Internal app builders shift the boundary of software work
Thesis: Internal AI app builders move some routine software creation to domain employees while keeping data and design constraints inside a company-controlled environment. The Pragmatic Engineer reports Ramp's Glass and Stripe's Harbor as examples, not proof of universal adoption.
Mechanism: An internally built AI-assisted environment lets non-engineers assemble and share tools within company-specific data and design boundaries, while engineers provide the platform and its constraints. Ramp's Glass lets employees create and share internal tools; the report says nearly every non-engineer uses it, and simple tool-building is part of new-hire onboarding. One marketer built a custom CRM. Stripe's Harbor took one engineer about six weeks. The report says internal control can reduce the risk of sending company data to an outside vendor and enforce company design rules.
Why now: The examples suggest that the cost of assembling a tailored internal builder may be low enough for companies to consider it, while generic external products do not necessarily meet internal data or design requirements. The evidence is a reported profile of two companies, with no comparative cost, quality, security, or productivity measurements; it cannot establish that an in-house platform is safer or better in general.
Implications: Domain employees may be able to address small workflow gaps without waiting for a conventional engineering queue. Engineers' work may shift toward providing the platform, governing data access, and reviewing tools rather than implementing every small request. First implication: Wider participation only creates durable value if teams can inspect, maintain, and permission the tools that employees create.
Watch: Whether these tools move beyond prototypes into maintained workflows, and what review and access controls companies put around them. First watch condition: Look for evidence about who owns, reviews, and maintains employee-built tools after launch.