Overview
This presentation asked a practical question: what does AI-powered utility data actually change for the engineers and agencies who have to design around what’s buried?
Not in theory, but on a live project with a hard deadline, a site nobody had mapped, and a client who needed a cost estimate before anyone could put boots on the ground.
Two perspectives shaped the discussion: a utility engineer with nearly three decades of transportation design experience, and the AI utility mapping platform his firm has used for close to four years. Together they pointed to the same conclusion:
AI utility mapping does not replace subsurface utility engineering. It moves the start line — so the engineering begins on day one instead of month four.
The second lesson was equally important. The costliest utility problems are not the ones found in planning. They are the ones found by an excavator. Every unknown that survives into construction becomes change orders, RFIs, delay claims and, potentially, liquidated damages.
The less you know early, the more you pay late.
Key Takeaways
Utility risk is mostly an information problem: inaccurate records, and worse, unknown records.
811 is not the whole picture. Some municipal water and wastewater systems are not in the 811 system at all.
Traditional Quality Level C mapping can take two to four months — most of it waiting on, finding and interpreting records.
AI utility mapping compressed a 150-acre feasibility study to four days — delivered one day ahead of deadline, with three costed servicing options.
AI-mapped data is not yet an ASCE 38 quality level. It is a tool, like CAD, and it still requires engineering review.
Spot-check, then feed corrections back. The platform improves as practitioners report what they find in the field.
Move utility coordination earlier. VRX has shifted it from the 60% design stage into the schematic phase.
Construction is where utility surprises cost the most — and where the audience said they most often encounter them.
Presentation One
Utility Risk Starts With What You Don’t Know
Jon C. King, P.E. — Vice President, Engineering Manager, VRX, Inc.
What utility risk management actually is
Jon King defined it plainly: identify the risks on a project, then take deliberate steps to reduce them. The goal is safety for property, equipment and — above all — the people on the job. Reducing uncertainty about subsurface infrastructure also reduces contingency cost.
The more we know about what’s underground, and where it is, the better the project — and the more constructible it is for our clients and the contractor.
— Jon C. King, P.E.
The four-step mitigation loop
| Step | What happens |
|---|---|
| 1. Identify | Build a task list of everything that could be a risk to the project or to the contractor. |
| 2. Evaluate | Review existing controls and processes against those risks. |
| 3. Plan | Define an action plan to reduce each risk. |
| 4. Verify the data | Confirm the underlying utility data is accurate. The design must be data-driven and verifiable in the field. |
The most common risk at project start
Inaccurate data — or, more importantly, unknown data. His examples from North Texas:
- 811 gaps. Some municipalities do not register their water and wastewater systems.
- Old infrastructure, older drawings. The Dallas–Fort Worth area still has water lines installed in the 1930s and cast-iron gas lines from the 1940s and 1950s.
- Drift between the drawing and the ground. Property has been acquired, lines rerouted, and plans never updated.
The traditional path: SUE quality levels
| Level | What it involves | In practice |
|---|---|---|
| D | Records research | Call utility owners, submit 811 requests, request as-builts and compile results. The lowest level. |
| C | Records plus surface survey | Survey visible valves, manholes, handholes, poles and abandoned structures, then correlate them with records. |
| B / A | Field designation and verification | Higher-confidence field work to locate and, ultimately, expose utilities. |
The problem is time. A records request can take weeks to return in a usable format. The data then has to be interpreted, checked against the current right-of-way and matched to the project footprint.
To get it into this type of plan is going to take roughly two to four months. That’s a lot of man-hours, a lot of field time, a lot of coordination.
— Jon C. King, P.E.
What changed with AI utility mapping
- Time cut by more than half. What used to take about four months now takes weeks, or even days.
- Near-instant foundation data, exportable to CAD or GIS and overlaid on aerial imagery.
- A faster on-ramp to Level B and A work, because verification starts from a complete working map.
