Overview of Our Journey
At the recent YC Alumni reunion, I encountered a few fellow entrepreneurs who expressed surprise at how drastically our business direction has shifted over the years.
This article recounts how my co-founder, Aaron Villalpando, and I have weathered 12 major pivots while maintaining our partnership intact.
Our current focus revolves around BAML, a brand-new programming language tailored to capitalize on large language models (LLMs).
After 3.5 years of experimentation, we’re finally experiencing a burgeoning user base that genuinely values what we offer.
The First 1.5 Years: A Phase of Discovery
This initial stretch was marked by our approach of experimenting with several ventures while relying on stable income sources.
We hesitated to fully commit, often waiting for validation before making any intense investments.
Throughout this time, we explored various ideas, including:
A six-month stint running an online coding bootcamp funded by income-share agreements.
Although our intention was to inspire social change, we only attracted 15 users and soon realized it wasn’t a sustainable business model.
A six-month initiative with interactive Twitch ads meant to monetize our efforts.
Despite speaking with numerous streamers, interest dwindled to just one response.
After a tough Y Combinator interview where a key figure questioned our knowledge of Twitch, we shifted gears.
A brief six-week venture attempting to create an SDK for multiplayer games, inspired by smash hits like “Among Us.” Although we secured a test with a single streamer, a lack of gaming expertise led to yet another pivot after Aaron left his position at Amazon.
A four-week exploration that allowed fans to invest in creators’ future earnings.
With Aaron’s previous YouTube experience, we thought we had an advantage, but the economic feasibility proved insurmountable.
A quick three-week project focused on employee engagement surveys to improve team dynamics.
Despite discussions with human resources executives, the traction we sought eluded us, prompting Aaron to question the rationale behind his recent career shift.
Facing Challenges and Transforming Ideas
Throughout these early attempts, we struggled with fluctuating objectives and a reluctance to charge users.
The projects often felt misaligned with our passions.
However, we acquired invaluable lessons about the need for rapid pivots and established a strong foundation of trust between us.
In one of our early challenging conversations regarding equity, we agreed on a fair compromise where I initially contributed funds to a joint account until I transitioned to working full-time, leading to a 50-50 equity split.
Our commitment to Gloo, a voice-first competitor to platforms like Slack and Discord, devoured 11 months of our time.
Despite the enthusiasm we poured into Gloo—logging over 1,000 hours each—our analysis revealed a lack of product-market fit.
Reflecting on our Gloo experience:
We aimed to revolutionize remote team communication, but our offering did not resonate with the intended audience.
When we entered Y Combinator Winter 2023, we hit a significant revenue milestone, but feedback urged us to pivot.
Many companies had faced similar failures, yet our attachment to Gloo held us back.
Conversations on February 10, 2023, with co-founders of a comparable platform laid bare the reality that, despite substantial funding and users, a scalable market remained elusive for them.
This realization hit hard as we noticed our own user base dwindling, shattering the confidence we had built.
Revealing BAML: A Fresh Programming Language
One evening on August 10, 2023, we began brainstorming a syntax for a new programming language.
By that Sunday, we had developed a compiler in C++, and by Monday, we started transitioning our existing customers to this new library.
Despite underestimating the challenges ahead, we encountered a pivotal moment when one customer requested a shift back to Python.
After thoughtful deliberation, we forged ahead with a trial period, gathering feedback and refining our product over the next three months.
By March, we welcomed about ten new customers and enhanced the syntax for easier accessibility, ensuring a smooth transition for our existing clients.
As time went on, we noticed a compelling pattern: once users engaged with BAML, they remained committed, often inquiring about potential new features.
We rolled out a series of improvements, including evals and support for various LLMs, reaching an important milestone when we allowed compatibility with any existing programming language in under a minute.
One bright spot at this year’s YC reunion came when unfamiliar faces recognized me as “the BAML guy,” eager to share their excitement about coding in BAML during their travels.
While we still have a long road ahead with BAML, we feel optimistic that we have entered a stable phase, leaving behind the relentless pivoting.
Our next goal is to engage 1,000 active developers weekly on the BAML platform.
If you’re curious about BAML, check out this YouTube video showcasing the coding experience with BAML.
Documentation is available at https://docs.boundaryml.com.
For developers interested in creating their own language, visit our GitHub repository at http://github.com/boundaryml/baml.
Vaibhav Gupta is a co-founder of BoundaryML, dedicated to developing an expressive language for structured text generation.
Source: Producthunt