A VALUED CUSTOMER OF THIRDEYE DATA
ProChain Solutions: a help centre assistant that keeps itself current
ProChain Solutions builds project management software used by organisations to focus effort on the tasks that actually determine delivery. Their customers ask detailed questions, and the answers sit across product documentation, PDFs, and the company website. Support in a specialist product is largely the work of locating the right answer rather than composing one.
LOCATION
Lake Ridge, Virginia, United States
COMPANY SIZE
Small-scale Enterprise
INDUSTRY
Information Technology › Explore solutions for this industry
GenAI help centre assistant
ThirdEye Data built a generative AI assistant that gives users a straightforward conversational way to get answers, drawing on ProChain’s own material rather than on general knowledge.
The system uses retrieval augmented generation, which matters for accuracy in a specialist product. The assistant answers from the company’s actual documentation rather than from a model’s impression of how project management software generally works, and that distinction is the difference between a support tool and a liability.
Two design choices did most of the work. The index updates automatically when content on the website changes, so the assistant does not slowly drift away from the product as documentation is revised. And it collects user feedback on the answers it gives, which is used to improve the quality and relevance of what comes back rather than sitting in a report nobody reads.
The self-updating index is the part that determines whether a system like this survives. Assistants built on a snapshot are accurate at launch and progressively wrong afterwards, and the people who built them are rarely still watching when that happens.
Delivered and completed. The assistant reindexes automatically as product content changes.
Related solution: generative AI support assistants
A support assistant is judged on one thing, which is whether it is right. A general purpose model answering questions about a specific product will produce plausible answers that are wrong, and in a technical product a customer usually cannot tell the difference until they have acted on it.
ThirdEye Data builds these to answer from the organisation’s own material and to say so, with the retrieval layer doing the real work and the model handling the phrasing. Two things then decide whether the system stays useful: whether the index keeps pace with the content, and whether user feedback actually changes anything. We build both in from the start, because an assistant that was accurate at launch and has not been touched since is the most common version of this project and the least useful. We have built support assistants for software products, help systems for enterprise teams, and document assistants for regulated material.
Explore our generative AI assistant solutions ›
DO YOU HAVE A SIMILAR USE CASE?
Talk to our team to learn more about what we did for ProChain Solutions and how that experience applies to your own environment.
Explore What We Did for Other Customers