- Safer, office-first discovery. Features surface in the platform before crews walk streets.
Presentation Two
Inside the AI Conflation Engine
Nissa Guerra — Enterprise Account Executive, 4M Analytics
Nissa Guerra positioned 4M as the first utility AI mapping platform — a “Google Maps for the subsurface” covering utilities both above and below ground. The company was founded in 2019 and is headquartered in Austin, Texas.
Who uses it
| Customer group | Examples |
|---|---|
| AEC consultant firms | Engineering firms such as VRX |
| DOTs and municipalities | Public agencies planning and delivering projects |
| Utility owners | Public and private water, wastewater, transmission, distribution and transportation organizations |
How the engine works
| Stage | What it does | Output |
|---|---|---|
| 1. Collect and digitize | Gathers GIS databases, as-builts, GeoPDFs, permits and imagery | Digitized source records |
| 2. Conflate | Removes duplication and attaches attributes and metadata | A cleaner, unified utility map |
| 3. Detect | Machine learning identifies hydrants, valves, poles and manholes | Clickable detected objects |
| 4. Read the paint | Identifies and classifies 811 locate marks | Paint-mark evidence with visual date |
| 5. Build the network | Assembles connected line work within each utility sector | Sector-by-sector utility networks |
A deliberate design choice: the platform does not rely on requesting data from utility owners. Doing so, she noted, would leave 4M in the same position as the rest of the industry — waiting to hear back.
The platform, live
- Define the area — draw a linear project with a buffer, or freehand a polygon.
- Instant line work — utilities within the boundary, broken down by sector.
- Object detection — the sample project returned close to 3,000 surface objects in one to two hours.
- Attributes on click — owner, installation date, dimension and material where available.
- Project analysis and export — owner, crossing and parcel information in GIS and CAD formats.
At the time of the webinar, 4M reported coverage in 19 states and targeted nationwide coverage by the end of December 2026. Confirm current coverage with 4M before publication.
Presentation Three
Case Study: A 150-Acre Feasibility Study in Four Days
Jon C. King, P.E. — VRX, Inc., with Nissa Guerra — 4M Analytics
The driver
VRX was serving on the program/construction management team for development at George Bush Intercontinental Airport in Houston. A client needed utility routing scenarios and a cost estimate for a planned development on the airport’s north side.
The problem
- A 150-acre, largely undeveloped site with no known utility information.
- Adjacent to FM 1960, a seven-lane TxDOT roadway.
- A five-day deadline from project inception to feasibility estimate.
- A remote site where the old approach would have consumed most of the week.
What the mapping showed
| Utility | Finding | Implication |
|---|---|---|
| Power | Main lines across FM 1960; upgraded overhead power on a side street | The side street became a candidate western feed |
| Water | Potable water near the site and another line on the side street | Candidate connections for the development |
| Sewer | None along FM 1960; sewer on the side street | Proposed tap points located there |
| Gas | Multiple large midstream lines; no distribution gas | Three crossings and a separate gas solution required |
| Telecom | Multiple fiber providers across FM 1960 | Provider options available; crossing required |
The outcome
VRX delivered one day ahead of the deadline: three servicing options for power, water and sewer; natural gas and fiber options; and costed routing, relocation and new-utility requirements.
It goes back to risk management. The more we know about utilities, the more we can budget for them — in the fee proposal and in the construction contingency.
— Jon C. King, P.E.
Field Guide
The practical guidance that emerged across the presentations and Q&A, consolidated. Print this section.
Before you design — utility pre-scoping checklist
- Map the corridor before scoping the fee. Unknown utilities belong in the budget and contingency, not in change orders.
- Don’t treat 811 as complete. Confirm whether municipal water and wastewater systems are registered.
- Check the right-of-way on every as-built against current right-of-way.
- Flag records that may be analog only and budget time for retrieval.
- Make record requests specific to avoid irrelevant plans and repeated emails.
