Custom, Secure AI
Without Engineering

Josh leads a sales team who sell insurance products. They support prospects through phone, email and social media enquiries.

Across his 70 person team there’s always a range of knowledge, experience and skill level, as new hires are trained, and new products are introduced. 

Josh’s goal is to have the knowledge and experience of his best people at the fingertips of his entire team. That means being able to automatically respond to tickets about any product or sales question on the spot, without having to waste time manually retyping those answers. 

The knowledge and expertise of Josh’s team is a hard won asset that has taken years to develop. It gives them a real advantage in a competitive marketplace.

He’s been impressed with the potential of ChatGPT but sharing his team's wisdom with a public AI Chat model, poses a real security risk. Josh is also working with a limited budget. He doesn’t have the engineering resources to dedicate to a pricey experiment. He doesn’t want to start a risky development project where the costs might spiral out of his control. And he doesn’t want to try and build this himself. 

Using Tennr, Josh was able to quickly build a custom AI assistant that could intelligently generate inbound replies to sales tickets his team received with complete knowledge of their products. A tool specifically focused on understanding his insurance products, his customers, and his tried and tested sales methods, inside and out. This custom model, trained specifically on what has worked best in the past, ensures that any member of his team can automatically send out the perfect message to prospects, every single time.

Dan leads an operations team that processes tens of thousands of documents each month. They have an OCR system that automatically reads common forms and files, but rely on manual efforts to dig through lengthier nuanced documents that required generating detailed answers and next steps.

Dan's goal was to use AI to augment his team to move through tickets faster, with fewer mistakes. He was learned about generative AI through ChatGPT, but when he tried to ask ChatGPT the way his team processes each document, he found it tedious and error-prone. Furthermore, unlike humans, ChatGPT had no way to automatically know what to look for based on a given document.

By training, fine-tuning and monitoring through Tennr, he was able to develop a custom large language model (AI) that was trained to accomplish this exact task. By embedding directly into his team's current workflow, he didn't need to get engineering involved or take a risk on a lengthy product cycle.
‍
His ops team is now able to focus on less tedious and time-consuming client-service tasks.