This report is built on Wave 3 of the Georgian + NewtonX Applied AI Benchmarks, an ongoing study of how B2B software companies are turning AI ambition into operating reality. Each wave re-fields a consistent core of questions so that change can be measured over time, alongside new benchmarks that track emerging behaviour.
At a glance
Wave 3 at a glance
| Metric | Value | Detail |
|---|---|---|
| Total respondents | 501 | B2B software decision makers |
| Technical decision makers | 252 | Longitudinal Wave 2 → Wave 3 panel |
| Fielded | Mar–Apr 2026 | Wave 3 |
| Confidence | 95% | Significance threshold for reported changes |
The benchmark series
The Applied AI Benchmarks are a longitudinal series. Wave 1 established the baseline, Wave 2 was fielded in June 2025, and Wave 3 — the basis for this report — was fielded in March–April 2026, roughly nine months later. Tracking the same questions across waves is what lets us report movement (for example, agentic AI implementation rising from 45% in Wave 2 to 81% in Wave 3) rather than a single snapshot.
The research is produced by Georgian in partnership with NewtonX, which manages respondent sourcing and fielding.
Sample & respondents
Wave 3 surveyed 501 B2B software decision makers. Respondents fall into two role groups, each crossed with company stage:
- Technical decision makers (TDMs) — leaders accountable for engineering, product, and AI/ML systems. A matched subset of 252 TDMs was tracked directly from Wave 2 to Wave 3 (shown as “Direct W2 → W3” on the relevant charts), which is what enables true wave-over-wave comparisons for technical metrics.
- Go-to-market (GTM) leaders — leaders accountable for sales, marketing, customer success, and revenue.
Each role is reported across two company stages — Enterprise and Growth-stage — producing the four segments used throughout the report (Tech × Enterprise, Tech × Growth-stage, GTM × Enterprise, GTM × Growth-stage). See Definitions for how each segment and the maturity tiers are defined.
How to read the data
- Base sizes. Every chart carries the question wording and the base it was asked of. `n=501` denotes the full sample; `n=252` denotes the technical decision-maker panel. Always read a figure against its stated base.
- Significance. Changes flagged “Significant (95%)” clear a 95% confidence threshold versus the comparison wave or segment.
- Points vs percent. A change expressed in points (pts) is an absolute difference between two percentages (e.g. 45% → 81% is a +36-point move), not a relative percentage change.
- Rounding. Percentages are rounded to whole numbers, so stacked figures may not sum to exactly 100.
Longitudinal comparisons
Wave-over-wave figures labelled ”Direct W2 → W3” are drawn from the matched 252-respondent technical panel — the same people answering in both waves — which is the most rigorous way to measure change and removes sample-composition noise. Comparisons without that label are wave-to-wave at the sample level. New benchmarks introduced in Wave 3 have no prior-wave comparison and are labelled accordingly.
Limitations
- Findings reflect self-reported practice and sentiment from decision makers, not instrumented telemetry from their systems.
- The sample is limited to B2B software organizations and does not generalize to other sectors.
- Correlations between maturity and outcomes describe association, not proven causation — for example, the report does not imply that any single guardrail causes a change in incident rates.
For definitions of the terms used here, see Definitions; for external studies referenced in the report, see Data sources.