Rez Karim
@rezkhere • 38,092 subscribers
Giving your AI agents superpowers at @unsora_ai. I talk AI video, ads & prompt engineering.
Shorts
Videos
This is wild... A1 sets a new standard for live avatars, and my proof is a goose personal trainer. I built Coach Honk with A1 from Vivix.AI. The first thing I told him: “I came here for arms.” He took it personally. Then I asked for a proper squat and he dropped into a full-body demo. After that, he walked over, grabbed a kettlebell and showed me a goblet squat. The kettlebell stayed a kettlebell the whole time. Which is rare in gen AI lol. The whole thing was way smoother than I expected, especially the convo. Answered quickly and kept the back-and-forth natural. Best of all, the characters know how to work with interruptions unlike most AI avatars. Making one is simple. — Go to the Agents tab in Vivix AI — Describe the character you want and how it should behave, then hit enter. It builds the character step by step and once it's done, you can start talking to it. You can also publish your character and share with friends. Talk to Coach Honk here:
Rez Karim41,936 次观看 • 1 个月前
Gave Teamily AI an Edinburgh trip: 4 friends, 5 days. 3 agents on 1 shared whiteboard, basically a team AI for getting trips finally out of the group chat. - One found where to stay and how to get around. - One planned out the days from that. - One kept a running budget off those days. I even interrupted, telling them that one friend could only stay for the first 3 days. The days got reshuffled so all the highlights land before they leave, and the budget split into 2 totals, all from the same plan. Then my collaborator asked for a checklist right in the chat and the agents picked it up. Once we were happy, the Webpage Developer turned the whole plan into a trip page with Website Builder. Start to finish in one chat and you can also edit it as needed in Studio. Trip page: Wild stuff Teamily AI
Rez Karim19,946 次观看 • 17 天前
Holy shit... I think I created a monster. 1. Upload your product 2. Select from past successful ads in your niche/from competitors. 3. Get as good or better ads for your brand Comment 'ads' and I will DM you the access. Also share your product below. Imma make you some ads :)
Rez Karim194,824 次观看 • 1 年前
Step 2: Get ChatGPT to plan the video Use this prompt. Change video topic as needed. "I want to create a storytelling video for my coffee shop 'Drip Drop' using Visla Plugin. Brainstorm the other things about the video that Visla plugin takes as input and outline them below."
Rez Karim180,129 次观看 • 3 年前
Animation ads are ripping rn Small brands usually never ran them. One 3D ad meant a studio, a storyboard and a month of revisions. So when people are suddenly coming across a clay motion ad for a $30 serum. It's scroll stopping. Arcads just shipped the animation studio and it squashes that whole process into one prompt. Here is what I would do: - Make a list of the top 5 brands in your space - Search each one in the Meta Ad Library - Find the oldest ads they're still running Those are probably the best performers with the highest ROAS. A brand that keeps paying for the same ad for months is getting paid back. - Send them over to the Arcads agent, Mark and ask for the hook, the pain point, the offer - Put in your URL and product image and build that winning angle into your prompt - Generate the same ad in all 6 styles - Keep the first round short (I'd start around 15 seconds and only go longer once a style proves itself) Now you're testing style and angle at the same time, and each creative costs you a few minutes of waiting. - Run every version on Meta as its own ad set - Cut the losers after a few days and keep the style with the highest ROAS - Make 10 more variations in that style with fresh hooks - Find the winners and double down - Repeat with the next brand's angle Animation used to be a brand awareness play for companies with money to burn. Now you can run it as a performance ad and shut it down it in 3 days if it flops.
