Rate how well the tools supported reliable testing, meaning correct behavior could be proven before work went live

The mean rating was 4.6 and the median 5.0 (n = 792 respondents). Ratings of 3 or below, the distressed range, made up 32.70% of the answers. The rating was associated with all four study measures. Higher ratings came with less cancellation (r = −0.67, 95% CI −0.75 to −0.57, p < 0.001, n = 150 engagements), higher customer satisfaction among active engagements (r = 0.66, 95% CI 0.53 to 0.76, p < 0.001, n = 94 engagements), a higher engagement indicator mean computed without the indicator (r = 0.87, 95% CI 0.82 to 0.90, p < 0.001, n = 150 engagements) and lower intent to leave (r = −0.66, cluster bootstrap 95% CI −0.71 to −0.61, p < 0.001 with standard errors clustered by engagement, n = 741 respondents). After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 51.82, Holm p < 0.001, omega squared 0.441), respondent role (Welch F 10.38, Holm p < 0.001, omega squared 0.018) and vendor type (Welch F 4.18, Holm p = 0.006, omega squared 0.021).

Of the 38 written responses, nearly all from C177, the comment field on the AI tools, 26 were negative and 10 positive, and 17 carried the strongest negative score of −10. Respondents said correct behavior could not be proven before work went live, in several cases because no test environment was provided, so errors surfaced on live runs and real cases. Across engagements, a more negative engagement mean written score came with cancellation, and respondents whose written responses were more negative reported higher intent to leave.

Ratings of 3 or below, the distressed range, accounted for 259 of the 792 answers (32.70%). The engagement mean fell in the distressed range in 32 of 150 engagements.

The primary responses numbered 34, all written in answer to C177, “Add any comments on the agentic AI tools that the ratings in this section do not capture.” The remaining 4 responses, 10.53% of the total, were secondary and came from 3 other questions. That secondary share ranked 42nd from the highest among the 71 Part 2 indicators with 30 or more written responses, against a median of 18.92%. Respondents who wrote both a primary and a secondary response numbered 0. Negative responses made up 67.65% of primary responses (23 of 34) and 3 of the 4 secondary responses.

Twenty-two responses said correct behavior could not be proven, or was proven only thinly, before work went live. Seven described no test environment, or none at realistic volume or fidelity: changes validated on real employee tickets during business hours, no test environment for billing, only a prod-like setting for remediations, and ten sample rows where real volumes were needed. Seven described problems found only on live runs or outputs already sent: on a live NAV run, in a false positive report, on real cases, and in a regulatory submission file, a point the respondent called too late for regulatory work. One respondent said the tool gave a different pass or fail on the same test run.

Five responses described manual controls or hand-built test material in place of testing support: a supervisor sampling closed dispute cases for an hour a day, two runs of the same schedule compared side by side in a spreadsheet, a day spent writing fake employee requests because no test set existed, and last month's closed batch rerun before trusting a new run. Two responses to C198 described model version changes: one had two weeks' notice and compared the old and new prompt sets before switching, the other described two weeks' notice to rebuild a pipeline that had taken two months, and an interface change with no notice at all.

The ten positive responses described a sandbox, validation pack or test environment that let the team test a job before it went live, several as an improvement on an earlier engagement, and placed their concerns in cost, logging and consistency.

Of the 38 responses, 26 (68.42%, 95% CI 54.05% to 81.58%) were negative and 17 (44.74%) carried the strongest negative score of −10. Positive responses numbered 10 (26.32%), and 2 were neutral. The mean written score was −4.3 (95% CI −6.2 to −2.4, n = 38 responses), and the median was −5.0.

Of the 38 responses, 38 came from respondents who also rated the indicator. Among them, the written score rose with the rating (Spearman rho 0.69, 95% CI 0.50 to 0.82, p < 0.001, n = 38 responses). The written responses corroborated the rating. Of the 26 negative responses with a rating, 15 (57.69%) came from respondents whose rating was in the distressed range.

