Customer Story
AI for tower operations: why Terrion chose Scout
Terrion runs more than 3,000 wireless sites with a deliberately lean IT organization. Here’s why they chose Scout and what early results look like.
Terrion is a Canadian tower company focused on acquisition, construction, and co-location leasing across a portfolio of more than 3,000 wireless infrastructure sites. Established as an independent towerco in 2025, Terrion operates with a lean IT organization by design, selecting Sitetracker as the operational system of record and Scout as the core agentic AI platform.
Terrion, a Canadian tower company managing more than 3,000 wireless infrastructure sites, was not short on AI options. As a carveout from a major mobile network operator in late 2025, the company inherited deep AI fluency and access to every major model, with AI in use across several parts of the business. The question was which AI to put in the business’s critical path, working directly with the people and systems its tower operations depend on. Terrion selected Scout, Sitetracker’s agentic AI platform, and began deploying it inside Sitetracker and across its operational workflows. The decision came down to where the value actually sits: Scout works on live Sitetracker data from day one, inside the permission model and data structure Terrion’s team already trusts, and it advances on two fronts, with Sitetracker curating the harness and model layers while Terrion’s team deploys the Skills, agents, connectors, and use cases specific to its operations.
We sat down with Michael D’Itri, Terrion’s Head of Digital Innovation and Technology Strategy, to talk through the decision, the early results, and where he sees the value heading next.
Why Terrion needed an off-the-shelf system of record
Most independent tower owners evaluating new technology tend to replace legacy systems or add to an existing tech stack. As a carveout, Terrion could keep what was working and decide where it made sense to buy or build anew.
The business is focused on site acquisition, build-to-suit construction, and co-location leasing to national carriers. D’Itri describes it simply: “Most people would equate Terrion to a tower company. I like to think of us more as a real estate company. We just happen to be leasing space on our towers.”
The carveout created specific constraints. Terrion’s IT organization is deliberately focused. Building fully custom systems from scratch wasn’t realistic, and D’Itri’s mandate was to find a platform that lined up with the business and could be deployed without heavy customization. He evaluated the market and kept coming back to Sitetracker. “The roadmap we had been planning to build ourselves had actually evolved in exactly the same direction,” D’Itri says. “There was a natural fit.”
Sitetracker became Terrion’s operational system of record. And before the team had even finished onboarding users, they were ready to deploy Scout.
What Terrion looked for in AI for tower operations
Terrion’s criteria for evaluating AI were the same ones that decided Sitetracker: off-the-shelf and ready to go rather than built and maintained in-house; integrated closely enough that users don’t swivel-chair between systems; and operating within the security and role structure already in place rather than requiring a second one. The evaluation didn’t start from a position of unfamiliarity. D’Itri spent 12 years heading up business systems in the wireless infrastructure domain. “We didn’t have a shortage of AI tools,” he says. Building its own AI layer was a real option.
“We could have built something ourselves,” D’Itri says. “But it comes back to the same reason we selected Sitetracker: the idea of picking a solution that is curated by the vendor, that sits within and is fully integrated with Sitetracker.”
The reasoning came down to two factors. First, Scout reaches Terrion’s users directly inside Sitetracker, without swivel-chairing between systems, and runs within existing workflows, sometimes executing in the background with results available for user review. Second, a vendor-supported platform approach meant Terrion’s team didn’t have to own ongoing optimizations.
“Doing it on our own would require a lot of effort to keep tweaking prompts, playing with settings, picking the right models,” D’Itri says. “There are new models every week, almost every day. Having that curated for what’s best-in-class to do the job today lets our team members focus on the task at hand.”
“We could have built something ourselves,” D’Itri says. “But it comes back to the same reason we selected Sitetracker: the idea of picking a solution that is curated by the vendor, that sits within and is fully integrated with Sitetracker.”
Scout inside Sitetracker: access, repeatability, and control
Three things follow from Scout operating directly within Sitetracker.
Scout is where the work already happens. Terrion’s team members negotiate leases, build towers, and manage co-location agreements. They aren’t IT specialists, and D’Itri didn’t want them to need prompt engineering skills to get accurate results. Because Scout is available directly in Sitetracker and has direct access to the data, users ask questions in plain language and get answers grounded in the system of record. “Scout filters the context correctly so that our end users can get the insights they need with the least amount of effort,” D’Itri says.
Skills make the work repeatable. Scout Skills define the logic behind how work gets done, so a workflow that’s been configured once runs the same way every time it’s needed. That repeatability turns an individual win into something the broader organization adopts, and it lets the team codify a process without writing and maintaining custom code.
Existing configurations are respected. Tower companies manage confidential lease agreements, tenant information, and site-level operational data. “Security and privacy are among the most important things my team thinks about,” D’Itri says. “We need to make sure the right people have access to the right data at the right time. We also need to ensure that nobody accidentally makes changes to records they shouldn’t.” Because of the secure connection between Scout and Sitetracker, the role-based access controls Terrion already configured carry straight through. “We can jump in, use the platform, and know that it’s not going to allow our users to do anything they’re not already entitled to do,” D’Itri says.
