Prosodica Innovation & Technology Culture

Our data scientists rapidly develop new models, but their agility is only possible due to the highly abstract and careful architectural work that enables our platform to seamlessly interact with diverse types of models. Combining these distinct cognitive strengths has made our organization much more resilient and adaptable.  

What is the unique story that you feel your company has with AI? If you were writing about it, what would the title of your blog be?

Our unique AI journey began more than a dozen years ago, well before AI was prominently featured in commercial products. We turned to AI out of necessity, driven by a desire to solve repetitive, tedious problems without resorting to the kind of brute-force human effort that could drain the joy out of our work. We simply couldn’t accept the idea of hiring talented individuals only to burden them with monotonous tasks, while the truly interesting challenges awaited attention. Instead, we harnessed AI as a willing collaborator to tackle these mundane problems, enabling our team to engage in stimulating, meaningful work. We took inspiration from Tom Sawyer, inviting AI models as eager “friends” to help paint the proverbial fence, keeping our team’s efforts focused on innovation and creativity.

What was a monumental moment for your team when it comes to your work with AI? 

For us, as for many others, the monumental turning point was the release of OpenAI’s GPT-3 in 2020. Already deeply involved in machine learning, the sudden availability of this powerful large language model opened up entirely new horizons in our ability to understand and interact with human language. This breakthrough presented an exciting yet challenging moment: we had invested considerable resources into developing specialized, bespoke models for understanding caller intent. GPT-3 offered similar capabilities, with greater flexibility but higher costs. Anticipating that costs would inevitably decrease, we boldly embraced LLMs to position ourselves at the forefront of language-driven AI innovation.

What challenges did your team overcome in AI adoption?

As a company rooted in software engineering and data science, our challenges with AI adoption weren’t typical. Instead of resistance, we faced overwhelming enthusiasm, rapidly integrating AI into our processes. Initially, we believed our in-house, domain-specific models would outperform general-purpose large language models. Yet each quarter, generative AI advances quickly outpaced our custom solutions, prompting us to adopt these models while staying focused on customer needs and preserving the intellectually stimulating aspects of our work. 

Today, as AI encroaches on tasks we once saw as uniquely ours, we’re experiencing the same challenges we’ve long helped customers navigate. It’s a demanding but thrilling landscape, rich with opportunities for growth. And as generative AI evolves, we find ourselves once again blending general models with finely tuned, domain-specific solutions — creating uniquely powerful results.

On Prosodica's Auto-Evalutation Product:

"Prosodica built Auto-Evaluation to address a core problem in contact centers: Human quality assurance teams cannot realistically review enough conversations to provide fair, consistent feedback at scale."