Introducing Lindy: Your AI Assistant

Today, we’re excited to reveal Lindy, an AI assistant designed to manage your daily tasks, giving you the freedom to focus on what matters. 

Lindy’s skills are general, but she particularly excels at tasks like organizing your calendar and email, coordinating travel bookings, finding potential customers or new team members, and summarizing content from videos, blog posts, and podcasts, making your daily life more efficient and saving you precious time.

In this post, we'll introduce Lindy's capabilities, outline our principles for building the product, and share our long-term vision for enabling individuals to harness the power of software.

What Lindy Does

Perfect Scheduling

Lindy puts order in your calendar. If you ask her to “help you find 30min with Alice,” she's clever enough to figure out who you're referring to by looking at your contacts, emails and calendar, and, if still unsure, to simply ask you for more information. 

Lindy can handle both internal and external meetings, finding common availabilities, sending options to external parties, and managing any back-and-forth communication, while always keeping you in control.

She can also automatically detect and resolve scheduling conflicts — for example, if you double booked a 1:1 on top of a board meeting, she’ll infer that the board meeting is probably most important, and offer you to reschedule the 1:1 to a time that works for both parties. 

These workflows illustrate two of the constitutional principles we describe below: high agency and comfort with ambiguity.

Recurring Workflows

Lindy can handle recurring workflows for you, triggered by events like new emails or calendar invites. 

For example, you can ask Lindy to send you a note 5 min before each meeting, with the Zoom link to the meeting and details around the attendees and your last interactions. Lindy will pull up your last few email exchanges and meeting recordings with these attendees, and send you a summary in Slack:

Tame Your Email Inbox

Lindy can triage your emails, categorizing them as FYI, urgent, or archive, depending on your preferences. 

She can even pre-draft responses in your voice, tailored to each recipient, so you wake up to an inbox with replies pre-drafted, which you can review, edit as needed, and send.

Omni-Channel Communication

One of the key features of Lindy will be her ability to communicate with you through any channel you prefer. Whether it's through your favorite messaging app, email, or even voice commands, Lindy will meet you where you are. 

Record and Summarize your Meetings

Meetings is one of the channels supported by Lindy. She’s able to join them on Zoom, Google Meet or Teams, record them, and summarize them.

You can then even ask questions to Lindy against your corpus of meetings, or get her to perform tasks based on their content, like updating a “Next Steps” field in Salesforce.

A Fast Learner

Lindy quickly adapts to your preferences, communication style, and working habits. 

By observing your interactions and receiving your feedback, Lindy continually refines her understanding of your needs and preferences, ensuring that her assistance becomes increasingly personalized and efficient. 

Other Capabilities

Besides managing your calendar and email, Lindy will be able to help you with:

Lindy’s Constitutional Principles

At the heart of Lindy are seven constitutional principles, all laddering up to one core goal: to save you time.

  1. Reliability.
    Lindy is exceptionally reliable and attentive to detail — she doesn’t screw up.
    In practice, this means we’ve built many layers of guardrails: we constantly monitor both Lindy’s perplexity when she’s about to do something, and the potential impact of getting that action wrong, taking both factors into account to decide whether to move forward or to wait for your confirmation.
  2. Privacy.
    The word “secretary” comes from “secret” — this is core to Lindy’s job, who respects your privacy and will never sell or divulge your information to third parties.
  3. Personalization.
    Lindy learns from both your explicit and implicit preferences, fitting into your existing tools and workflows.
  4. Proactivity and High Agency.
    Lindy takes initiative, proactively offering solutions and anticipating your needs.

    This always comes after Reliability: Lindy only takes initiative when she’s highly confident, or when the stakes are low.

    For example, she may reach out about your wedding anniversary, offering to send flowers and make a restaurant reservation.
  5. Embracing Ambiguity.
    Lindy's can handle ambiguous requests, figuring out what you mean and acting accordingly.

    For example, if you ask her to "find 30 min with Bob," Lindy will check things like your calendar or email to identify Bob, and only ask for more details if needed.
  6. Speed.
    Lindy is fast and able to learn quickly from your feedback.
  7. Invisibility.
    Lindy aims to be invisible, seamlessly fitting into your life and existing workflows.

Looking Further: Unleashing Software for Everyone

German monk Martin Luther set out to change the relationship between people and God, advocating for a direct connection between the two without the need for the church as an intermediary.

We are now in a pre-Lutherian era for computing, relying on a clergy class of software engineers to access the immense power of software. 

AI assistants like Lindy will usher in a Lutherian revolution for computing, by giving everyone a direct relationship with software. 

When you ask Lindy to book a meeting or send an email, it writes code to execute those tasks — in effect, you’re getting a custom app just for the problem at hand. Sometimes, Lindy will even build a custom UI to help you, using high-level building blocks at her disposal:

For now, these apps are simple. But Lindy will soon be able to handle much more complex tasks. Eventually, we envision a world where you can describe intricate workflows and get sophisticated apps to automate them. This will unleash the power of software for everyone, enabling you to harness the full potential of computing without the need for technical expertise.

This article was co-written with GPT-4. Thanks to Geoffrey Lit for reading a draft.