Scout in early use: three workflows at Terrion
Terrion rolled Scout out in stages, validating each use case before moving to the next. A few early wins are worth exploring in more detail:
- Site inspection processing
- Lease document handling
- Operational data validation
EARLY SCOUT USE CASES AT TERRION
| SITE INSPECTION PROCESSING | Scout extracts deficiency data from vendor inspection PDFs and automatically creates prioritized follow-up jobs in Sitetracker. Substantially reduces the time needed to process each report. |
| LEASE DOCUMENT HANDLING | Real estate teams query the lease portfolio in plain language. Scout flagged conflicting expiration dates across multiple lease documents and declined to populate the field, surfacing a data conflict issue for the team to review. |
| OPERATIONAL DATA VALIDATION | Team members query site status, maintenance schedules, and project milestones through Scout’s conversational interface, grounded in live Sitetracker data, delivering insights where users need them. |
How Scout automates tower inspection report processing
Tower site inspections generate vendor deficiency reports as PDFs. Getting that content into structured records has always required someone to read the report and transcribe each line item by hand. As D’Itri puts it, “the truth of the matter is we’re going to always have PDFs in some way, shape, or form, or emails and so on.”
With Scout, Terrion’s team drops the PDF in. Scout extracts every line item, including the full deficiency list and priority classifications, maps those priorities to Sitetracker punch list values, and creates the follow-up jobs required to correct them. No transcription, no manual mapping.
D’Itri expects a substantial efficiency gain in the time it takes to extract value from the document and input it into Sitetracker.
The larger significance, in his view, is what it does for data quality at the source. “That is a huge win when it comes to processing the data, getting good data in, and laying the foundation for not just looking for information or searching with AI, but actually improving the data layer with AI.”
What happened when Terrion asked Scout to process a lease agreement
Terrion’s team uploaded a lease agreement and several related documents and asked Scout to create the corresponding lease record in Sitetracker. A request of about a sentence and a half.
Scout began extracting and populating the record, then stopped short of completing it. When the team asked why, Scout explained that the source documents didn’t agree: one file gave one expiration date, and another gave a different one. Rather than selecting one and populating the field, Scout left it empty and surfaced the conflict.
“Scout was able to bring back to the user: I would love to fill it in, but I don’t know which to pick,” D’Itri says. “It quickly surfaced an insight that the documents aren’t aligned, and that’s important. We need to be able to make sure that all of our legal documents are matching up.”
A tower company’s lease portfolio is its revenue base. An incorrect expiration date entered into the system of record propagates into renewal tracking, revenue forecasting, and tenant negotiations.
Scout caught the discrepancy at the point of entry and handed the decision to a person. In this example, it surfaced the conflict for human resolution rather than guessing. That behavior comes from operational context, an understanding of the systems, the data, and the nature of the work itself. For a tower operator managing thousands of lease agreements, it is worth more than speed.
How Scout validates operational data against the system of record
On their first day, before any formal onboarding, a team member looked for information about a specific site. They entered the site ID into Scout.
Scout checked it against Sitetracker and reported that the ID didn’t exist in Terrion’s records, then asked whether the user was certain it was a Terrion site. It wasn’t, and the team member found out in seconds rather than after hours of work on the wrong record.
“That was already a very quick win,” D’Itri says. “An easy thing that someone with no training was able to do, get to the result much quicker, and stop doing a task they shouldn’t have been doing.”
The value isn’t only speed of retrieval. Scout answers against the system of record, so when a question references something that doesn’t reconcile with Terrion’s data, that gap surfaces immediately instead of propagating into downstream work.
What’s next for Terrion and Scout
D’Itri is clear-eyed about the value AI should deliver in practice. “I have a strong belief that it’s great to be able to document time efficiency, but you’re not going to reduce staff,” he says. “The intent is to give your team more time for high-value-added activities.”
The next phase applies Scout to the commercial side of the business, using it to draw insight from Terrion’s co-location and revenue data and answer questions that would be impractical to answer by hand across a portfolio of this size.
“Things like opportunities to improve revenue visibility and billing accuracy, where there’s so much data you can’t possibly figure it out manually,” D’Itri says. “Having Scout connected to that data, organized in a way that it can find what’s relevant, provides a quick way to improve profitability without a lot of effort.”
D’Itri advises other operators to start with the business outcome rather than the tool. “It’s really important that you understand what you’re trying to drive and what kind of business outcomes you want to achieve,” he says. For Terrion, that meant putting AI where the tower business runs: on the system of record, from day one.
Having Scout connected to that data, organized in a way that it can find what’s relevant, provides a quick way to improve profitability without a lot of effort.
FAQs
Scout is Sitetracker’s agentic AI platform for critical infrastructure operations. It operates within Sitetracker’s data model, inheriting existing security and permission structures, and supports workflows across the full asset lifecycle, from document processing and data access to automated job creation and portfolio-level analysis.
Scout inherits the role-based access controls already configured in Sitetracker. Users can only access data they are already entitled to see, and Scout is designed to respect those entitlements when taking actions. This reduces the need to build a separate security layer for AI.
Early use cases at tower operators include site inspection report processing, lease document inquiry and validation, and operational status reporting. Scout can also be extended to new workflows through natural language prompts without pre-configured agents.
General-purpose tools work from what you give them in the moment, which suits open-ended work. Scout runs against live Sitetracker data within the permission model already configured, and it can execute work in the system by creating jobs, populating records, and running configured Agents and Skills.
Yes. Sitetracker is the system of record Scout reasons against, and connectors extend that context to adjacent systems including ERP, CRM, GIS, construction, email, and document storage. That matters because most operational questions, from billing accuracy to asset performance, cross system boundaries. Administrators provision connectors centrally and control which systems Scout can reach, with every agent execution traced and auditable.