- Identify midstream and transmission lines early.
- List every highway, pipeline and rail crossing — and the permits each triggers.
- Start utility coordination at schematic, not at 60% design.
- Decide the target quality level for each segment and where AI-mapped data will feed it.
- Record the source of each line and object relied upon.
Where AI-powered utility data fits
| Question | Answer from this session |
|---|---|
| Is AI-mapped data an ASCE 38 quality level? | Not currently. Industry discussion about how to classify it is ongoing. |
| What is it, then? | A tool — comparable to AutoCAD or MicroStation — that accelerates records research and surface-feature discovery. |
| Does it carry an accuracy rating? | No accuracy level is currently attributed. Source indication is shown where available. |
| Does it need review? | Yes. Engineering judgment determines whether data aligns with right-of-way and expected line locations. |
| Does it replace field verification? | No. It makes Level B and A work faster to plan and better targeted. |
A phase-by-phase playbook
| Phase | How AI utility data was used | Value |
|---|---|---|
| Feasibility | Instant map, routing scenarios and costs | Four-day turnaround on 150 acres |
| Schematic | Early route feasibility check | Kills unworkable routes before detailed design |
| Field planning | Overlay bore-hole coordinates before mobilizing | QC check alongside 811 |
| Design | Foundation for Level C and targeted B/A work | Months of records work compressed |
| Construction | Quick reference after a strike or discovery | Faster conflict resolution |
Verification practices
- Spot-check every project.
- Report discrepancies back so the engine improves.
- Expect abandoned utilities; each can take hours to confirm as inactive.
- Keep engineering in the loop. Never send an AI map out as “all utilities.”
Questions to ask any AI utility data provider
- Which sources feed the line work, and is the source shown per feature?
- Is any confidence or accuracy level attributed?
- Which states or regions are covered today?
- How long does object detection take?
- Which export formats support your GIS and CAD environments?
- How are field corrections incorporated?
- Which integrations exist with tools your team already uses?
A crawl–walk–run roadmap
Run AI mapping on one pursuit or feasibility project. Compare it with traditional records research.
Make it a standard schematic step and a QC overlay before field crews mobilize.
Build it into risk registers and fee proposals. Track time and cost saved.
Audience Polls
Results from the live webinar polls. Utility collection and project-phase questions allowed multiple selections, so their percentages do not total 100%.
Who joined the session
| Question | Answer | % of votes |
|---|---|---|
| Organization type | Local Government | 44% |
| Industry | 28% | |
| Other | 17% | |
| State Government | 6% | |
| Academia | 6% | |
| Location | United States | 100% |
| Business sector | Public Works / Utilities | 44% |
| Other | 28% | |
| Public Information | 11% | |
| Land / Public Administration / Planning | 6% | |
| Municipality population | Over 100,000 | 44% |
| Under 25,000 | 17% | |
| 25,000–50,000 | 11% | |
| 50,000–100,000 | 11% |
How attendees collect utility data
| Method | % of votes |
|---|---|
| Rely on utility owners for records | 64% |
| Use private utility locating vendors or subcontractors | 27% |
| Contract Subsurface Utility Engineering specialists | 27% |
| File an 811 Call Before You Dig ticket | 27% |
| Other | 18% |
| All of the above | 9% |
Where attendees encounter utility issues
| Project phase | % of votes |
|---|---|
| Planning | 43% |
| Design | 43% |
| Construction | 43% |
| Procurement | 14% |
| Maintenance | 14% |
| All of the above | 14% |
Panel reaction: Jon King called the construction result “the most alarming” — a conflict found in construction means equipment on site, lane closures, rental costs and a contract clock running. Nissa Guerra said catching unknowns at the lowest-cost point is the core problem 4M exists to solve.
Full Q&A
Questions came from the live audience and the moderator.
Q. How accurate is this data compared to actual GPS or survey data?
Nissa Guerra: 4M does not attribute accuracy levels to the data. The goal is to provide utility information so firms can do their own due diligence and bring it up to whichever quality level the project requires.
