Leobit
AI Sales Assistant Powered by Anthropic’s Claude
Custom AI agent development for Tipco, a U.S.-based industrial distributor and engineering solutions
ABOUT the project
- Client:
- Distributor fluid conveyance and sealing solutions
- Location:
-
USA
- Company Size:
- 500+ Employees
- Industry:
-
E-Commerce
- Solution:
- Custom AI software
Services:
Technologies:
Leobit helps Tipco, a U.S.-based retailer of industrial hoses, build a smart AI agent powered by Anthropic’s Claude AI. The solution can quickly retrieve data from a large catalog of the company’s items and provide suggestions relying on Tipco’s knowledge base. Claude also supports the system’s analytical capabilities.
Leobit’s work has been released on time and within budget. The team has been highly engaged and effective, providing constant communication. They have spent time learning the client’s business and offer solutions to help build the best product.
Customer
Tipco is a US-based provider of fluid conveyance and sealing solutions, specializing in hose assemblies, custom gaskets, and products for industrial, hydraulic, and high-purity applications. Rooted in the 19th century, the company now serves customers across multiple industries through a nationwide network of locations and strategic partnerships. They are running a comprehensive e-commerce platform providing diverse categories of products from multiple manufacturers.
Business Challenge
Tipco was handling a growing volume of customer requests and needed a tool to manage that load more effectively. With a large database of assemblies, the goal was to make that data usable — giving employees and customers a way to build AI-validated assemblies.
Why Leobit
The customer had previous successful experience of cooperation with Leobit and was satisfied with our deep technical expertise and responsible development approach. They were also aware of our deep expertise in developing custom AI solutions across industries, which was crucial to the project’s success.
Project
in detail
The project is ongoing and involves multiple phases, as we support the customer all along the way from planning to execution.
Our BAs worked with the client to define project requirements and map out the AI agent’s core logic. One of the primary challenges was establishing a reliable data foundation for the AI models. Tipco maintained a large catalog of product data from multiple manufacturers. We organized data and trained the model on different datasets in phases instead trying to cover all the client’s data at once. This enabled us to establish more reliable foundations for an AI agent.
Following the research phase, the team developed a proof of concept for the AI agent. Our specialists integrated and configured Claude’s Sonnet model to retrieve product information from Tipco’s knowledge base, providing valuable support to the client’s regional managers. Once the PoC was released internally, it was piloted with regional directors, whose structured feedback was collected and analyzed to identify priorities for the next development phase.
We are currently optimizing the AI agent and preparing it for a full-scale release for users. We plan to expand the product by integrating it with the website, enabling Tipco’s end-customers to interact with a smart assistant while navigating its pages.
E-Commerce AI Assistant
We built a smart AI-powered assistant that helps the customer’s regional managers handle end- client requests more efficiently. Powered with Claude’s Sonnet model, it helps them navigate the product catalog and quickly advise customers on hose and fitting types, sizes, male/female fittings, and related specifications. Currently, the workflow looks as follows:
- An end-client contacts a regional manager with a product request
- A regional manager forwards this request to an AI assistant
- The assistant processes the request, retrieves relevant data from the database via tool use (function calling), and provides product recommendations that match the customer’s requirements
To ensure a more efficient data foundation for the agent’s operations, we retrieved information from various manufacturers’ websites and fed it into the model. Over time, we managed to partially automate this workflow. Leobit specialists are now preparing a website integration, so end-clients will also be able to submit requests directly through the assistant.
Analytical Capabilities
The AI agent uses Sonnet’s analytical capabilities to process product data alongside user requests and sales patterns to deliver insights relevant to business forecasting and commercial analysis. In particular, the solution can identify the most popular product configurations across regions, capture emerging trends, and surface correlations between seasonal demand shifts and specific item categories. As a result, regional managers and business analysts get a data-driven foundation for inventory planning and sales strategy.
Technology Solutions
- An AI sales assistant powered by Anthropic’s Claude AI that processes requests and provides product recommendations by retrieving item data from the customer’s database through tool use.
- Integration and custom configuration of Anthropic’s Sonnet model.
- Strong data foundation for the AI agent’s operations.
Value Delivered
- Comprehensive support across the full project lifecycle, including project management, business analysis, software development, QA and testing, and support.
- An AI PoC that showed excellent performance during user acceptance testing.
- Ongoing product development, with upcoming stages including a full product release and public availability through website integration.
- Strong technical support and a proactive approach to the client’s technology initiatives over an extended engagement period, currently exceeding five months and ongoing.
