OS Manifesto
Scroll to explore
Preface
User interfaces have drastically evolved over time, moving from the command line, accessible only to technical users, to the graphical user interface, which democratized computing for everyone. That was in 1973, and besides the routine UI trends that come and go, the GUI has remained largely the same.
Thesis
We believe that AI will help usher in a new kind of interface. We call it the language user interface (LUI). We wholeheartedly think that all work will one day be done with natural language, and we want OpenSesame to be the company that makes that possible.
The current state
Although we’re still in the early days of the LUI, we’re already seeing a transition away from the GUI, bolstered by the mass adoption of ChatGPT and its canvas feature, developer-focused companies like Cursor and Bolt, which have redefined the way developers work, and feature releases from Crunchbase, Figma and Canva, which all promise the same thing: letting their users use their software through natural language.
Why isn’t it widespread?
We expect that most, if not all, SaaS companies will want to implement their own LUI, in order to meet customer demands, but are putting it off due to a lack of in-house AI talent, a lack of R&D budget, and a realization that an LUI isn’t a feature they can just ship and forget about, it’s a perpetual commitment to continuous improvements and iterations.
Where does OpenSesame fit in?
We offer a product called Cell, the LUI for SaaS companies.
Right now we’re creating a Canva-like interface to build and deploy AI-native experiences on your product in minutes. Our intention is that you should be able to design beautiful agents that understand your product, connect to your APIs and deploy in your front end without rewiring your current infra.
What a user is able to customize on this interface is our early version of the LUI called Cell. When a user customizes a Cell, it can tap into an application’s underlying APIs to act on the user’s behalf. To create a Cell, all we need is API documentation in the form of openAPI specs or Swagger docs and the product URL. The whole process takes less than 5 minutes.
Instead of spending upwards of $500,000 to hire the necessary talent and wait 8-12 months until the LUI is shipped, companies can get access to Cell on a subscription basis for a fraction of that price.
What needs to be true for Cell to hit critical mass?
We’ve distilled our technical goals down to 6 pillars:
The time to satisfaction must be shorter than the time it would take a user to do the action manually.
Cell must never make any mistakes; a button click always works, so should a query response.
Cell must feel bespoke to the software it is embedded on, all the while…
…Cell must feel bespoke to each user. Everyone works and thinks differently; a good LUI should reflect that.
Cell should be able to communicate with external tools and other agents.
As a company, we need to strive to get our time to value down as much as possible. Customers must instantly see and feel the value of what we’re offering
Each and every one of our product decisions runs through these 6 pillars to ensure that we’re not deviating from the correct path and building for the sake of it.
Past, present, and future
We expect OpenSesame to take up more screen real estate from the host SaaS product over time, ultimately taking up the entire screen, allowing for remote product usage through different modalities and for a viable monetization path for the age of autonomous agents.
Our world-class team of advisors and angels, which includes:
Aidan Gomez (CEO @Cohere)
Andreessen Horowitz (a16z speedrun)
Kevin Wang (CPO @Braze)
Spencer Burke (SVP, Growth @Braze)
Matt Bilotti (Product Lead @Angel List, Employee Number 6 @Drift)
Robert Williams (Head of Sales @Merge)
Comma Capital (NYC)
Daniel Nieto (Principal @BDC Seed Fund)
Parting words
We see the jump from the GUI to the LUI as one of the most significant opportunities unlocked by the democratization of large language models. We believe that a successful implementation of embeddable AI coworkers like Cell will spark a productivity boost on par with the arrival of the GUI and lay the groundwork for AI agent networks on a global scale.
-A & J