Future Unicorn #225: Mem.ai

The Quild Future Unicorn highlights one future of work early-stage startup with statistically significant signals of becoming a unicorn.

Mem is a productivity platform to streamline tasks and note-taking workflows. The company's platform allows users to quickly capture and share information through its AI-powered, self-organizing workspace.

Founders: Dennis Xu (Co-CEO), Kevin Moody (Co-CEO)

Signals:

1. Top company alumni

• Kevin was a product manager at Google (3+ yrs)

• Dennis was a product manager at Yelp (2+ yrs)

2. Top university alumni

• Kevin graduated from Stanford University

• Dennis graduated from Stanford University

3. Top investors

• a16z led seed

• Unusual Ventures participated in seed

• Floodgate participated in seed

Origins

Mem's origin goes back to Korean fried chicken. In 2014, Dennis and Kevin met at the now-defunct Korean restaurant SO during their summer internships after freshmen year. They talked about how information about us seems to live in silos. Wouldn't it be great if our preferences from Netflix could just be ported over to HBO or Hulu? They dreamed of a "Me API" that captures knowledge that can be ported over to any application. That idea stuck with them over the next years as they graduated and went on to build products at Google and Yelp.

As product managers, they had to synthesize a lot of information. I'm sure that when they were onboarded, both had to learn about how things worked inside their companies, what their teams had done in the past, why decisions were made, and so on. And I'm also sure that they went on an expedition searching through documents spread across several applications and talking to a dozen people. Wouldn't it be great if knowledge is just captured all in one place?

The "Me API", operationalized as a knowledge graph, could only be realized if knowledge is captured. So in 2019, Kevin and Dennis started building Mem to be the most ubiquitous note-taking tool.

Pain point and persona

Mem is solving every knowledge worker's problem of knowledge management, which follows the hierarchy of capture-retrieve-share.

The first step is to capture and generate information. At the basic level, we just need a digital notebook. But fragmentation becomes a problem. The problem of scattered notes in the analog days remains the same in the digital age. I use Mem, Notion, Gmail, paper notebooks, and loose-leaf papers to take notes. This is not just a "me" problem. In organizations I've worked at, both big and small, information is captured across different systems from Evernote to Confluence to a cell inside an Excel file version FFFFF.

After knowledge is captured, retrieval becomes the next challenge. Have you ever tried to search for that one paragraph that is buried in a 20-level deep email thread after trying several keyword search combinations? Yep. Retrieval is a difficult problem because most systems rely on matching keywords. This sucks because we use the same words all the time.

Lastly, when working in a team we share documents. For the most part, it is as easy as sending a link. But working in a team also exacerbates the first two problems of capture and retrieval. Every team member writes notes uniquely in their own preferred tool. This makes searching for and referencing others' work tougher.

Product

Mem.ai approaches these problems differently by being ubiquitous and using the latest in artificial intelligence.

Capturing

Capturing in Mem is intentionally easy. Users can use the web app, desktop app, or iOS app. Users can even text notes to Mem via SMS, WhatsApp and Telegram. They make it as frictionless as possible to take notes inside Mem. One unique aspect of Mem is that I feel less pressured to take detailed notes. I can just write one bullet point and move on to the next note. Mem promises to organize every note so that it is easy to retrieve when searched for. When I create notes in other tools, I feel angst when a page seems empty. That one bullet point might be important but it will likely be buried under all other notes.

Retrieving

Retrieving notes in Mem is done intelligently. As a note gets created, Mem automatically surfaces relevant notes by looking at the text, entities (e.g. people, projects, companies) tagged, dates, links, and attachments. Users can even ask questions in the search bar and Mem will surface the right notes and files. It is building a knowledge graph of your notes behind the scenes just like how Google has a knowledge graph of the web.

Collaborating

As for sharing and collaborating, Mem has the standard features of sharing links, commenting, and a workspace to manage access & permissions. More to come here as Mem shifts its messaging from being a knowledge tool for individuals to a self-organizing workspace for teams.

Creating

With its recent series A announcement led by OpenAI, Mem beta released its Smart Write and Smart Edit features. This is next-level note-taking because it builds a personalized large language model (learn more about these models here) as your writing assistant. Based on the demos, users can instruct Mem to rewrite notes, generate ideas, summarize long-form content, and pull relevant information from different notes in a single command.