TOP LAUNCHES

Cursor for your pitch deck

TOP LAUNCHES

DeckSpeed listens to your meeting, brainstorm, or off-the-cuff voice note and spins it into a finished slide deck. It writes the storyline, chooses layouts, drops in charts, images, and even 3-D models, then hands you a PowerPoint file.

Granola for iOS is an AI note-taking app for people stuck in nonstop meetings. It records the call, transcribes it, and turns the mess into a clean, structured summary you can actually use, all from your phone.

Substage slips a natural-language command bar under every Mac Finder window. Type “zip these,” “convert to 1080p mp4,” or “word count,” hit return, and it does the thing—no menu spelunking, no Terminal ritual.

Antispace connects Gmail, Calendar, Notes, GitHub, and more into one command bar. You type what you want and your AI sidekick does it. No tabs. No toggling. The new update lets you control how your AI looks, thinks, and behaves.

FineTuner lets you upload content from PDFs, websites, or videos and fine-tune a GPT or Claude model without writing code. It builds the dataset, runs training, and gives you an API.

THE BIG IDEA

OpenAI split in two

OpenAI just split its soul in two—and gave each half a new job title.

The company is restructuring into a capped-profit parent and a fully nonprofit research lab. The new setup separates frontier model research (run by OpenAI Nonprofit) from the productized ChatGPT empire (run by the for-profit OpenAI Global), with CEO Sam Altman leading both. Chief Scientist Ilya Sutskever and alignment researcher Jan Leike have exited stage left, but not before Sutskever’s now-famous “superalignment” team was quietly dissolved.

Under the hood, this looks like a response to mounting pressure. Between safety critics, government scrutiny, and uneasy investors, OpenAI’s trying to reassure all sides: moonshot research can still happen, and ChatGPT can still ship updates on time.

It’s also a sign that alignment and commercial viability are drifting apart. One group will chase AGI safeguards. The other will chase monthly active users. Ideally, they’ll still talk.

Just maybe not over Slack.

TRENDING DISCUSSIONS

Break down those barriers

Wawi kicked off a discussion that’s still waiting for its first replies: what actually stops people from adopting AI products?

They point to human bias, low awareness, and unclear value props as some of the hurdles—and invite others to share what’s worked (or flopped) in their own experience launching AI tools.

The thread’s quiet for now, but it asks the right question. If you’ve ever tried convincing someone to trust your AI product, this could be your cue to chime in.