- Atomic AI, which uses machine learning to explore the potential of RNA in drug discovery, has raised $35 million in Series A funding.
- They developed a machine learning model that can accurately predict the structure of RNA molecules based on limited data sets.
- They then refine the model and use it to pursue their own drug discovery programs. The program generates candidate molecules that may be useful in treating conditions notoriously difficult to treat or drug resistant.
I’ve finished the article and think it’s a pretty solid summary, but (hopefully) brief enough to be interesting. Not everyone visits articles and clicks links. Are you going to open 10 tabs just to look it up, or are you guessing from the URL? Popups are a pretty nice touch to me and only appear when needed.

Animated example of the Summary popup.
Try it live here in this post. You may or may not care about football salaries, but you can watch them pop up and disappear. You can change everything from the length of the summary, the bullet points, if it triggers and when it triggers. So what happens when Link Preview gets this rich AI summarization?
“We’re seeing a 50% increase in page views,” says Shrager. “I’m a fairly analytical person and I was nervous because what we were seeing was a little counterintuitive. search costs are reduced.”
Having a bigger hint of what the hovering thing will be reduces the “should I click this or not?” question. It’s the smallest hump in the world, sure, but if you tell web publishers you can increase clicks by 1 percent, let alone 50 percent, they’ll jump at it. Time on site and engagement are valuable metrics, and finding ways to increase them is a big part of a product manager’s job.
Different links, like affiliate links, can provide different previews, for example, product pros and cons summarized from the last 100 reviews. Alternatively, you can leave the external links alone — probably (honestly) so that the same click effect doesn’t benefit them.
Shrager said he’s done a lot of work under the hood to ensure that the abstract doesn’t exhibit the kind of “creativity” that language models are notorious for. For example, names, dates, quotes, etc. are always preserved. Changes in wording are limited to places where the meaning remains the same. “All of our valuable IP is all the know-how and knowledge before and after AI models,” he said.
Ultimately, users benefit, but Summari’s customers are now website operators. The company charges a flat fee for accessing the tool, followed by a small usage fee.
“If your average article has 1,000 words, and you have 5 of them, we’ll summarize the 5,000 words and write 500 words, and charge maybe 50 cents.” trying to make Deminimis Therefore, it does not prevent sales. The only way to scale quickly is to find a large distribution channel. ”
After the initial partners, Summari will be published in major academic publishers and large news sites (both unnamed). “The whole industry definitely has his eye on FOMO,” he said, Shrager. “There are people who are starting to realize its value and are sniffing around. We have a natural network effect because our backlinks are aggregated. No need to start over.”
I suggested that with major technology players such as Microsoft and Google playing big roles in AI, it shouldn’t be surprising if the company slips an offer or two under the door. After all, a summary like this would be great for a search engine or an algorithmically curated news site. But Shrager said they weren’t asking for a quick exit.
“My job is to maximize shareholder value. If we have an attractive offer from Google or Yahoo, it could be of great strategic advantage… I am not stupid. We are all looking for big wins.”