WATCH: lwj.dev/leet-heat/s1e1
Reposted by Keanan Koppenhaver
🔥 NEW WEB DEV CHALLENGE 🚀
there’s chaos is Santa’s workshop, and I challenged @aaronfrancis.com, @kentcdodds.com, @devdevcharlie.com, and @domitri.us to fix it in time to save Christmas
@sanity.io sponsored this episode on location in San Francisco — don’t miss it!
youtube.com/watch?v=_H5p...
there’s chaos is Santa’s workshop, and I challenged @aaronfrancis.com, @kentcdodds.com, @devdevcharlie.com, and @domitri.us to fix it in time to save Christmas
@sanity.io sponsored this episode on location in San Francisco — don’t miss it!
youtube.com/watch?v=_H5p...
4 Hours to Save Santa's Workshop — Web Dev Challenge S1E10
YouTube video by Learn With Jason
m.youtube.com
December 17, 2024 at 3:13 PM
🔥 NEW WEB DEV CHALLENGE 🚀
there’s chaos is Santa’s workshop, and I challenged @aaronfrancis.com, @kentcdodds.com, @devdevcharlie.com, and @domitri.us to fix it in time to save Christmas
@sanity.io sponsored this episode on location in San Francisco — don’t miss it!
youtube.com/watch?v=_H5p...
there’s chaos is Santa’s workshop, and I challenged @aaronfrancis.com, @kentcdodds.com, @devdevcharlie.com, and @domitri.us to fix it in time to save Christmas
@sanity.io sponsored this episode on location in San Francisco — don’t miss it!
youtube.com/watch?v=_H5p...
This is what happens when I'm stuck on a plane with limited WiFi 😅
Where do you think we are (or where are you personally) in the AI hype cycle?
Where do you think we are (or where are you personally) in the AI hype cycle?
December 16, 2024 at 8:15 PM
This is what happens when I'm stuck on a plane with limited WiFi 😅
Where do you think we are (or where are you personally) in the AI hype cycle?
Where do you think we are (or where are you personally) in the AI hype cycle?
Reposted by Keanan Koppenhaver
The only time I ever go all “USA! USA!” is when talking about our range of beautiful and very ecologically diverse natural landmarks. Wherever you are in the US, you’re not really ever more than a day’s drive away from some sort of beautiful natural feature that will stir your spirit
December 14, 2024 at 5:50 AM
The only time I ever go all “USA! USA!” is when talking about our range of beautiful and very ecologically diverse natural landmarks. Wherever you are in the US, you’re not really ever more than a day’s drive away from some sort of beautiful natural feature that will stir your spirit
After 3 full days of using the badge scanners at AWS re:Invent with the Retool team, I had a few ideas for additional features and other ideas (because of course I did 🤓).
Turns out, after about an hour in Retool, I was able to build our own version!
youtu.be/q4n1Ob2Jegc
Turns out, after about an hour in Retool, I was able to build our own version!
youtu.be/q4n1Ob2Jegc
Rebuilding the NFC badge scanner from AWS re:invent in Retool
YouTube video by Keanan Koppenhaver
youtu.be
December 12, 2024 at 7:08 PM
After 3 full days of using the badge scanners at AWS re:Invent with the Retool team, I had a few ideas for additional features and other ideas (because of course I did 🤓).
Turns out, after about an hour in Retool, I was able to build our own version!
youtu.be/q4n1Ob2Jegc
Turns out, after about an hour in Retool, I was able to build our own version!
youtu.be/q4n1Ob2Jegc