Kai: An AI Guide for Labour Relations Software

Two HR colleagues reviewing documents at a laptop with linked knowledge cards
Kai helps teams find Graylark's maintained product guidance as they work.

Labour and employee relations teams work across complex processes: preparing change proposals, managing consultation, tracking engagement and keeping work moving across countries and stakeholders. Software can bring that work together, but people still need to learn where to start and how each part of the platform fits the job they are doing.

That is why we built Kai. It is an AI-powered guide inside Graylark LRM, designed to help people use the platform. We also use Kai internally to answer product questions and improve the guidance behind those answers.

AI help while using Change Proposals

On Graylark's Change Proposals page, a colleague can ask a simple question: “how do I use this?” Kai can answer with steps from the Change Proposals guide, starting with how to open and orient yourself in the module. A colleague can ask the question without leaving their work to search a manual or find someone who remembers the right menu.

Kai can use the person's role, the modules available to them and the route of the page where they opened the chat. That context helps it select the right product guide. If the question is about the current page, Kai can give that page's guide priority; if a module is unavailable to the person, its guide is left out of the answer search. It can also suggest links to documented destinations in the platform.

Page awareness means knowing which part of Graylark is open. Kai does not inspect the proposal rows, field values or other business records visible on that page. It cannot tell someone what a particular proposal says simply because the proposal is on screen. That distinction matters: Kai is a guide to using Graylark, not an analyst silently reading the user's workspace.

Role-aware guidance for HR and employee relations teams

The same words can mean different things to an Employee Relations colleague and an employee representative. Their starting points, available modules and next steps may differ. Kai uses that role and module context to search the appropriate guidance, so a short question can still lead to a relevant explanation.

People can ask follow-up questions in ordinary language. When the documented answer includes a destination, Kai can offer a route into the platform; those suggested links are checked against known destinations and the person's available modules. If the guides do not establish a procedure, Kai is designed to say what it cannot confirm rather than invent one.

For HR and labour relations leaders, the practical promise is straightforward: new colleagues can learn a workflow in the place they use it, experienced colleagues can recover an infrequent step, and specialist teams can spend less time repeating basic navigation advice. The guidance can also stay consistent across offices because it comes from maintained product documentation. Those are the outcomes we are designing for, rather than a claim of measured time savings.

How Kai fits into Graylark LRM

Kai's source material is Graylark's platform guides. As the product changes, we can rebuild its searchable knowledge base from those guides and test whether it retrieves the right sections for real product questions. The answer is grounded in that maintained material, with a clear path for improving the documentation when it falls short.

Graylark also has AI features for work with organisational information. Those are distinct from Kai's product-guidance role. A question about how to use Change Proposals belongs with Kai; analysis of a proposal's contents needs the appropriate business-data workflow and controls. Keeping those jobs clear makes it easier for teams to understand what the assistant knows and when human review is needed.

The result is a more approachable platform for complex work: help that meets people where they are, knows the part of Graylark they are using, and stays within the guidance it can support. For the engineering behind Kai's retrieval pipeline, read Building Kai: an AI knowledge agent with CAG and RAG. To see Kai in the context of your team's work, contact us about a Graylark demo.