• 4149
  • Posts
  • Proactive AIs are here

Proactive AIs are here

What would it feel like to have an AI teammate? 4149 is the world’s first proactive AI teammate.

Happy spring all 🌸 As the flowers bloom and the sun gets ready for the summer season, team 4149 too is emerging from our building slumber, ready to start showcasing the world’s first proactive AI teammate.

Enjoy this weekend read which sets the tone for what’s to come!

- Adrian

What would it feel like to have an AI teammate? No, not an AI assistant that notifies you about your next calendar appointment. Not a search engine that lets you find docs across all your apps. Not even an automator that chains together button clicks. 

To answer this question, we need to drop the “AI” and consider what we expect from a teammate in general. 

We obviously expect them to get along with the rest of the team. A teammate should be honest and even more importantly keep their teammates honest. A teammate should know what matters to the team and understand their priorities. We certainly care about a teammate actually being able to get their work done.

Most importantly though, we expect teammates to be proactive.

What’s the point in bringing on a new person to the team if that person requires your non-stop attention and direction? We want a teammate that takes their own initiative to pitch in – we all know the energizing feeling of a new teammate’s contributions making our work and lives better.

Over the last year we’ve seen AI have its moment. What started as playful chats and copy generation has now entered the realm of generating complex code and lifelike movies. We are living through the exciting transitional phase from discovery to revolution, where new product categories will be defined that transform how we get stuff done.

At the heart of this transformation: AI’s ability to ask itself questions.

What does this have to do with teammates and global transformation? While on the surface this ability might seem inconsequential compared to being able to throw a Dostoevsky novel at an AI and get back a 100-word summary, the consequences of self-inquiry are huge when you combine them with an objective and the right context.

Given an objective, an AI can ask itself the questions needed to make a plan. With a plan, the AI can start using basic tools like web browsers & APIs to progress through the steps. Each step of the way, the AI can ask more questions about whether or not it is getting closer or farther from the objective. We are now no longer dealing with a fancy auto-complete engine, but with something that only needs a goal and it will then do its best to accomplish it.

AIs that can reflect and make decisions unlock opportunities that can exceed our imaginations because they are no longer taking an input and manipulating it into an output of likely tokens. Instead, AIs are starting to plan and reason.

We no longer need to concern ourselves with the precise implementation of how to get the output we want, we can now focus on what to do with the output, and why it matters. 

So what are these objectives? It’s easy to imagine clear-cut examples, like “plan a travel itinerary” or “research a topic”, but can we think grander and set the objective to something like “support our team”? Given the right context – in the case of a work team, this might entail meetings, chats, docs, trackers, etc. – the AI can not only start asking itself the questions needed to figure out what supporting a team means but also can begin proactively taking action.

Here’s a concrete example…

Our team has a launch coming up later this week, which we’ve talked about the last few months during team meetings and across various team chats. As a participant in these meetings and chats, our AI teammate picks up on this upcoming event in its reflections on what the team is working on, connecting its learnings to priorities, blockers, individual deliverables, important dates, company values, shout outs and many (many!) more aspects of “supporting our team”. 

Leading up to the launch, our AI teammate notices the team hasn’t put together a launch plan of contacts to help spread the word, so it proactively decides to start compiling a list of people who the team had noted as fans and supporters in prior chats, meetings & online interactions, and drops a message in Slack to share the list as a Google Sheet.

As the team is busy making sure that everything works in time for the launch, our AI teammate also warns us about a few security questions left unresolved in our discussions of one of the new features we’re launching, and (kindly) calls out one of our human teammates to please get to the bottom of this ASAP. 

At last, it’s time for launch! We receive a message in Slack from our AI teammate:

Hats off to the team, you've made it through,
Your hard work paid off, kudos to you!
Here's to a launch that's stellar and true!

Software now becomes a relationship not a transaction.

We can check off all our expectations for a great teammate:

✅ An AI can get along with the team – even credit and celebrate its human teammates for their wins and contributions.

✅ At the same time, an AI can make sure the work the whole team is taking on actually aligns with our goals, and when it does not, call us out and keep us honest.

✅ Lastly, and again most importantly, given the ability for an AI to plan, it can pitch in and take on the tasks it’s best suited for on its own. 

In other words, an AI teammate is proactive.

This is just a glimpse of what having an AI teammate will be like – a teammate that is always on and focused on the sole purpose of helping elevate their human teammates and our team’s work. 

The value of AI will not be measured in productivity gains. Like all transformative technologies, the value AI unlocks will be in the new product categories it creates and new perspectives it unlocks for humans. 

Here’s the kicker: we’re already working on this! 4149 is the world’s first proactive AI teammate. For sure, more products like this will come out in various flavors and capacities as the field matures – the future will be bursting with autonomous AI. However, you can check out 4149 for yourself right now by joining our beta

It might feel weird at first, but soon you won’t know how you got anything done without an AI teammate. 

— Adrian, Greg & Stephen

Why am I getting this? You’ve either signed up for the 4149 beta or have had a conversation where you asked to learn more about what we are building.