Beyond Coding: How Parrot is Leading the Next Wave of AI-Driven Legal Tech
In 2004, Mark Kislingbury made history with a stenographic speed of 360 wpm at 97% accuracy. This achievement, remarkable in its time, served as inspiration as I delved deeper into the world of software engineering. The power of AI in transforming spoken words into text intrigued me, hinting at the potential beyond manual transcription.
At Parrot, where I spearhead an AI-centric team, we're not just developing; we're pioneering. We're harnessing machine learning to reshape how US legal depositions are digitized. Given the decline of traditional stenographers and our ambitious pursuit of real-time, precise transcription, our tech stack is nothing short of cutting-edge.
However, this innovation demands unique skill sets. Our engineers don't just code; they ideate, understand the complexities of transcription holistically, and preemptively tackle challenges embedded within the model. It's about syncing up tech with linguistic nuances - accents, slang, background disturbances, and more.
Our current metrics? An error rate slashed from almost 30% to 7%, with a goal to rival human accuracy at 4-5%. Achieving this involves a blend of AI transcript review, human refinement, and iterative model training, a perfect playground for tech enthusiasts hungry for challenges.
Technically, we’re exploring uncharted territories, leveraging avant-garde technologies like Argo Workflows and Gantry. Our aim? An evolved speech recognition system, equipped to generate summaries, cross-compare testimonies, and delve deep into the intricacies of Natural Language Processing (NLP).
For the tech-savvy reading this: If you're craving a dynamic, agile environment, where weekly sprints morph ideas into solutions, and where you’re not just coding but innovating, Parrot is where you need to be. Together, let's redefine the limits and make Kislingbury's achievements a legacy, while we script the next big revolution in legal technology.