Skip to main content

Getting the Most Out of Holly's AI

A practical guide to matching Holly's AI features to the right kinds of work — and knowing where your judgment should lead.

Not Every Task is a Good Fit for AI

Holly's AI is built for the parts of government HR work where it genuinely helps: reading and comparing large volumes of documents, finding patterns across classifications, and drafting content that your team then reviews and owns. It's fast, consistent, and doesn't get tired of running the same analysis across 200 class specs.

But AI works differently depending on what you ask of it. Understanding which tasks are a natural fit — and which ones still need your expertise front and center — helps you get more out of Holly and use your own time well.


Where Holly's AI Performs Best

These tasks share a common profile: they're document-heavy, repeatable, and produce output that a human reviews before anything happens.

Classification and compensation work

Comparing classifications against market data and peer agency specs is exactly the kind of work AI handles well. The analysis is structured, the inputs are consistent, and every recommendation is traceable to source data. What used to take weeks of manual comparisons can happen in a fraction of the time — with your team doing the review and making the final call.

Contract compliance screening

AI can read a pending contract and flag where the scope of work overlaps with your classified positions. For California agencies, this is especially useful for AB 339 compliance. The AI surfaces the overlap; you determine what to do with it.

Class spec drafting and updates

Holly can generate a first draft based on the role level, duties, and context you provide. AI is well-suited to this because it has learned the patterns of how government class specs are written — but your knowledge of your agency's specific classifications and culture is what makes the final product accurate.

ChatMOU

Asking questions about bargaining agreements is a strong use case because MOUs are structured documents with consistent patterns. AI can surface the relevant language quickly. That said, always verify that the answer applies to your specific MOU and situation — especially for questions involving eligibility, entitlements, or recent amendments.

Workforce analytics and reporting

Identifying patterns across your workforce data — retirement trends, vacancy clusters, turnover by department — is where AI adds real speed. These reports are inputs to your planning, not decisions themselves.


Where to Stay Closely Involved

Some use cases are valuable, but require more active oversight on your part.

Interview question generation. Holly can generate interview questions from a class spec, and they're often a solid starting point. Review them carefully before use — make sure they reflect your agency's actual selection criteria and comply with your jurisdiction's interview practices.

Hiring and candidate screening. If you're using AI tools to screen or compare candidates (within Holly or elsewhere), human review of every hiring decision is essential. AI can help organize and surface information, but the selection decision belongs to your team. This is also the area most scrutinized from a legal and equity standpoint, so document your process.

Anything involving individual employees. Performance decisions, discipline, grievances, and similar matters require context, judgment, and due process that AI can't reliably provide. Holly doesn't offer automated recommendations in these areas, and we'd recommend being skeptical of any tool that does.


Where AI Isn't the Right Tool

A few categories where the work is better done by people, full stop:

  • Labor negotiations — AI can help you prepare (modeling fiscal scenarios, pulling precedents), but the negotiation itself is a relationship. The dynamics at the table belong to your team.

  • Novel legal questions — AI is trained on existing text. When a new statute or a recent court decision changes the landscape, human expertise leads.

  • Relational and political judgment — Reorgs, difficult employee situations, and leadership transitions require the kind of contextual intelligence that AI doesn't have.

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.

For more information on AI in the Holly platform, see How AI Works in Holly. For questions, reach out to [email protected].

Did this answer your question?