|
In partnership with
|
 |
|
OpusClip started with clips. Now it edits entire videos.
|
|
|
|
You probably know OpusClip as the tool that turns long videos into short clips.
|
|
With OpusClip 4.0, it's now a full AI video editor.
|
|
Drop in your raw footage or paste a link, toggle on the types of edits you want applied, and get back your finished video. You never have to touch a timeline!
|
|
The new OpusClip has it all:
|
-
Viral Hook opens your video with a flashy hook style from a library of proven templates.
-
Motion Graphics turns what you say into animated icons, numbers and charts.
-
Speech cleanup, captions and AI B-roll are one toggle each.
-
Recipes save your setup, so next week's footage gets edited the same way.
|
|
Full podcast episode cleanup, 60-second highlight clips, low-fi yapper videos: OpusClip handles all of it.
|
|
Want to try it out? Get 40% off your first month of Pro Plan.
|
|
Try the new OpusClip →
|
 |
Today's drop
π€ Kitesurf Lets Agents Browse, No Chrome
⭐ 1,545 · cloudflare/kitesurfRepo
Many AI agents that read the web run a hidden copy of Chrome to do it. Cloudflare has an experimental project that tries another way, and it is today's lead pick. Further down: an agent that works the apps on your Android phone or Mac the way you do.
An experimental browser engine that runs entirely on Cloudflare Workers, no Chrome or browser container involved.
That hidden Chrome is often the slow, costly part of a scraping or screenshot job. Kitesurf loads the page, runs its code and takes the screenshot itself. It speaks the Chrome DevTools Protocol, so a Puppeteer script can connect to it.
Quick start:pnpm install --frozen-lockfile after cloning, then pnpm run dev, which builds with Bazel and starts it on your machine.
Catch: it is experimental, running it on Cloudflare needs the Workers Paid plan, and the first build compiles a Rust core, which takes a while.
Open the repo →
HubSpotAffiliate
HubSpot builds AI agents on the customer data already in your CRM, like a Prospecting Agent that watches for buying signals and starts the outreach. HubSpot's free version never expires. See how HubSpot worksWe may earn a commission.
andeeTool⭐ 6,144
An agent that lives on your Android phone or Mac and uses any app the way you do, no root or integrations needed. Catch: it can pay, send messages and act as you, and its README calls it experimental. The recommended install runs through CodeFlying, a separate service with its own terms and pricing.
inferssdRepo⭐ 1,175
Runs AI models too big for your computer's memory by reading parts of them from your SSD as they are needed. Catch: the largest models are a download of about 470 GiB, and Windows is still listed as in progress.
modern-web-guidanceRepo⭐ 17,449
Gives your coding agent current web best practices and browser support data, so it stops writing outdated code. It is stewarded by the Chrome, Edge, Firefox and MDN teams. Catch: it only covers web work (HTML, CSS and browser features), and you add it to your agent with an npx install.
|
|
|
Speech models ace the lab. Canto's built for your commute.
|
|
|
|
Speech recognition is great on clean studio audio. Real dictation happens at your desk, on a walk, or in a loud cafe. That's why Wispr Flow built Canto, its own speech model, trained on how people really talk.
|
|
On 10 hours of real dictations from 2,300+ speakers, it had the lowest word error rate of every model tested, ahead of models from Google, OpenAI, AssemblyAI and Deepgram. Background music, traffic, someone talking next to you, a whisper: on a challenge set full of exactly that, Canto had the lowest error rate among real-time models tested.
|
|
Talk naturally and Wispr Flow types clean, formatted text in any app. 4x faster than typing.
|
|
Try Wispr Flow free
|
 |
New here? Start with The Awakening. It's free, it takes one sitting, and it's where every reader should begin.
Take The Awakening →
Turn a cold email that went nowhere into three follow-ups that actually earn a reply
Start Here is one prompt from our pack, every issue, tested on ChatGPT and Claude. Most people send one follow-up that just says "bumping this" and then give up. This prompt builds a real three-message sequence instead, with each one carrying something new. Paste in the original email, who you sent it to, and what's changed since, and the AI does the rest. Tap Copy prompt, paste it into the AI app you already use, and fill in the bits in capitals there.
The line that makes it work: "Each message gives them a reason to reply that was not in the one before it. A new thought, a short example, or real news counts." That line is what stops the sequence from reading like three copies of the same nag. Most people skip straight to writing message two before checking whether it says anything message one didn't.
TRY THIS: Pull up the last cold email you sent that got silence. Copy it in, tell the prompt how many days it's been, and let it hand you the next move.
This is one of 250 in the AIgent prompt pack. The rest, plus guided builds with Mike on video, are inside the AIgent Academy on Skool: Join the AIgent Academy.
|
|
EARN FROM HOME
Get paid to train AIAffiliate
AI companies pay people to rate, fix and check what their models write. Open projects list pay of up to $17 an hour for AI evaluation and up to $125 an hour for specialist roles, US-based, part-time on your own schedule. Browse live roles and apply free.
See open AI training roles →
We earn a referral fee if you're hired through this link. It never changes your pay.
|
|
 |
|
OpenAI introduced ChatGPT Wallet, which lets its agents pay without seeing your card details. Every purchase needs a security check, and your card keeps its rewards and dispute protections. It starts with ChatGPT Pro subscribers in the US, and OpenAI is also releasing tools and guidance for businesses that want to sell to agents. (OpenAI Blog)
llama.cpp hit 1.0, cutting 35 deprecated functions from its public header. The project is declaring its interface stable, which should mean fewer surprise breakages if you run AI models on your own machine. If your pipeline pins an old build, this is your prompt to test the upgrade path now. (Latent Space)
datasette-llm now supports OpenAI-compatible endpoints, so a local model running in LM Studio, Ollama or llama.cpp can power it. It is the piece other Datasette plugins use to reach a model, so those plugins can now run on one that lives on your own computer. Worth a look if you're trying to cut inference costs out of a side project. (Simon Willison)
A disgruntled researcher dropped a Windows zero-day called SignDrop that turns Defender's own signature-update function against itself to grant System rights. The tool meant to protect your machine can become the attack surface. It needs someone already on the machine as an ordinary user, so if you run agents on Windows, be careful what you let them run until Microsoft ships a fix. (Ars Technica)
|
|
Recommended reading
If you like The AIgent, a small group of operator-tier publications worth your inbox: see the shortlist.
|
|
What do you spend on AI tools in a typical month?
|
|
|
 |
|
Hit reply and tell me which one you're pulling this week: Kitesurf for the scraping bill, or Andee for the app you gave up trying to integrate. I want to know what breaks first when you run it.
|
|
See you Tuesday.
|
|
Haven't taken The Awakening yet? It's free and it's where we'd start. Take The Awakening →
|
|
|
|