Write it down before they retire

On 2 December 2020, the human resources manager at Inland Empire Utilities Agency, a municipal water district in Chino, California, gave the board a staffing workshop. The slides are in the public board packet. One is titled "Loss of Knowledge and Experience." It lists nine operations positions, from an Operator III with 15 years to an operations supervisor with 38, and adds them up: 268 years of experience. The next slide has one line: "It takes 5 years to develop an operator's ability to be on-call."

I believe the knowledge in senior staff's heads is the first thing any AI tool at a utility will depend on. A tool can read work orders, SCADA history and the CMMS. It has no way to know why an operator runs a lift station by hand on a stormy night if nobody wrote it down. So my suggestion to utility managers: write it down before they retire, and before the tool arrives.

The numbers behind the slide

IEUA's slides put numbers on its own risk. The agency averaged 9.6 retirements a fiscal year over seven years, with 13 in FY 2019/20. A quarter of its full-time staff were eligible to retire at the time of the workshop, and 41% would be by 2025. The slides list operators, electrical and instrumentation technicians, control systems analysts and lab scientists among the critical positions.

The national figures point the same way. EPA's water workforce page says roughly one-third of the water sector workforce is eligible to retire in the next 10 years. EPA's August 2024 report to Congress, from the interagency water workforce working group, cites Brookings: roughly one-third of water and wastewater treatment plant and system operators were 55 or older in 2021. The Brookings 2018 report, Renewing the Water Workforce, found that 78.2% of water jobs need at least one year of related work experience, and that 10.2% of water workers were under 24, against 12.5% of all US workers.

The same EPA report says what leaves with the people: "The source of water, necessary treatment, equipment idiosyncrasies, water quality issues that may occur seasonally, and the concerns of consumers are all system-specific and take experience to manage." It also notes that the federal infrastructure money, and the work of finding and replacing lead service lines, add demand on the same workforce.

What recent board minutes say

IEUA's workshop is from 2020. Board minutes from the last two years carry the same concern. On 30 July 2025, the commission of the Green Bay Metropolitan Sewerage District (NEW Water) voted unanimously, as Motion #25-057, for a temporary over-hire "to support succession planning efforts." The item gives the reason: "The department is forecasting 1 to 2 Operator retirements within the next 6 to 18 months, with an additional 2 expected in the next 24 months." At the Los Angeles County Sanitation Districts, the District No. 2 minutes of 28 May 2025 say that "In the 1980s, the Districts hired a significant number of Blue Collar staff, who are now retiring." The same minutes say those blue-collar vacancies "have not been challenging to fill." The experience those retirees take with them is harder to replace. In Georgia, an Albany Water, Gas & Light Commission agenda item of 13 March 2025 recommends a contractor for tasks that "COA crews are unable to complete promptly due to staffing shortages. (42% Vacancy)"

We counted how often these topics come up. From 2023 to 2026, 29 of the utility boards we read mentioned succession planning or knowledge transfer in a water context, 11 mentioned operator retirements, and 11 mentioned a staffing shortage or hard-to-fill positions. These are minimums, from the boards we read, and they say nothing about a national rate.

A book about companies, read from a utility

A new book gave me a way to think about this. The Enterprise Brain, by Ragy Thomas, Sravan Vadigepalli and Chandhu Nair (2026), argues for a shared layer that holds what an organization knows, so that people and AI agents can work from it. The water sector has a name for this problem: the silver tsunami. Experienced operators from the baby boomer generation are retiring now, and they take 30 years of operational history with them. The people coming in do not have that history, though they are comfortable with technology. Codifying what the retiring operators know into the utility's brain, and making it available to the agents, lets the new operators take over smoothly. Three ideas from the public material about the book fit a utility well.

The first is about fear. In an advance blurb for the book, Keith Ferrazzi writes that the authors treat AI resistance as an identity problem: people fear AI for what it reveals about their work. At a utility, what a tool reveals is often a workaround. A senior operator has a way of handling a nuisance alarm, or a valve sequence that keeps one pressure zone stable, and it works. It may differ from the written procedure, or there may be no written procedure. Ask that operator to help "train the AI" and it can sound like an audit. Ask them to write down how they do it, with their name on the record as the person who knows, and it sounds like credit. Both requests capture the same knowledge. I would expect the second to get more of it.

I don't have a case of my own to tell here. I have heard it many times, in conversations with utilities, in interviews and at industry events, and the board minutes above show the same thing in public.

The second comes from the book's description, which promises "the self-scaling enterprise where revenue grows and head count doesn't." That is private-sector language, and it reads badly at a public utility, where staffing is set in a budget the board adopts in public. IEUA's own slide on employee engagement carries the line "The pendulum has swung too far on doing more with less." The version that fits a utility is the retirement gap. The work grows with lead service lines and the capital program, experienced staff leave, and a new operator takes years to train (five years to on-call, by IEUA's count). A tool that helps a newer operator find what a retired one knew covers part of that gap. It needs the knowledge on paper first.

The third is from a Forbes column by Sandy Carter (11 August 2026), where Sravan Vadigepalli describes the larger challenge with AI as organizational: how decisions get made and how knowledge moves across the organization. At a utility, that means someone owns the record. If keeping it current is nobody's job, it goes stale, and an AI tool will repeat the stale version with confidence.

