An analyst needs to update a job spec, so she downloads the old version from the shared drive, opens a blank Word doc, and starts pulling salary data from three different PDFs. She reformats everything to match the template, emails it out for approval, and waits. Two weeks later she does the same thing for the next one.

The software she uses stores information, but it doesn't help move it. That's a system of record. Holly is a system of action - software that doesn't just hold your class specifications but rewrites them, benchmarks them, routes them for approval, and tracks every change. It's the difference between a filing cabinet and a flight deck. One stores information, the other helps you do the work.

That difference matters more now than it used to. Government agencies are facing a staffing crisis they can't admin their way out of, and when updating a job spec takes months, you lose candidates. Holly compresses that timeline by doing the mechanical work so analysts can focus on the judgment calls.

Why systems of record aren't enough anymore

Systems of record were built for a different era - when the goal was just getting documents out of filing cabinets and into databases. That was progress. But storing information isn't enough anymore. Government agencies don't need better archives; they need software that helps them act.‍

That's the shift to systems of action. Bessemer and Tidemark are writing about it now because they've started noticing the pattern across enterprise software. Holly is already there, moving agencies from questions to answers:

"Where's the spec for Administrative Analyst III?" becomes "Here's a modernized draft - ready for review." 

"Who has the latest version?" becomes "Here's the full revision history with every change tracked." 

"What are other agencies paying for this role?" becomes "Here's the market median across 12 comparable agencies - and a button to update your salary range."

"Did the department head approve this?" becomes "Approved yesterday at 2:47pm. Now routing to the Personnel Board."

The old way required HR analysts to be the glue - downloading documents, chasing approvals via email, manually pulling salary data from PDFs, copying and pasting between systems. Holly eliminates that coordination tax.

Modernizing specifications - one or five hundred

Classification libraries drift over time. Job specs written five years ago describe roles that no longer exist, language gets inconsistent across related positions, and minimum qualifications stop matching the labor market. Most agencies know their specs need updating - they just don't have the bandwidth to do it.‍

Holly works at both scales. When an analyst needs to modernize a single classification, they don't start from a blank Word doc. Holly generates a draft that pulls from your existing library, applies consistent formatting, and updates language to current standards. The analyst's job shifts from drafting to reviewing.

When a county needs to modernize 500 classifications, reviewing them one-by-one is a multi-year project. Holly's bulk scanning runs AI-powered analysis across your entire library in an afternoon, flagging outdated language, inconsistencies between job families, and specs that haven't been touched in years.

Snohomish County, Washington, manages over 1,100 job classifications for 3,500+ employees. Before Holly, the team couldn't systematically review that library - "It's just impossible to do," said Lindsey Jones, HR Business Partner for Classification & Compensation. When the county hired five consultants for a $120,000 airport classification study covering 50 specs, the results came back inconsistent - some consultants rewrote entire descriptions, others changed a few words. Holly cleaned up and standardized the output so the team could actually present it. Now, Lindsey's team creates four new, fully researched job classifications in a single day - work that previously took over a week per classification. Classification analysis worksheets dropped from four weeks to one or two days.

The analyst's expertise goes into the judgment calls - does this spec capture what the role has become? Are the qualifications right for today's labor market? Holly handles the rest.

Salary benchmarking: from data collection to decision-making

Here's how salary benchmarking typically works: An analyst identifies comparable jurisdictions, requests salary data, compiles everything into a spreadsheet, calculates medians and percentiles, presents findings to decision-makers. Weeks later, someone acts on it.

"It's very manual and it's very incumbent upon those people responding to us," says Wendy Ross, HR Director at Yavapai County, Arizona. "Just the difficulty in trying to compare apples to oranges, but also in trying to gather the information up in the first place."

Holly collapses that workflow. When you open a class specification, Holly automatically pulls benchmarking data from comparable agencies. You see market medians, percentile rankings, and how your current salary range stacks up - right there, in context. And when you're ready to adjust, you make the change in the same place you saw the data.

The dispatcher test: Yavapai needed to benchmark their dispatchers - but dispatchers don't always go by "dispatcher." Different agencies use different titles, different levels, different minimum qualifications. Manually gathering and normalizing that data across counties, cities, and fire districts would have taken days. Holly pulled the information, matched job descriptions, and showed how Yavapai's compensation compared to the market.

"The aha moment is when the system shows you the percentage of match," Wendy says. "I can lay two job descriptions next to each other and say they sort of match. But I can't calculate a percentage in my head. Holly can and made it immediately visible to us."

No exporting to Excel. No follow-up meeting to discuss findings. Decision and data stay together.

There's also the credibility factor. When it comes to pay, people get emotional. "They attach their personal value to what they're being paid," Wendy says. "It's nice to have a way to take that out. This is factual information, statistics and data. There's no emotion involved." Having an outside tool helps when elected officials or department heads push back - it's harder to say "HR just doesn't want to give our people a raise" when the analysis is right there.

Why this matters now

A new generation of HR professionals is entering government - and they won't tolerate the old ways. They've grown up with software that anticipates what they need. They're already using AI in their personal lives. They know that compiling data in spreadsheets and chasing approvals via email isn't how work has to be.

AI has crossed a threshold. It can draft a job spec, match positions across agencies, and flag inconsistencies in a classification library - tasks that used to take hours. Holly is built for this moment. The agencies that move now won't just be more efficient - they'll attract talent that refuses to work with outdated tools.

The shift from record to action

For government HR, class specifications are the control point - upstream of hiring, compensation, and workforce planning. An agency that manages specifications in Holly doesn't just have document storage. They have the orchestration layer for how classifications get created, reviewed, approved, and maintained.

We're building the operating system for government HR. The agencies that move first will set the standard. The rest will spend years catching up.