The product catalog had been waiting for a year. At Pattern, the ecommerce company, a backlog of features sat paused while data managers faced the decidedly unglamorous job of making product information agree across marketplaces. Amazon, eBay, Mercado Libre: each brought its own attributes and demands. Somewhere inside that tangle was software that needed to launch.
ITX entered with inherited code and an architecture assessment. Its team built a catalog whose attributes could change without a developer rewriting them, then moved to releases more frequent than the usual two-week sprint. The system reached production ready for an audit. The interesting achievement was giving the people managing products more room to manage them.
The useful bits, before the deep dive
- What you hire: product strategists, researchers, designers, developers, and specialists who work alongside your own team.
- Where it fits: custom workflows, inherited software, integrations, and products that need continuing care.
- The transferable idea: investigate the user’s task before committing to the next feature.
01 / THE INHERITED PROBLEMThe catalog that waited a year
Pattern’s Shelf Product Catalog needed more than a tidy front end. Its data had to be reliable enough for internal financial systems, with controls and records that would survive an audit. ITX used an object-oriented database model so data managers could modify properties without coding and upload products and listings in bulk.
ITX reports that a budget originally allocated for three months supported five months of work. Pattern doubled the budget for the next phase. Those are engagement results, not a universal bargain. They do, however, illustrate what a custom software partner can do: understand the existing system, make decisions about its structure, and get a team delivering again.
“ITX’s sense of ownership is incredible.”
DAVID LAKE / SENIOR DIRECTOR OF PRODUCT, PATTERN
02 / A CHANGE OF PERSPECTIVEThe user is not the developer
Ralph Dandrea founded ITX in 1997 as a two-person consultancy handling telecommunications projects. In a conversation marking the company’s twenty-fifth anniversary, he recalled an early weakness: developers imagined themselves as the users. Real people arrived with needs and edge cases that the builders had overlooked.
Product thinking gave ITX a way to examine those differences. It meant considering the various groups using a product rather than treating the developer’s own experience as sufficient. That is a modest-sounding change with expensive implications. An elegant screen can still be an obstacle if its designer misunderstands the person looking at it.
Today, the offering runs from product vision and roadmaps to research, interaction design, architecture, security, development, and support. Interviews and prototypes help establish what to build; engineers and testers make it work. Customers can commission a website or mobile application, connect systems through APIs, modernize a platform, or bring in a team to help their existing people deliver.
A working sequence / not a product menu
An editorial reading of ITX’s strategy, delivery, and post-launch services.
03 / AI MEETS THE CAFETERIALunch is a design problem
For Heartland School Solutions, the user was a school nutrition professional. Menu planning involves dietary needs, changing supplies, and USDA requirements. Add a generative AI tool, and a new difficulty appears: someone accustomed to forms and spreadsheets may now have to become adept at writing prompts.
ITX designed the interface and requirements for Mosaic AI Assistant around guided steps. Detailed prompts and safeguards sit behind the interface, with USDA guidelines incorporated into the experience. Heartland’s developers brought the product into production. The distinction matters: this was collaborative client work, not an ITX subscription app.
The practical lesson is to use what people already understand. A guided form can collect the context a model needs without making the user compose an essay. That approach depends on knowing the workflow and preserving human control. A familiar screen cannot, by itself, make an unreliable answer safe.
04 / CONFIDENCE, TESTEDA pilot earns its keep
At ePlan Services, part of Paychex, service staff needed nuanced answers about retirement plans and regulations. Research across IRS and Department of Labor material took time. The experiment was whether an AI agent could help without sacrificing accuracy.
ITX worked on a six-month proof of concept, organized into eight agile sprints. The agent connected with Salesforce and regulatory resources; staff took part in tuning, feedback, and evaluation. ITX reports a 20% reduction in cost per service interaction during the project. That number describes a pilot, not a promise about every future support call.
According to the case account, staff skepticism softened as they saw the tool’s potential. Demonstration and participation changed minds. The sensible thing to copy is the bounded experiment: choose a task, involve the people doing it, and assess quality alongside speed before expanding the scope.
ITX’s separate Miro AI pilot took that logic into planning. Across two teams, a six-week trial used AI for user stories and dependency mapping. The reported planning session fell from two full days to four hours. It is a specific result in a specific process. Its usefulness lies in identifying a measurable task before automating it.
05 / THE SOFTWARE THAT STILL PAYSKeep the old engine running
WineDirect faced a less fashionable problem. While moving from a monolithic application toward microservices, it still needed its ColdFusion platform to serve wineries. The legacy application was large, complex, and poorly documented. Turning it off would have been an exceptionally efficient way to stop selling wine.
ITX embedded specialists to audit code, document development, automate deployments, and stabilize the application. During the initial ten months, it reported an 83% decrease in the time needed to close a ticket. Keeping an old platform functioning while a replacement takes shape is valuable work, even when it offers little material for a launch party.
At AssuredPartners, acquisition created another recurring software problem: subsidiary websites that needed migration, consistent design, and accessibility work. ITX built reusable Sitecore components and a repeatable migration process, delivering the initial project three weeks early. Reuse mattered because the wider objective involved more than 100 subsidiary sites, not just one attractive redesign.
These cases locate ITX in the market. It combines product and design consulting with engineering and continuing platform care. An internal team, a specialist agency, or another development consultancy could be an alternative. When packaged software already fits the workflow, commissioning custom work deserves a harder business case.
06 / MONEY & MOMENTUMThe price of participation
ITX earns its money through client services. Its published June 2024 terms put prices in the statement of work and allow for fixed-price engagements. Buying the service means specifying outcomes and scope, rather than choosing a seat count from a SaaS price card.
Those terms also make the client part of the delivery system. ITX needs business knowledge, access to platforms and personnel, and timely feedback. Delayed inputs can increase costs and schedules. The implication is straightforward: an outside team cannot decide everything your organization has been avoiding.
Before commissioning a build, define the user task, the acceptance criteria, and who can answer questions. Make access and review time part of the plan. That advice is especially relevant to legacy work, where an undocumented business rule can be more troublesome than an old programming language.
One modernization engagement
A delayed roadmap, compressed
ITX reports completing three epics for a nonprofit client using an additional team and GitHub Copilot. This compares planned and actual effort in one engagement; it does not isolate AI’s contribution.
ITX’s nonprofit case illustrates the opportunity and the limits. A modernization project was already delayed when an additional team joined, using Copilot to analyze legacy code and help with delivery. The company reports completing work scoped at fourteen sprints in 7.25. Extra people, tooling, and the choice of work all belong in that story.
07 / THE PEOPLE BEHIND THE SYSTEMA remote team, with ice cream
ITX describes itself as remote-first. Its careers materials offer details more revealing than the usual culture adjectives: individual development plans, training access, live onboarding with leaders, and salary benchmarks based on the market rather than bargaining skill.
The firm has also added expertise through acquisition. In September 2023, it bought SUMO Heavy, the Philadelphia ecommerce consultancy. This was ITX purchasing a specialist team and practice; it was not a venture funding round for ITX.
Security is part of its proposition, too. In August 2024, ITX announced SOC 2 Type II compliance after an independent audit by MHM Professional Corporation. The announcement reported no deficiencies or remediation issues. That is a concrete institutional checkpoint alongside the softer promise of being easy to work with.
In June 2026, Dandrea and technology leader Michael Lesher argued that AI shifts value toward professional judgment and accountability. For a services company, the argument is commercially convenient. The test remains practical: whether the catalog launches, the staff can trust the support tool, and the old system keeps doing its job while the new one arrives.