tutch - agentic services
by Tutch Pty Ltd
tutch - agentic services enable retailers to use the power of AI without all the heavy lifting
The tutch Agentic AI is an AI platform based on the the principles of Generative AI and is created for retailers to configure AI agents (called tutch Assistants) that can engage with a customer in a conversational manner across their shopping journey.
The objective is to use customer’s engagement context to understand a customer’s needs and engage with them on any channel (website, mobile app, or in-store kiosk) to offer options that that satisfy those needs – some examples of such needs include “designing a wedding outfit”, or “designing a trip”, or ”remodeling a room”.
While the first implementation of the platform is designed to be used in the retail industry, the application can be across industries where intent based conversations can lead to better results for customers and lower the cost of businesses via agent-based automation.
A retailer has the option to create unique AI agents (Assistants) that fit their needs regarding digital engagement priorities and what they want to offer in terms of market/sell as products. These AI agents from tutch can be made available across any of their channels.
With the number of large language models (LLMs) available in the market, it is often challenging to identify the right LLM that fits the need. The tutch Agentic AI platform can select the right LLM to solve its needs. As an example, an AI Agent needing to seem trends in the market so that it can identify products may need a different LLM, than the AI Agent recommending a visual design for a home improvement. Similarly, and AI Agent helping a customer visualize a complete outfit may require a very specialized model (known as small language models pr SLM) in combination with a general-purpose trends LLM model.
The tutch Agentic AI platform also has built in choreography and orchestration that allows for several AI Agents to interact with each other to help achieve a customer’s goal. As an example, an AI agent configured to help a customer with a specific task, will interact with other AI Agents (also configured and re-used) through the platform’s orchestration layer.
The AI agents that live within the AI agentic AI platform have the following key capabilities:
Automatically identify and configure itself with the skills required to complete the current ‘task’ providing the customer what the need
Automatically develop questions (also known as prompts) using advanced techniques such as self-reflection to discover customer’s intent.
Recognize patterns using natural language skills within customer’s responses and develop insights about what a customer is seeking
Look for actions that can be performed on behalf of the customer to support the objectives (e.g. look for alternative products or compare pricing)
Enhance the customer journey experience through automated actions (e.g. notify a store staff member when a high loyalty customer is in store)
The following are technical design specifics that make up the tutch Agent AI platform
Dynamic skill resolution, question generation, ai-routing, multi-hop querying and reasoning based on conversational context
Dynamic LLM allocation (both model and hosting) per specific agent task per client profile
LLM & RAG ( Response Augmentation ) text to touch translation to support visual engagement
Augmented search and inference engine with pluggable recommendation algorithms