Grid Dynamics Q2 2026 Shows 90% Code Generated by Agents

How Grid Dynamics discusses “code generation” in its earnings transcript

Where “code generation” appears (with context and speaker)

1) Prepared remarks: “code generation and testing” as a core capability in Horizon One

In management’s prepared remarks, Grid Dynamics frames its agentic modernization platform around using agents to accelerate work across the modernization lifecycle, calling out “code generation and testing” explicitly. ¹

“Agents can now accelerate work across most of the modernization life cycle, particularly code generation and testing.” ¹

The CEO then clarifies the operating model: while an agent can write code, “a person still makes a call and stands behind it,” positioning the company’s value around governance and accountability rather than fully autonomous engineering. ¹

The same remarks also connect this capability to specific delivered outcomes for clients—e.g., a program where “seven COBOL services were moved” and “approximately 90% of the code [was] generated by agents,” with all services entering production in the quarter. ¹

2) Broader strategy framing: differentiation beyond code generation

In the earlier excerpt describing the company’s technology and differentiation, Grid Dynamics says its differentiation is “not limited to code generation,” emphasizing that its agents can incorporate additional context (e.g., security advisories, dependencies, upstream changes). ²

This matters because management positions code-only tools as insufficient for enterprise-grade outcomes, and instead stresses that richer context helps identify problems those tools can miss and supports the traceability/evidence requirements of regulated enterprises. ²

Thematic synthesis: how the company’s messaging evolves around code generation

Across the excerpts, “code generation” is presented in two reinforcing themes:

1. Execution acceleration inside modernization workstreams
   Code generation is treated as a tangible, near-term accelerator within the modernization lifecycle—especially paired with testing to drive adoption and reliability. ¹

2. Enterprise-grade governance and completeness vs “code-only” approaches
   Grid Dynamics explicitly downplays code generation as the whole story, arguing that agents must use broader operational/security/dependency context to catch issues that simple code generation might miss, and to produce evidence suitable for regulated environments. ²

Strategic analysis: management’s positioning, metrics, and implications

Strategic positioning

Grid Dynamics positions “code generation” as a capability enabled by its broader agentic platform and tooling (e.g., Rosetta and Allium) rather than as a standalone feature. ¹² It emphasizes that while agents can generate code, human judgment remains the accountability layer—a key enterprise adoption message. ¹

Additionally, the company links code generation to “modernization remains the foundation of our business” and to its AI-first transformation (“AI is changing how the work gets done”), implying code generation is part of monetizable workflow transformation. ¹

Quantitative/impact details tied to code generation

Management provides an explicit scale/impact example:

• In a modernization program at a Fortune 30 home improvement retailer, “seven COBOL services were moved to a modern technology stack” and “approximately 90% of the code [was] generated by agents.” ¹
• All seven services entered production this quarter. ¹

Grid Dynamics also quantifies engineering adoption at another client context: approximately “550 of the client’s engineers are working with the platform.” ¹

Forward-looking/commercial implications (as stated)

Grid Dynamics connects code generation-driven productivity to commercial model expansion:

• Productivity is said to be creating opportunities for more fixed-price and outcome-based commercial models when scope/accountability are clearly defined. ¹
• The emphasis on governance (“domain knowledge, governance, and control”) suggests a commercialization strategy centered on enterprise reliability, not just faster coding. ¹

Competitive/market narrative

Management contrasts its approach with “code-only tools,” asserting that their broader context incorporation (security advisories, dependencies, upstream changes) improves problem detection and provides the traceability regulated enterprises expect. ²

It also frames open-source contributions (e.g., SpecFlow) and platform openness as a credibility/adoption mechanism rather than a licensing revenue strategy—supporting the idea that code generation capability is meant to be validated and adopted at enterprise scale. ²

Bottom line

In Grid Dynamics’ earnings transcript excerpts, “code generation” is not discussed as a generic AI feature; it is presented as a measured, workflow-embedded capability within modernization and testing, enabled by enterprise governance. ¹ The company supports this with a concrete metric example (~90% of code generated by agents) and ties the outcome to production readiness and to commercial model opportunities (fixed price/outcome-based), while also emphasizing differentiation beyond code generation through broader operational/security context and traceability. ¹²