hunter peratt

Introducing One Click Finish. The first of its kind, fully custom ai model trained specifically to return interior design products. The model curates fixtures and finishes in seconds, with a users text prompt or with a reference photo. We return real products from real home retailers, with links that take you to the product page. We also generate an image, after the list of products is returned, to help visualize the space. You are able to add rooms to a design, and the new room products stays cohesive to the current room. You are able to shift a design is a new directions. Ex: “make it warmer, add more brass” We have many pre made packages you can browse for free, and see the prompts that were used to make them. Also a fun feature, we have semantic search. Search a vibe you want, and a list of pre made packages will populate, tailored to that search input. (Screenshot below) Infinite styles, infinite results. The bathroom image below is for visualization. The product list was curated by our model. The bathroom below was made with a Mediterranean home image from Pinterest.

@TanayVasishtha bro posted this 4 months from now

I was way ahead of my time, but it’s still performative.

@DTC_Quizbuilder Yeah ad to purchase, ty 🙇

@_hunterperatt In terms of ad to purchase? At scale (depending on APV) roughly 2-3% should be good

Although I am number 1 shiller of quiz funnels I think listicles are incredibly powerful - especially if you're a beginner without much copy experience They're almost like dumby product pages & easy to churn our with congruent headlines Not hard to throw 5 reasons together Make the copy super simple Reasons are easy to digest Customers will easily be able to digest this information In some niches where advertorials or VSL's get rinsed - it becomes harder to make yours stand out they also require a much higher skill lvl with good research documents & claude you should be able to get enough proof of concept with your listicle & fill your PDP with social proof

@DTC_Quizbuilder I had a longggg landing page with CTA to quiz. Drop off was 80-85%. Switched to simple intro page with headline, sub head, and CTA to start quiz. Drop off now at 60%

Immediate test ideas: 1) Remove the pre-quiz page & CTA drop down (didn't even see the quiz) Either go Bioma style landing page with CTA > quiz on diff page or Just go straight into DR style quiz and capture your core offer in a headline + subhead If I had to guess most users bounce and start rate is quite low? DM me happy to help

@DTC_Quizbuilder What are your thoughts on a landing page with CTA to start quiz?

@skirano @rameadows What’s new model called??

@rameadows New Meta model

Visual AGI is here btw

@JasonJh1319 @iblamejulius thoughts here?

Among all the Creative Strategist skills tests I've seen recently (30+): Copywriters surprisingly make the worst candidates when they try to fit this role. They often miss or dismiss the importance of visuals and focus too much on overt direct response.

@LeCodeBusiness @unusual_whales And many homeowners have lots of equity (if they bought before 2020) so if they have to fire sale then they will be just fine

@unusual_whales The demand collapse is real but the supply side is the bigger story. Homeowners locked into 3% mortgages won't sell. So you get falling buyers AND frozen inventory, prices don't crash, they just seize. Different dynamic than 2008.

"California housing crash fears as buying rates plummet below Great Recession level," per NYP

@LeCodeBusiness @unusual_whales 100% spot on. Very different that 08

@BigBearBummers Similar thing happening in Denver. I see new listings everyday where the seller paid way too much in 2021-2022. They aren’t willing to accept a loss though. Many are listed 100k-200k more than what they paid

In '22 they paid a jaw-dropping $415,000 – a valuation only possible during a speculative bubble. But that bubble popped and now they're facing utter financial devastation. This is beyond short sale territory – they need to mail the keys back to the bank.

@thedanielokon @theryandoney It’s gonna be technical to manually implement. I’m sure people are already building software on top of it. Maybe I’ll mess around with it in my free time and see what happens.

@_hunterperatt @theryandoney is there an easy way to get this setup? Seems like an incredible tool to implement

You can make ads with this. Give it every creative asset, every review, every competitor ad. Point your AI strategist at your new vector database. Brief new ads. Now it pulls context from EVERYTHING YOU'VE EVER LAUNCHED to inform new creative. Others will do boring stuff like economic data analysis and high-stakes legal work. But we can make ads with this.

@SahilAutomates @thedanielokon @theryandoney Tokens would count when the LLM uses the data to write something back to you.

@SahilAutomates @thedanielokon @theryandoney The embeddings don’t count towards token count. The embeddings just store in depth information about the data

@thedanielokon @theryandoney I’ll revise, probably 1,000,000x better

@thedanielokon @theryandoney It’s like uploading a bunch of creatives, ad copy, metrics, etc, to Claude and asking it for advice but 10,000x better. You can embed infinite data with it, and ask an LLM to do anything with the data.

Apparently I need to repeat myself because I’ve been on too many consulting calls with yall and you keep fucking this up when you’re running ads for lead gen, calls, webinars, etc. all this copy and messaging and position and creative strategy and CRO shit is great but if you don’t have a 9.3+ event match rate literally NONE of it matters. Your CPA is going to be double what it should. Your costs are going to fluctuate like crazy as you scale. You might as well just Venmo daddy zucc directly and help him buy another yacht. I repeat: IF YOU DO NOT HAVE A 9.3+ EVENT MATCH SCORE ON THE EVENT YOURE OPTIMIZING FOR STOP EVERYTHING YOURE DOING AND FIX THAT SHIT RIGHT NOW. Lmk if yall need a guide on this

@thisisnikolai @Andersonmancini Looks amazing! When is beta live?

Excited to show off the level of realism we've achieved in the browser… We're calculating realtime light bounces with SSGI - all running in WebGPU + ThreeJS (no AI) This was a Herculean effort - kudos to the HomeFig team + special thanks to @Andersonmancini for helping us

@cursor_ai Can two cursor agents talk to each other when running in different windows? If I have back end and front end running can they team up?

Cursor can now control your browser. Agent can take screenshots, improve UI, and debug client issues. Try our early preview with Sonnet 4.5.