You've probably heard terms like "AI," "ChatGPT," or "LLM" used interchangeably. Here's what's relevant to your work in Holly.
The kind of AI Holly uses is called a Large Language Model (LLM). It was trained on enormous amounts of text — job descriptions, classification specs, policy documents, MOUs, court opinions, and more. From all of that, it learned the patterns of how people explain, compare, and reason through HR work.
When you give Holly a task (drafting a class spec, comparing a contract against your classifications, or answering a question about a bargaining agreement) the AI doesn't look up a pre-written answer. It generates a response based on the patterns it's learned. Think of it less like a search engine and more like a very well-read analyst who can process hundreds of documents at once.
The most important thing to understand: AI can produce confident-sounding output that is still wrong. Human review is always the final step.
Note: Holly's AI is designed to support your professional judgment, not replace it. Always review AI output before using it in official decisions, reports, or communications.
Where AI Shows Up in Holly
Holly has AI built into several parts of the platform. Depending on the feature, it works a little differently.
Class Spec Writing and Editing When you're writing or updating a class spec, Holly can generate a draft based on your inputs — including the classification series, level, and duties. You can also ask it to revise specific sections. For more information on spec writing with AI, see How to Generate a Specification with AI or How to Rewrite a Class Spec with AI.
Interview Question Generation Holly can generate interview questions tailored to a classification's duties and requirements. In some cases, this runs automatically when you click a button. The questions are a starting point — review and adjust them to match your agency's process.
ChatMOU ChatMOU lets you ask plain-language questions about your bargaining agreements. Instead of searching through an MOU manually, you can ask things like, "How is overtime calculated for employees on a 9/80 schedule?" and get an answer with the relevant contract language cited. For more information on ChatMOU, see ChatMOU: Querying and Comparing MOUs with AI.
Benchmarking and Compensation Analysis Holly uses AI to compare your classifications against market data and peer agency specs, surfacing patterns across large datasets that would take weeks to analyze manually. For more information on AI compensation analysis, see How Benchmarking Works in Holly.
What Makes this Different from a Search or Database
A common question: Isn't this just a fancy search?
No — and the distinction matters. A search returns documents that match your query. An LLM generates a response based on patterns it learned during training. That's why it can draft a full class spec, summarize a complex contract clause, or explain how two job classifications compare — tasks a search engine can't do.
It also means the output isn't pulled from a stored record. It's constructed. Which is why it requires your review. The AI doesn't know your agency's specific history, culture, or the nuances of your labor relationships — you do.
FAQs
Is your agency's data used to train Holly's AI? No. Your agency's data is not used to train the underlying models. Holly processes your data to generate results for your agency only.
Can Holly's AI make employment decisions? No. Holly generates analysis and recommendations. All employment decisions remain with your agency's authorized personnel.
Where does Holly's AI get its information? Holly's AI doesn't search the internet. It works from two sources: the documents and data inside your agency's Holly workspace (class specs, contracts, benchmarking data, and similar inputs), and the patterns it learned during training. It doesn't pull in outside web results or reach into other agencies' data.
What should you tell employees or union representatives about AI in Holly? You can explain that Holly uses AI to help analyze HR data — class specs, contracts, compensation benchmarks — and that all outputs are reviewed and decisions are made by HR staff.
For questions about AI in a specific Holly feature, see the related article for that workflow, or reach out to [email protected].