If you’re having trouble keeping track of your work life as it spreads across 20 different apps and productivity suites, to the point where you have a meeting in 5 minutes that you don’t even know you created, then you’re not the only one. We can’t all have a PA to keep track of for us, but Augment hopes that a PAI, if you want, might be the best, learn from every digital interaction you have and provide the information you need before you get there. even ask for it.
“Wait a minute” I hear you typing. “Didn’t you write this story about Atlas Informatics five years ago?” Why yes, I did – what a memory you have. But Atlas sank without a trace, as founder (and co-founder of Napster) Jordan Ritter disappeared from my radar until a few weeks ago when Augment reached out to talk about what amounts to the next generation of the idea, with a decidedly more modern approach.
The basic idea of Atlas was that “everything seen is remembered”, making your entire online world locally searchable, from appointments and contacts to tab groups and purchases. But linking to search seems to have been a misstep, Ritter explained to me, for several reasons, but perhaps only because searching for something assumes you know what you’re looking for. The new problem is that our data is so dispersed that you may not even remember that there is anything to remember.
“Instead of searching, we are building a learning AI. Basically, it teaches what’s important to you to help you be more productive,” Ritter says. The system will make use of modules that cater to different needs, the first of which is “for people with stuck lives. We don’t remember everything; we are unable to locate all communications or documents at this time; in the meeting we are busy and at the end we don’t have time to do the follow ups because it is time for the next meeting. Augment does that for you, alongside the apps you know and love.”
If you’ve ever worked with a talented EA or PA, you know how invaluable it is to have such information at your fingertips – and it’s largely a matter of good organization, not deep familiarity with the people or concepts involved. Not everyone has those skills, and it has become more difficult as the tools we use have multiplied and compartmentalised themselves.
“We used to think of the app ecosystem as the solution,” Ritter says. “Now we have tons of point solutions and pages of apps.”
One solution is to do everything on one service, or a few that integrate very closely with each other. Nice if you don’t mind being completely at the mercy of Google, Microsoft, Apple or Salesforce. “Or you can use us as a bridge between services and use whatever calendar you want,” he continued.
Augment’s CTO Dan Cintra, with a resume that includes Google and Axon, described what the product is supposed to do for a person who is often unprepared or completely unaware of what’s going to happen next. (As you might be able to see, I’m one of these people. In fact, I was late to the Augment meeting for no good reason.) “What we’re solving is context,” Cintra said.
He showed a sample of what popped up before their meeting with me: contact information, recent discussions in email or other apps, information pulled from whatever database I’m in (normally this is LinkedIn and so on), as well as soft information such as the topics I normally cover, some personal details, and so on. It contained documents sent between us in relevant conversations, and if we had had previous meetings there would be links to those recordings and summaries. After the meeting, you’ll get action items tagged to people, a transcript and summary, and other follow-ups.
This was all rendered via a native app that appears before and after meetings, but browsers can also be “extended”, with a browser overlay where the AI pops up with information where appropriate – on a calendar item, an invitation to a meeting , or next to a name in an email thread.
Here’s a video of Augment’s first Augments in action:
None of this is brought in via APIs with Gmail, Zoom or anything else. It’s all collected and organized by Augment’s agent and collected on their own systems.
“Because of where we are in the stack, we stream the data as it comes,” Cintra says. And where exactly is that? In a very privileged position, obviously, since the agent has access to your browser, sound in and out, etc. That’s necessary for it to work, but it’s not just screen scraping or anything hacked together.
“The main approach is through accessibility and tools — you can think of it as an automatic Evernote,” suggested Ritter. I replied that anyone working in IT or security probably couldn’t hear him because of the alarm bells going off in their heads. After all, it is a single point of failure that collects data from every service you use.
Ritter acknowledged that there may have been some skepticism, but they were careful from the get-go to build for security and privacy, achieve SOC 2 certification, and make sure users own their data from top to bottom. It may take companies a while to get to grips with this level of meta-organization, but he pointed out that it also took years for them to become familiar with Dropbox, iPhones, and other now must-have technology. For now, they’re targeting individual prosumers, perhaps freelancers juggling multiple clients.
Augment is coming out of stealth today after raising a $3.5 million starting round led by Flying Fish and JAZZ Venture Partners, with the participation of Incisive Ventures and the Allen Institute for AI’s Incubator (which I’ve discussed before).