Rez Karim12,036 次观看 • 1 个月前
![Remember when AI couldn't draw a hand? Seven fingers, knuckles pointing backwards. And the AI spaghetti videos. That was three years ago. Images are done now. Video is close enough that you scrolled past AI ads this week and clocked exactly zero of them. Code writes itself and there are like 40 coding agents. AI voice spent that entire stretch sounding like the lady voice in a 2014 GPS. Flat, evenly spaced and every sentence landing with the same weight, like it's reading from a phone book. Here's why it stayed broken. Bad images are funny. You screenshot the seven fingers, it goes viral for being bad, someone fixes it. Bad audio is just boring. It doesn't fail spectacularly, so it never got that pressure. The bigger problem was the scoring. The whole industry graded AI voices on whether you could make out the words. So the models learned to over pronounce everything, hitting every syllable like a newsreader. Perfectly clear but robotic. Everyone was chasing a score that had nothing to do with sounding human. Meanwhile a small open-source team was doing something harder. Their lead researcher, an ex-NVIDIA engineer, went all in on an approach the rest of the field had written off. Two years early. No funding announcements or launch tour. He just put the whole thing on GitHub for free. It's sitting at 50,000+ stars now. Then they ran the test everyone else avoided. For 10 days they piped real users through their model and every big competitor with the listener never told which was which. Thousands of real people, real scripts. Whichever voice you actually preferred, they logged it. Theirs came out on top. It beat ElevenLabs about 6 times out of 10, head to head. It beat OpenAI's voice model 8 times out of 10. The gap was widest on the breathing, the pauses, the little hesitations, which is exactly the stuff that makes a voice sound like a person instead of a machine reading. They ran on real users rather than a lab, which is more than most of these claims can say. That's Fish Audio. This week they shipped S2.1 Pro: - Clone anyone's voice from 15 seconds of audio - Fast enough to hold a live conversation - 83 languages, one model - Type [whisper] or [sigh] mid-sentence and it does it - Around 70% cheaper than ElevenLabs - Free to download and run yourself Voice was the last thing on the list. Around 20 people with a free repo got there before other billion dollar companies did.](https://image.24vids.com/tw-2082409789385224219/amplify_video_thumb/2082408743611113472/img/Iub3c6E4muQa_yeC.jpg)
Remember when AI couldn't draw a hand? Seven fingers, knuckles pointing backwards. And the AI spaghetti videos. That was three years ago. Images are done now. Video is close enough that you scrolled past AI ads this week and clocked exactly zero of them. Code writes itself and there are like 40 coding agents. AI voice spent that entire stretch sounding like the lady voice in a 2014 GPS. Flat, evenly spaced and every sentence landing with the same weight, like it's reading from a phone book. Here's why it stayed broken. Bad images are funny. You screenshot the seven fingers, it goes viral for being bad, someone fixes it. Bad audio is just boring. It doesn't fail spectacularly, so it never got that pressure. The bigger problem was the scoring. The whole industry graded AI voices on whether you could make out the words. So the models learned to over pronounce everything, hitting every syllable like a newsreader. Perfectly clear but robotic. Everyone was chasing a score that had nothing to do with sounding human. Meanwhile a small open-source team was doing something harder. Their lead researcher, an ex-NVIDIA engineer, went all in on an approach the rest of the field had written off. Two years early. No funding announcements or launch tour. He just put the whole thing on GitHub for free. It's sitting at 50,000+ stars now. Then they ran the test everyone else avoided. For 10 days they piped real users through their model and every big competitor with the listener never told which was which. Thousands of real people, real scripts. Whichever voice you actually preferred, they logged it. Theirs came out on top. It beat ElevenLabs about 6 times out of 10, head to head. It beat OpenAI's voice model 8 times out of 10. The gap was widest on the breathing, the pauses, the little hesitations, which is exactly the stuff that makes a voice sound like a person instead of a machine reading. They ran on real users rather than a lab, which is more than most of these claims can say. That's Fish Audio. This week they shipped S2.1 Pro: - Clone anyone's voice from 15 seconds of audio - Fast enough to hold a live conversation - 83 languages, one model - Type [whisper] or [sigh] mid-sentence and it does it - Around 70% cheaper than ElevenLabs - Free to download and run yourself Voice was the last thing on the list. Around 20 people with a free repo got there before other billion dollar companies did.
Rez Karim18,446 次观看 • 2 个月前
没有更多内容可加载