Jon C. King: In SUE terms, AI utility mapping is not currently classified as any quality level. It is a tool, like AutoCAD or MicroStation. Engineering judgment must assess alignment, right-of-way and expected locations before the data is issued.
Q. How complete is it? Are missed utilities fed back to 4M?
Jon C. King: Yes. VRX spot-checks every project. When something does not match field observations, the team reports it. More practitioner feedback helps the engine improve.
Nissa Guerra: Progress on comprehensiveness and accuracy has come from close partnership with firms, DOTs and the wider industry. Customers report catching issues early that would otherwise ripple forward.
Q. How has early AI analysis aided planning for on-site verification?
Jon C. King: It is a planning tool and QC check. Before geotechnical crews mobilize, bore-hole coordinates can be overlaid on utility data alongside 811 results. At schematic level, it gives an early read on risk and route feasibility.
Q. How is the data compiled — existing records, or 100% AI from imagery?
Nissa Guerra: Both. The conflation engine collects mostly public GIS, as-builts and permits, then digitizes and reconciles them. Imagery from multiple vendors is also processed through machine learning to detect objects such as manholes and valves.
Q. Does the data include an AI-estimated veracity or confidence level?
Nissa Guerra: Not currently. Where possible, line work identifies its source. 4M is working toward greater transparency around the confidence it can attribute.
Q. How has your utility research process changed?
Jon C. King: Speed. The old process could mean driving to a municipal GIS office, searching blueprint vaults and index cards, receiving dozens of irrelevant plans, and repeating requests. Now a new project can be mapped within about two hours.
Q. What upcoming AI capabilities do you see?
Nissa Guerra: The next focus is industry workflow and integration, so utility data flows directly into tools teams already use throughout the construction lifecycle.
Q. What advice do you have for utility owners, DOTs and municipalities?
Jon C. King: Invest in utility research early. VRX once began utility coordination at 60% design; it now starts during schematic design because relocations consume substantial money and time.
I don’t ever feel confident giving a contractor a utility plan that doesn’t have some kind of utilities identified on it. It’s just a major risk.
— Jon C. King, P.E.
Greg Babinski: Moving from months to days is anecdotally a major saving. A formal return-on-investment or benefit–cost study would strengthen the case for firms and agencies.
Glossary
- SUE
- Subsurface Utility Engineering. The engineering discipline of investigating, mapping and managing risk from existing underground utilities.
- ASCE 38-22
- The American Society of Civil Engineers standard guideline for investigating and documenting existing utilities, which defines quality levels A through D.
- Quality Level D (QL-D)
- Information from existing records and verbal recollections. The lowest level.
- Quality Level C (QL-C)
- QL-D records correlated with surveyed, visible above-ground features such as valves, manholes and poles.
- Quality Level B (QL-B)
- Utilities designated in the field using geophysical methods such as electromagnetic locating or ground-penetrating radar.
- Quality Level A (QL-A)
- Utilities physically exposed, typically by test holes or vacuum excavation, and precisely surveyed. The highest level.
- 811
- The national “call before you dig” number that notifies member utilities to mark buried lines before excavation.
- Locate / paint marks
- Color-coded marks sprayed on the ground by locators to show the approximate location of buried utilities.
- AI conflation engine
- 4M’s process for collecting, digitizing, deduplicating and attributing utility records from many sources into one map.
- Conflation
- Merging overlapping datasets into a single, reconciled version.
- Object detection
- Machine learning that identifies specific features, such as hydrants or poles, in imagery.
- As-built
- A drawing intended to show what was actually constructed, which may differ from the design.
- GeoPDF
- A PDF with embedded geographic coordinates.
- KMZ
- A compressed Google Earth file format, often used for quick overlays.
- Midstream pipeline
- A large pipeline moving oil or gas between production and distribution. It cannot serve an individual site.
