The CTO in a recent Larridin demo already had an in-house system for connecting AI spend with engineering work. His description of it was, “bailing wire and toothpicks.” The problem was how much custom machinery it took to keep the measurement working.
The instinct behind a homegrown AI ROI dashboard makes sense. In the Larridin demo, the CTO wanted to correlate AI spend with merge requests and story points, see who was up to speed, and identify where mentorship could help. His organization included 80–85 engineers in the U.S. plus another large group internationally.
For a technical team, building a first version may be faster than waiting for a new platform. Python scripts and API calls can pull usage and cost data, while engineering systems already contain delivery data.
The difficulty comes as the measurement needs expand. Cost, delivery, and project context live in different systems, and every new AI tool or API change adds more upkeep. Answering the next question can mean building yet another connection between the data.
At that point, the dashboard is no longer a side project. It is measurement infrastructure that engineering owns.
The cost of maintaining a custom build grows as the system gets more complex, usually in three areas:
That last point was the real issue for the CTO in the demo. He wasn’t looking for a better cost dashboard. He wanted to know whether AI investment was translating into engineering results and where the organization should intervene.
Larridin’s Developer Productivity platform connects to GitHub, Jira, and the broader development toolchain to measure AI adoption alongside engineering velocity, quality, cost, and ROI.
That changes the build-vs-buy calculation. Engineering teams no longer have to create every connection between AI activity and delivery data themselves just to answer the next leadership question.
Larridin also measures AI proficiency and adoption patterns and provides visibility into AI tools across the organization. Those signals can help leaders investigate why two teams using similar tools are getting different results without turning a custom cost-tracking project into an ever-larger internal analytics platform.
The deeper proficiency comparison question is its own problem. A principal engineer in the same recent demo was less interested in getting cost data “more accurately” than in comparing high performers with peers on the same tools and finding the behaviors worth spreading. That’s where behavioral measurement starts doing work a cost script was never designed to do.
The question is whether the custom system still earns the engineering time it consumes.
A homegrown build is easier to justify when it answers a narrow, stable question with little maintenance. The case for a purpose-built platform gets stronger when leadership needs broader tool coverage, repeated integration work, cross-team comparisons, or connections among AI spend, adoption, quality, and delivery.
CIO.com’s discussion of the broader AI execution gap makes a related point: deploying technology is not the same as turning it into measurable enterprise value. For a CTO, the measurement layer has to keep up as the AI program moves from experimentation into day-to-day operations.
Yes. A focused custom build can be useful when the organization has a narrow measurement need and the engineering effort to maintain it stays low. The tradeoff changes as the number of tools, data sources, teams, and questions grows.
A cost script can show usage and spend from the systems it connects to. Larridin connects AI activity with broader engineering signals such as adoption, velocity, quality, and ROI, while its AI Fluency capability adds proficiency measurement.
No. A parallel comparison can help engineering leaders see where the two systems overlap and whether a purpose-built platform answers questions the existing build cannot. The transition approach depends on the organization’s current integrations and reporting needs.
The CTO wanted to know whether AI spend could be connected with engineering delivery, including merge requests and story points. Larridin connects AI measurement with GitHub, Jira, and the broader development stack so leaders can evaluate spend alongside the engineering outcomes they already track.
Larridin gives engineering leaders a shared view of AI adoption, cost, proficiency, and developer outcomes without requiring their teams to keep expanding a custom measurement stack.
Book a discovery call to see how Larridin connects AI investment with engineering output.