What to do on Monday

Pick one recurring decision. Keep it small and real: the response to a SCADA alarm that operators handle by habit, the valve sequence for isolating a main in one pressure zone, or how the capital plan scoring committee breaks a tie. Then write down five things:

1. What the decision is, in one sentence.
2. Who knows how to make it today, and who the backup is. If there is no backup, write "none".
3. Where it is written, if anywhere. A shift log entry or an old email counts, so note it.
4. The steps, as the person who knows them tells them, and why they do it that way.
5. Who owns the record, and when it will be reviewed next.

Have a newer staff member do the writing, sitting with the person who knows. That pairing is the passing on of system-specific knowledge the EPA report describes, done one decision at a time. Small systems can start from EPA's Knowledge Retention Tool, a free Excel spreadsheet that helps operators "consolidate system information into one location" and is designed for staff transitions.

After one record you will know how long the next one takes, and roughly how many your plant needs before the next retirement party.

For vendors

If you are piloting an AI or analytics tool at a utility, the person who checks its output is often a senior operator or engineer close to retirement. Ask at kickoff who that person is, and whether the knowledge your tool depends on is written down.

Where I found this

The IEUA slides came from a public board packet. Reading board packets, capital plans and minutes in the public domain is what we do every day at AquaIntel, and workforce comes up in them alongside the pipes and plants. If your utility has a staffing workshop or a knowledge transfer plan in its own public record, I would like to read it.

Sources

• Inland Empire Utilities Agency, Regular Workshop of the Board of Directors, 2 Dec 2020, agenda and item 1A "Agency Staffing Update" (presented by the Manager of Human Resources): ieua.org. Slide 6 (retirements per fiscal year 13/14 to 19/20: 8, 5, 5, 13, 13, 10, 13; average 9.6; 13 in FY 2019/20; chart values checked against a render of the page), slides 7 and 8 (25% of FTEs eligible to retire; 41% eligible by 2025), slide 9 (critical positions), slide 14 ("The pendulum has swung too far on doing more with less"), slide 22 ("Loss of Knowledge and Experience": nine positions, 15 to 38 years, 268 years total), slide 23 ("It takes 5 years to develop an operator's ability to be on-call").

• US EPA, Water Infrastructure Sector Workforce: epa.gov. ("roughly one-third of the water sector workforce eligible to retire in the next 10 years").

• US EPA, Interagency Water Workforce Working Group Report to Congress, EPA 810-R-24-004, August 2024: epa.gov. p. 3 (roughly one-third of operators 55 or older in 2021, citing Kane, Brookings, 2022); p. 11 ("The source of water, necessary treatment, equipment idiosyncrasies..." quote); pp. 11-12 (BIL investment and lead service line replacement add demand on the workforce); p. 17 (Knowledge Retention Tool).

• Kane, J. and Tomer, A., Renewing the Water Workforce: Improving Water Infrastructure and Creating a Pipeline to Opportunity, Brookings Metropolitan Policy Program, June 2018: brookings.edu. (78.2% need at least one year of related experience; 10.2% of water workers under 24 vs 12.5% of all workers; 102,520 operators at utilities in 2016 and about 9,200 projected annual openings for water treatment operators, 2016-2026, used in Figure 2 only).

• US EPA, Interactive Tools for Owners and Operators: epa.gov. (Knowledge Retention Tool Spreadsheet for Small Water Systems: "helps operators consolidate system information into one location", "Designed to assist in personnel transition").

• The Enterprise Brain: Rewiring Your Business for the AI-Native Era (Ragy Thomas, Sravan Vadigepalli, Chandhu Nair, 2026). ("the self-scaling enterprise where revenue grows and head count doesn't").

• NetGalley listing for The Enterprise Brain, advance praise: netgalley.com (Keith Ferrazzi blurb, paraphrased here: the authors describe AI resistance as an identity crisis, and people fear AI for what it reveals).

• Forbes, Sandy Carter, 11 Aug 2026, column with author comments (Vadigepalli: the larger challenge with AI is organizational).

• Green Bay Metropolitan Sewerage District (NEW Water), commission minutes, 30 Jul 2025 (Motion #25-057, temporary over-hire for succession planning; forecast operator retirements)

• Los Angeles County Sanitation Districts, District No. 2 board minutes, 28 May 2025 (blue-collar staff hired in the 1980s now retiring; vacancies "have not been challenging to fill")

• Albany Water, Gas & Light Commission (GA), agenda item, 13 Mar 2025 (contractor approval, "staffing shortages. (42% Vacancy)")

• AquaIntel reading of public board minutes and agendas, 25 Oct 2026: 317 utilities with dated meeting documents in 2023 to 2026 (225, 240, 260 and 278 by year). Strict counts in a water context, 2023/2024/2025/2026: succession planning or knowledge transfer 18/13/12/10 (29 distinct); operator retirements 1/0/7/5 (11 distinct); staffing shortage or hard to fill 5/3/6/3 (11 distinct). Minimum values, of the boards read.