- Atomic AI, which uses machine learning to explore the potential of RNA in drug discovery, raised a $35 million Series A round.
- They have developed a machine learning model that can accurately predict the structure of RNA molecules from a limited set of data.
- They have since improved their model and are using it to advance their own drug discovery program, which produces candidate molecules that could work to treat conditions that are drug resistant or notoriously difficult to treat.
Now that I’ve written the articles, I think they’re pretty solid summaries, but short enough to (hopefully) pique your interest. Not everyone browses and clicks on links in an article, because who knows what else is relevant? Are you going to open 10 tabs to find out? Or will you try to guess from the URL? The popup seems pretty nice to me and only appears when you want it to:
Here, try it live further this post – you may or may not care about football salaries, but you can see it popping up and disappearing. That appearance is also custom; the length of the summary, bullet points, if and when it fires, all of that can be changed. So what happens when link previews get this extensive AI digest treatment?
“We’re seeing about 50 percent more page views,” says Shrager. “I’m quite an analytical person and I was nervous about what we saw because it’s a bit counterintuitive. But Wikipedia did this, and it was very successful. It reduces research costs for the user.”
Getting a bigger hint as to what they’re hovering over clears the hump of “should I click on this or not?” The smallest bump in the world, sure, but if you told a web publisher you could increase clicks by one percent – let alone 50 percent – they’d jump on it. On-site time and engagement are valuable metrics, and finding ways to increase them upwards is a big part of any product manager’s job.
Different links, such as affiliate links, can give different examples, for example advantages and disadvantages of a product summarized from the last hundred reviews. Or external links can be left naked – perhaps (to be fair) to avoid the same click effect benefiting them.
Shrager noted that they’ve done a lot of work under the hood to make sure summaries don’t exhibit the kind of “creativity” that language models are notorious for — names, dates, quotes, and other things are always preserved, for example, and changing wording is limited to places where it will not change the meaning. “All of our valuable IP is all the know-how and knowledge that is before and after the AI model,” he said.
Ultimately, though users benefit, Summari’s customers are now website operators. The company charges a flat fee for access to the tool and then a small usage fee.
“If your average article has 1,000 words, and you have five, we summarize 5,000 words and write 500, and maybe charge 50 cents,” Shrager offered as a very loose example of the scale. “We try to make the total price the minimum so it’s not a barrier to a sale — the only way we can scale quickly is to find major distribution channels.
After the initial partners, Summari goes live on a major academic publishing house and a major news site, both unnamed. “We’re definitely noticing FOMO across the industry,” Shrager said. “People are sniffing around, starting to see the value. There is a natural network effect as backlinks are aggregated. Once gotechbusiness.com summarizes the use of us, you won’t want to turn it off and do it all over again.”
I suggested that, with big tech players like Microsoft and Google making huge efforts in AI, the company shouldn’t be surprised if an offer or two slips under the door. After all, such summaries would be great in search engines or an algorithmically curated news site. But Shrager said they are not looking for a quick exit.
“My job is to maximize shareholder value. If I got an attractive offer from Google or Yahoo, it could be of huge strategic advantage… I’m not crazy, I’d go for it. But everyone here is going for the big win.”