- Distribution line
- A smaller utility line that delivers service to end users.
- Utility coordination
- The process of working with utility owners to identify, avoid or relocate utilities affected by a project.
- Utility conflict
- A point where a proposed design and an existing utility cannot both remain as they are.
- Abandoned utility
- A line or conduit no longer in service but still in the ground, often undocumented.
- Schematic phase
- The early design stage when alignments and layouts are set, before detailed design.
- Feasibility estimate
- An early cost and constructability assessment used to decide whether and how to proceed.
- RFI
- Request for Information. A formal contractor question during construction.
- Change order
- A formal modification to a construction contract’s scope, cost or time.
- Liquidated damages
- Pre-agreed daily charges assessed when a contract is not completed on time.
- Contingency
- Budget held in reserve for unknown conditions.
- PMCM
- Program Management / Construction Management.
- Benefit–cost analysis (BCA)
- A structured comparison of a program’s monetized benefits against its costs.
Speakers
Greg Babinski, GISP
Moderator · Founder, GIS Management Academy
greg@gismcs.comGreg Babinski is the founder of GIS Management Consulting Services and the GIS Management Academy. He is a past president of URISA, now the Geospatial Professional Network, and an American Geographical Society Ethical GEO Fellow. His background includes nearly a decade at a California water and wastewater utility, and a long-standing focus on GIS return on investment.
Jon C. King, P.E.
Vice President, Engineering Manager, VRX, Inc.
j.king@vrxglobal.comJon C. King brings more than 28 years of transportation civil engineering experience across roadway design, rehabilitation, bridge work, storm sewer, wastewater and utility coordination. His contribution centered on utility risk management, SUE quality levels and the IAH feasibility case study.
Nissa Guerra
Enterprise Account Executive, 4M Analytics
nissa.guerra@4m-a.comNissa Guerra works with AEC consultants, DOTs, municipalities and utility owners to bring AI-powered utility mapping into early project phases. She explained the conflation engine, led the platform demonstration and addressed data sourcing, confidence and the roadmap.
About the Organizations
4M Analytics
Presentation sponsor. Founded in 2019 and headquartered in Austin, 4M describes itself as the first utility AI mapping platform. Its conflation engine combines digitized public records with machine-learning object detection. Product coverage and capabilities are subject to change.
VRX, Inc.
A woman-owned engineering firm headquartered in Plano, Texas, providing civil and environmental engineering, construction management and inspection, maintenance management, wastewater and aviation services.
GIS Management Academy
Founded by Greg Babinski to advance professional GIS management practice.
Geospatial Professional Network
Formerly URISA, rebranded to center the working geospatial professional.
Resources
From this presentation
Watch the presentation archiveWatch AI-Powered Utility Data in Practice.Request speaker presentationsAsk the event team about slides and supporting materials.2026 Mapping the Hidden City SeriesExplore upcoming and archived presentations on subsurface utilities and infrastructure data.Organizations & tools referenced
4M AnalyticsAI-powered utility mapping platform.VRX, Inc.Civil and environmental engineering, construction management and utility engineering.Geospatial Professional NetworkProfessional education and networking for geospatial practitioners.Society for Benefit-Cost AnalysisProfessional society for benefit–cost methods.Standards & guidance referenced
ASCE 38-22Standard Guideline for Investigating and Documenting Existing Utilities.811 — Call Before You DigNational damage-prevention notification system.FHWA Subsurface Utility Engineering ProgramFederal guidance on SUE for highway projects.Continue the Series
Asset Mapping Intelligence brings together practical guidance from public-sector practitioners, engineers and geospatial specialists working across utilities, transportation and infrastructure delivery.
Explore AssetMapping.EventsThank you to 4M Analytics for sponsoring this presentation and the 2026 Mapping the Hidden City series, to VRX, Inc. for contributing its project experience, and to Greg Babinski for moderating.