Respondents who wrote about the indicator and gave a rating numbered 38, with a mean rating of 4.4. The 754 respondents who rated the indicator without writing about it had a mean of 4.6. The difference, −0.2 (95% CI −1.0 to 0.7), included zero. Ratings in the distressed range made up 42.11% of the ratings from respondents who wrote (16 of 38) and 32.23% of the ratings from respondents who did not (243 of 754).

Every figure in the segment tables is at respondent level. Segments with fewer than 30 respondents carry no interval. Segments with fewer than 10 report counts only.

Segment families recorded at engagement level were tested on engagement means. Respondent role and vendor type were tested on respondents. Where an expected count fell below 5, the test of the distressed share used Fisher's exact test for a two-by-two table, marked exact, or a Monte Carlo test, marked simulated. After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 51.82, Holm p < 0.001, omega squared 0.441). Active engagements in Build-Out rated highest (mean 5.5, n = 162 respondents) and canceled engagements lowest (mean 3.1, n = 307 respondents). Ratings also differed by respondent role (Welch F 10.38, Holm p < 0.001, omega squared 0.018). Portfolio directors rated highest (mean 5.6, n = 30 respondents) and individual contributors lowest (mean 4.4, n = 610 respondents). Ratings also differed by vendor type (Welch F 4.18, Holm p = 0.006, omega squared 0.021). Consulting firms rated highest (mean 5.1, n = 111 respondents) and BPO specialists lowest (mean 4.1, n = 183 respondents). Outsourcing model, institution type, process family and class of work showed no difference.

Written sentiment did not differ by engagement status and lifecycle stage across the 33 engagements with a written response (Kruskal-Wallis H 7.25, df 3, p = 0.064). Of the groups with 10 or more responses, responses from canceled engagements were the most negative (mean −8.1, 13 responses) and responses from active engagements in Transition/Cut-Over the least negative (mean −2.7, 15 responses). Responses from active engagements in Build-Out (5) and active engagements in Production (5) are reported as counts only.

No test across respondent roles was run, because fewer than two roles had 30 or more written responses. Individual contributors wrote 29, engagement managers wrote 6 and portfolio directors wrote 3 responses, reported as counts only.

Cancellation rate is the share of engagements canceled by the study date, measured at engagement level (150 engagements, 56 canceled). Canceled engagements were rated after cancellation. The association is therefore a descriptive contrast and not a forecast of cancellation.

Across the study set, higher ratings came with less cancellation (r = −0.67, 95% CI −0.75 to −0.57, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in fifteen of the 20 segment levels tested.

Customer satisfaction is the percentage recorded against each engagement. The primary estimate uses the 94 active engagements. A descriptive contrast uses all 150 engagements.

Across the active engagements, higher ratings came with higher customer satisfaction (r = 0.66, 95% CI 0.53 to 0.76, Holm p < 0.001, n = 94 engagements). The association held after Holm correction in twelve of the 16 segment levels tested.

The engagement indicator mean is the mean of an engagement's ratings across the rated indicators, measured at engagement level (150 engagements). The indicator is one of the rated indicators. Each comparison therefore uses the mean recomputed without it.

Higher ratings came with a higher engagement indicator mean (r = 0.87, 95% CI 0.82 to 0.90, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in twenty-four of the 24 segment levels tested.

Intent to leave is the answer to C019, “How likely are you to leave your current job in the next 12 months?”, rated 1 to 10. It is measured at respondent level on up to 798 respondents, with the engagement-level form reported alongside.

Respondents who gave higher ratings reported lower intent to leave (r = −0.66, 95% CI −0.70 to −0.62, Holm p < 0.001, n = 741 respondents). The association held after Holm correction in thirty-five of the 37 segment levels tested.

Two comparisons set the engagements with at least one negative response against all other engagements. Of the 126 other engagements, 117 had no written response addressing the indicator. These two comparisons therefore do not isolate written sentiment. Of the 24 engagements with a negative response, 10 (41.67%) were canceled, against 46 of the 126 other engagements (36.51%). The difference did not reach significance (Fisher exact p = 0.651). Customer satisfaction among active engagements did not differ between the two groups (mean 65.57% on 14 engagements against 63.84% on 80, Welch t 0.57, p = 0.573).

