Context.ai Aims to Combine Product Analytics Insights with Large Language Models (LLMs)

Since the launch of ChatGPT late last year, numerous companies have been developing generative AI tools to enhance customer interactions with their products and services. However, many of these vendors still lack insights into the effectiveness of the large language models (LLMs) they utilize and the quality of the responses generated.

This gap inspired Context.ai, which launched earlier this year to equip companies with better insights on user interactions with their LLMs. Today, Context.ai announced a $3.5 million seed investment to further refine its innovative approach.

Co-founders CEO Henry Scott-Green and CTO Alex Gamble, both former Google employees—Scott-Green in product management and Gamble as a software engineer—identified the pressing need for a service that evaluates LLM performance. Their discussions with numerous developers revealed a common struggle: the overwhelming uncertainty regarding how users engage with their models and assess their performance. “The phrase I hear repeatedly is: ‘my model is a black box,’” Scott-Green shared.

Context.ai operates similarly to product analytics tools like Amplitude or Mixpanel, which track user interactions within interfaces by monitoring clicks and page duration. However, Context.ai focuses specifically on analyzing the data produced by LLMs, aiming to determine whether the generated content effectively addresses customer inquiries. The overarching goal is to enhance model efficiency.

The process begins when customers share chat transcripts with Context via API. The platform employs natural language processing (NLP) to meticulously analyze these conversations, categorizing them by topic and assessing each interaction to gauge customer satisfaction with the provided responses.

After processing the chat transcripts, Context.ai presents a comprehensive analysis.

“We believe we’re witnessing a significant transformation with the rise of LLMs, leading to a proliferation of chat experiences in the coming years. As users increasingly engage through text interfaces rather than traditional graphical user interfaces, there’s a pressing need for specialized tools,” Scott-Green explained.

Context.ai started by developing an initial prototype, which has since been iterated upon through feedback from early customers and design partners. Scott-Green notes that product refinement is an ongoing endeavor, and they have already attracted considerable interest, including paying customers.

For those concerned about security and privacy, it’s important to highlight that Context.ai takes measures to remove personally identifiable information (PII) during data ingestion. The platform does not use content for model training or marketing and maintains data for a maximum of 180 days before deletion, according to Scott-Green.

Currently, Context.ai operates with a small team of six, but Scott-Green envisions a future of growth. He emphasizes the importance of cultivating a diverse workforce from the outset.

“Building representative, diverse, and inclusive teams is a challenge within the startup ecosystem and the tech industry overall. Both of us are committed to this mission, actively striving to ensure our employee base reflects inclusivity and diversity,” he stated.

Today's funding round was co-led by GV (Google’s venture arm) and Theory Ventures.