We just got backed by a16z speedrun.

We just got backed by a16z Speedrun.

Damn. Even typing that out feels surreal.

A year ago, OpenSesame didn’t exist. It was just the two of us sitting in a tiny meeting room on Bloor Street East in Toronto, whiteboarding out something that didn’t make sense to anyone but us.

No product. No funding. No customers. Just a dumb little idea and some conviction behind what we can build to change the next generation of software.

We’ve always been pure builders at heart, and in the middle of last year, we were hungry for something more.

Anthony and I left our last startups searching for something different.
We wanted to move faster, build with purpose, and, most importantly, work on something we couldn’t stop thinking about.

We wanted to work on something we couldn’t stop thinking about.
And we wanted to build something for people we actually enjoy selling to.

Today, we’re proud to share that we’re backed by a16z speedrun, joined by Aidan Gomez (CEO @Cohere), Kevin Wang (CPO @Braze), Spencer Burke (SVP, Growth @Braze), Daniel Nieto (Seed Investor), and Comma Capital alongside advisors from Merge, Drift and Angel List.

Check our Twitter post below :)

Chapter 1: Intro

OpenSesame actually started as something completely different.

We thought we were building a revolutionary hallucination tool. (lol) We launched the first version in 7 hours and went live with it.

It got some love. People used it. We even had some early paying users, and some pretty good tech that we still use today.

But something didn’t sit right. Every time we spoke to teams, we saw that the market was getting smaller and smaller. Why are we wasting our time going after a shrinking market?

So, we discarded the older idea and decided to start over with something new.

This was a pretty shitty time.

We had overbuilt parts of the product + lost one core team member.

There were a few crumbs of interest, a couple of logos here and there, but ACVs were low, and we weren’t excited about the potential of what could be.

It felt like we were fighting for scraps in a market that didn’t even know if it needed what we had. Worse than that, we weren’t proud of what we were making.

We weren’t waking up with conviction.

So we took a step back, went back to the whiteboard, stopped chasing random ideas workflows, and focused on basics.

Chapter 2: The Pivot

At this moment, we were exploring the concept of interfaces. Trying to deeply understand how newer AI interfaces were changing the way people work across software. This was motivated by interfaces like artifact, canvas, and, to some extent, perplexity.

We began with the idea that people should be able to interact with their tools using natural language, so we developed an algorithm that allows users to enter a prompt and generate an agent in under a minute, which they can then export as an endpoint to use in their product.

But what about the non-technical people? That’s when we moved away from endpoints and nodes and focused on a single interface that could be embedded into a SaaS product to make it AI native without changing the existing infrastructure.

These criteria led us to choose management software, specifically construction management, workforce management, and financial management software, as our initial focus. We took a page out of DoorDash’s book and chose to focus on the lucrative suburbs instead of the cities.

We asked the same set of questions to all the people we spoke to

• What’s stopping you from adding AI to your product?

• Do your users know what your product can actually do?

• How much are you spending on training new users?

• Do your customers want AI?

• Are you willing to spend 500k+ to hire a team of AI engineers to build and maintain this new AI infrastructure?

Most people didn’t have clear answers.

But they wanted the outcome, without burning resources across their team.

So we chose to build our first version of what we call the Language User Interface (LUI), one that understands your product, communicates with your APIs, and enables users to get work done on your Saas product using natural language.

That’s what we built.
We call it Cell.

It’s an embeddable agent that helps legacy SaaS companies become AI native quickly.

1. Upload your endpoints.

2. Add context and data.

3. Customize the look.

It takes minutes to go live.

And the impact is wild.

Users get work done faster. They discover features they never used before. Product teams get a signal on what people want. And companies unlock net-new revenue without spending 6+ months building AI-native features from scratch.

LinkedIn Launch Video

Chapter 3: a16z

We never actually pitched Andreessen. We ended up meeting a GP at @a16z in New York when flying over for Tech Week. Entering the office in Soho was wild. We both looked at each other, paused, grabbed the free canned water (twice), and said, “What?”

We spoke to the partner, explained our thesis, and outlined why we believe we will win in the long run. After that, the conversation faded into “sick. We think you guys are cool and would love to follow what you build.”

Later that night, we were compelled to do something more, so we emailed the partner saying, Hey, do you think we’d be a good fit for speedrun? (Got introduced to another partner)

Luckily, Anthony was in town for an extra day, so he ended up going back to the office to meet them in person. From there. We got on a call. Talked about what we were seeing, where this interface shift was going, and what it meant to be AI-native. We also did NOT send a deck; we just sent over our Manifesto (that we wrote down the night before in our Airbnb)

Chapter 4: Thesis

We think the GUI had a good run. But it’s starting to show its age.

We believe the future of work lies in natural language.
We think the next generation of software will run on LUIs.

Interfaces that talk to you.
Interfaces that do the work.

We’ve written a whole manifesto on what makes a good LUI.
But if we had to boil it down to five things:

1. It should be faster than the GUI

2. It should be accurate and repeatable

3. It should feel native to the product

4. And also feel personal to the user

5. And finally, it should let you do all your work, end-to-end

That’s what we’re building with Cell.

We aim to become the Canva for AI interfaces.
A simple, beautiful way to embed natural language into any product.

Chapter 5: Now

We’re still a tiny team.

Just a group of different people who think product interfaces are about to change forever. If this future makes sense to you, if you want to build, not just bet, come talk to us.

Or if you’re working on a product that should be AI-native but don’t want to spend months building it from scratch… we can get you set up with your very own LUI in a couple of minutes.

Trust us, we’re just getting started.

— A & J