Introducing Insight Streams, the Autonomous Analysts for Real-Time Customer Insights.

Today, I'm excited to share with you the latest innovation from Pathlight – Insight Streams. As many of you know, we've been at the forefront of Large Language Model-based Conversation Intelligence (CI). With Insight Streams, we’ve built a cutting-edge Generative Agent system that acts as an always-on team of research analysts who distill the thousands of insights we uncover into executive-level trends. As leaders, it’s our job to make rapid decisions with the information we have available to us. Insight Streams allows us to make better decisions, faster, with complete knowledge of what our customers are saying.

Starting with the customer: you don’t know what you don’t know

Earlier this year, Pathlight made waves when we launched the first Conversation Intelligence platform built natively on Large Language Models (LLMs), and in the span of a few months began processing thousands of customer interactions for leading companies every day.

Our hypothesis, now confirmed, was that customer conversations are the most valuable untapped source of data an enterprise has. They contain insights that could transform any business if only they could unlock them at scale. We saw that LLMs were the breakthrough the industry had been waiting for to unlock this data and built Conversation Intelligence to be the first platform to natively harness this profound technology.

We’ve seen customers set up Pathlight in a few hours and immediately start detecting and resolving important issues related to churn, conversion rates, agent performance, and compliance that had previously gone unnoticed for months.

“Within 24 hours of implementing Pathlight Conversation Intelligence, we discovered an opportunity to improve agent performance and quickly took action. The decisive measures we implemented, guided by Pathlight's insights, led to a double-digit percent increase in our QA scores.” Ken Shields, NativePath

But every single customer had the same request, which went something like this, “Pathlight is uncovering incredible insights from each conversation, but the amount of interesting data it is producing is almost overwhelming. We need help making sense of it all and distilling the data into the most important and urgent trends.”

We heard the same thing from CEOs. They told us, “When the company was smaller I had a complete understanding of what customers were telling us, and I was able to make rapid, unilateral decisions with confidence. As we’ve scaled, I’ve lost that connection.”

Leaders aren’t just yearning for a real-time, intelligent executive summary, they also have more specific, and often more urgent, questions that they need answers to, like:

• “Why are our customers canceling their subscriptions?”
• “Why aren’t our customers purchasing more product X or promotion Y?”
• “What product or supply chain issues are most deeply impacting customers?”

They could see CI pulling out these answers at the conversation level, but what they really needed was to be able to instantly hire a team of analysts and have them sprint 24/7 on reviewing every conversation with these questions in mind. For obvious reasons, this isn’t possible.

To address these burning needs, we built Insight Streams

Each Insight Stream acts as an “autonomous analyst”; you give it a specific research objective and it will go off on its own, review every customer conversation, and come back with a distilled analysis of the key insights related to the objective. Each Stream is fully transparent. You can drill down from analysis to themes to sub-themes all the way down to single conversations.

For example, let’s say you are launching a new product or promotion and want to understand how customers are responding to it. In Pathlight, you would simply create a new Insight Stream and give it a brief objective like, “Analyze how customers are responding to our new ABC product.” The Insight Stream would then begin looking for conversations that involve “Product ABC.” As it discovers product feedback, it creates categories of feedback and then distills those categories into multiple levels of themes. Finally, it summarizes those themes into an executive summary that answers the original objective.

We named this product Streams because this analysis is dynamic, not static. It is constantly being updated and adjusted as more conversations are streamed in. It’s running 24/7 in the background, reviewing every conversation.

Under the hood: AI Agents

Insight Streams are built on the latest research into AI Agents. Zapier has a good definition of the concept:

“Right now, interacting with AI follows a familiar ritual. You enter a prompt, and the AI model calculates a response based on the input. Every time you want a new output, you have to provide a prompt. There's always a human to start the process.

AI agents work in a different way. They're designed to think and act independently. The only thing you have to provide is a goal—be that researching competitors or buying a pizza. They'll generate a task list and get to work, relying on feedback from the environment and their own internal monologue. It's as if the AI agents can prompt themselves, constantly evolving and adapting to achieve their objective in the best way possible.”

Each Insight Stream is an AI Agent with custom instructions, complete access to its “environment” (the corpus of conversations), the ability to react and adjust to incoming stimuli, and remember its past actions and justifications. For more details, keep an eye out for a technical deep dive from our engineering team in the near future.

The future is autonomous: next steps for Insight Streams

In conclusion, Insight Streams is not just another feature; it's a monumental step forward in empowering executives with the intelligent tools they need to understand their customer base and make well-informed decisions swiftly. As exciting as this innovation is, it's just the beginning. We are already working on future versions of Insight Streams where AI Agents will become even more autonomous, capable of not only conducting detailed analyses but also of initiating new Streams based on emerging trends or opportunities. Imagine a future where higher-level agents, acting on their own accord, spin up new Insight Streams and ping you only when groundbreaking insights emerge that are worth your immediate attention, or even take direct action in your systems based on customer feedback. This is the frontier of decision-making in the enterprise world, and we're thrilled to be paving the way. Thank you for being part of this journey with us.

‍Insight Streams is now available to customers. To learn more, read our press release here or request a demo here.