Two correlations test written sentiment directly. Across the 33 engagements with a written response, a more negative engagement mean written score came with cancellation (point-biserial r −0.43, p = 0.012). Across 36 respondents, a more negative respondent mean written score came with higher intent to leave (Spearman rho −0.38, p = 0.021). Of the four comparisons, two reached significance at 0.05.

The four associations were compared in pairs on their engagement-level forms, each oriented so that a higher value is the better outcome, across the 150 engagements with all four measures. After Holm correction, five of the six pairs differed. The association with the engagement indicator mean was the strongest and the association with customer satisfaction the weakest. The associations with customer satisfaction and intent to leave did not differ (Holm p = 0.797).

Answers clustered within engagements (ICC 0.60). Clustering reduced the 792 respondent answers to the equivalent of about 222 independent answers. Respondent-level intervals therefore used the cluster bootstrap. Respondent-level tests of intent to leave used standard errors that allow for clustering by engagement.

No association was equivalent to zero. Each of the four differed from zero. The detection floor shown is the study floor at 150 engagements. Intent to leave was tested on its effective sample size.

Each change was computed before rounding and may differ by 0.01 from the difference between the two printed values.

All 22 checks held. A check holds when its test falls on the same side of 0.05 as the primary test. Where the primary test falls below 0.05, the check must also keep the sign of the primary correlation. The thin-base check removes the engagements in any level of outsourcing model or process family with fewer than 30 engagements. Institution type is not used for that check, because every one of its levels has fewer than 30 engagements.

Each of the four measures was computed from data the indicator does not contain. Each result is therefore a finding and not an arithmetic consequence. The indicator is one of the rated indicators, and the engagement indicator mean was recomputed without it. The correction changed r by 0.00, from 0.87 to 0.87.

Each deviation was computed before rounding and may differ by 0.01 from the difference between the two printed values.

The comparison covered 150 engagements. The indicator followed the study's typical pattern on all four measures, and no deviation held after Holm correction.

Among the 71 Part 2 indicators with 30 or more written responses, the indicator's share of negative responses, 68.42%, ranked 61st from the highest, against a median of 80.82%. Its mean written score, −4.3, ranked 45th from the most negative, against a median of −4.7.

The indicator most often present in the same responses was C157, “Rate the factual reliability of the AI tools' outputs on this engagement before any human review.”, in 7 of the 38 responses (18.42%). C159, “Rate the consistency of the tools' outputs when given the same input and instructions more than once.” (6 responses), C171, “Rate the overall production readiness of the AI tools for bank-grade work as experienced on this engagement.” (6 responses), C160, “Rate the end-to-end reliability of the tools on multi-step tasks, from first step to finished output.” (4 responses) and C172, “Rate how accurately the tools' own reports of success or failure matched the actual outcomes of their work.” (4 responses) followed.

The most frequent theme was “Comparison to a prior engagement”, in 6 responses (15.79%). The second was “Respondent proposes a remedy”, in 5 responses.

About a third of all ratings of how well the tools supported reliable testing fell in the distressed range (259 of 792 answers, 32.70%). The rating was associated with all four study measures, and every association held under each of the 22 robustness checks. Canceled engagements rated far below the three active stages (mean 3.1, n = 307 respondents).

The written evidence divides engagements into those given a place to test the work and those that tested it in production. Where no test environment existed, live operation became the test, and errors reached the financial institution's customers, employees or regulatory filings. Written sentiment was associated with cancellation and with intent to leave, on a small base.

The evidence is the reported experience of the vendor's delivery teams, collected in a single cross-section. The associations reported here are not causal.

Tests by respondent role and by vendor type treated respondents as independent. The section on clustering reports how far clustering within engagements reduced the information in respondent-level figures.

Customer management executives gave 2 answers, a base of anecdote strength.

Shares of written sentiment are shares of responses and not of the study set of respondents. The written score takes five fixed values, so its mean and median are coarse. Responses clustered within engagements (38 responses on 33 engagements). Intervals for written sentiment therefore used a bootstrap by engagement.