How sewer inspection data becomes a capital plan you can defend

Macomb County’s public works team had a recommendation in hand: make these repairs, and make them now. Something about it didn’t sit right. They had a limited budget and no room for a wrong call, so before they spent a dollar, they ran the inspection data through a second, AI-assisted read.

What came back changed the decision. Vincent Astorino, Operations Director at Macomb County Public Works, said the reassessment “saved us over a million dollars in repairs previously recommended as immediate.” The same 24-hour analysis also surfaced critical defects the first pass had missed.

The lesson isn’t that one assessment was wrong. It’s that the gap between collecting inspection data and turning it into a funded, defensible plan is exactly where good programs lose time and money. Inspection is the part the industry has mostly solved. The plan is the hard part.

Here is the path the strongest programs follow to get from a hard drive full of CCTV footage to a multi-year capital plan a board will approve, and an honest look at where the slow steps actually are.

A plan is only as defensible as the coding under it

Every capital decision traces back to how a single defect was coded. If one crew calls a condition a grade 2 and another crew calls the same thing a grade 4, the plan built on top inherits that inconsistency, and so does every budget meeting that follows. Standardized PACP coding isn’t paperwork. It’s the floor the whole plan stands on.

This is why coding accuracy matters more than it looks. Houston, the nation’s largest collection utility, measured AI-assisted coding at roughly eight times the accuracy of contractor averages. That consistency is what let the city trust the data feeding a multi-billion-dollar program instead of relitigating it segment by segment.

The takeaway: before you rank anything, make the coding consistent. A defensible plan starts with data you would be comfortable defending line by line.

Turn condition into risk, not just severity

A grade 5 defect on a quiet residential stub is not the same priority as a grade 4 running under a hospital. Severity is a property of the pipe. Risk is severity weighed against consequence: what fails, who it affects, how hard it is to reach, what it costs if you wait.

This is the translation that used to take a consultant and a quarter. Risk & Rehab does it continuously, turning coded inspection data into a live priority list that a director can read without GIS or modeling expertise. The ranking updates as new inspections come in, so the list reflects the system as it is, not as it was at the last study.

The takeaway: rank by risk, not by the worst photo. A board funds consequences avoided, not defect counts.

Bundle defects into projects, not pipe by pipe

Boards don’t fund defects. They fund projects. A list of 4,000 flagged segments is not a plan; it’s a spreadsheet that ends a meeting early. The real translation step is grouping segments into the work crews and contractors will actually bid and build: spot repairs, lining runs, manhole rehab, basin-level packages.

When Phoenix ran its small-diameter assessment program, AI-verified data across more than 181,000 feet sorted into clear work types, roughly 38% needing CIPP lining and 2% needing point repairs. That is a plan a contractor can price and a council can follow. And it cuts both ways: the contractors doing the lining and the point repairs are working from the same data, so better targeting means their crews show up where the work actually is, not where a stale guess sent them.

Smart Project Builder handles this grouping, connecting risk-scored data to the project packages that become the line items in a capital plan. The takeaway: the unit of a capital plan is a project, not a pipe.

Put real costs against it before anyone asks

A plan without numbers is a wish. Each project type carries its own unit costs: cleaning, traffic control, bypass pumping, lining, excavation, design, contingency. The programs that clear budget review are the ones that attached those costs at the project level early, while the data was fresh, rather than scrambling to defend a lump sum later.

The takeaway: the version of this plan that survives a budget meeting is the one that already answered “how much” before the question was asked.

Sequence it into a multi-year plan the board can read

Group the funded work into horizons: urgent failures in years zero to two, high-priority rehab in years three to five, monitored assets beyond that. Each line should say what, where, why, how much, and what risk it retires.

Houston’s program is the clearest example of this at scale, with inspection data informing how billions are spent across a 15-year consent-decree timeline. Greg Eyerly, Senior Assistant Director at Houston Public Works, described why the underlying data mattered for decisions: it is “readily available both in the app & via an API,” which cut internal coding and review hours and made the numbers more precise. Precise numbers are what let a program defend its sequence year after year.

The takeaway: a board approves a story it can follow. Condition, to risk, to project, to cost, to year is that story.

Keep it living, not a one-time report

The plan that ages worst is the static PDF filed after the study. Conditions change, crews re-inspect, priorities shift. Macomb is the proof of the opposite approach: a second read on existing data, done fast, changed a million-dollar decision and caught defects the first pass missed. That only happens when the data is something you can requery, not a report you archive.

The takeaway: treat the capital plan as a living priority list you can ask new questions of. The whole point of digitizing the inspection data was to be able to interrogate it later.

The plan was always in the pipe

None of this is new work for the people doing it. Directors, engineers, and contractors have always moved from condition to decision. What changed is the speed and the defensibility of the translation. The slow, contestable steps — coding consistency, risk ranking, project bundling, costing — are the ones AI now collapses from months into days.

The plan was always sitting in the inspection data. The job is getting it out fast enough, and clearly enough, that the people holding the budget can say yes.

Watch it work end to end. SewerAI’s live demo walks from raw inspection data into a risk-ranked, budget-ready plan using Risk & Rehab and Smart Project Builder.

Watch: From Inspection to Action

Want to see it on your own system’s data? Book a demo.