Rate the overall production readiness of the AI tools for bank-grade work as experienced on this engagement

The mean rating was 4.5 and the median 5.0 (n = 802 respondents). Ratings of 3 or below, the distressed range, made up 32.04% of the answers. The rating was associated with all four study measures. Higher ratings came with less cancellation (r = −0.69, 95% CI −0.77 to −0.60, p < 0.001, n = 150 engagements), higher customer satisfaction among active engagements (r = 0.64, 95% CI 0.50 to 0.74, p < 0.001, n = 94 engagements), a higher engagement indicator mean computed without the indicator (r = 0.86, 95% CI 0.82 to 0.90, p < 0.001, n = 150 engagements) and lower intent to leave (r = −0.67, cluster bootstrap 95% CI −0.72 to −0.61, p < 0.001 with standard errors clustered by engagement, n = 756 respondents). After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 56.41, Holm p < 0.001, omega squared 0.470), respondent role (Welch F 7.27, Holm p = 0.008, omega squared 0.014) and vendor type (Welch F 3.62, Holm p = 0.018, omega squared 0.017).

Of the 79 written responses, 60 were positive and 15 negative, the lowest share of negative responses among the Part 2 indicators with 30 or more written responses. Respondents described tools that worked, then named the limits they managed around: instructions dropped on long runs, tasks reported complete that were not, results that varied between runs and citations to the wrong source. The negative responses described correcting most of the tool's output. A more negative written score came with cancellation at engagement level and with higher intent to leave.

Ratings of 3 or below, the distressed range, accounted for 257 of the 802 answers (32.04%). The engagement mean fell in the distressed range in 36 of 150 engagements.

The primary responses numbered 55, 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 24 responses, 30.38% of the total, were secondary and came from 4 other questions. That secondary share ranked 30th 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 12.73% of primary responses (7 of 55) and 33.33% of secondary responses (8 of 24).

The fourteen responses to C030, the question on what the respondent loves most about the job, all named the technology itself, often with a qualifier: when it works, when it is allowed to run, when the data is clean. Most responses to C177, the comment field on the AI tools, rated the tools well and then named a limit. Nine described dropping an earlier instruction, a step or the middle of a long document on long runs or sessions, so that teams split cycles into small pieces or repeated the instruction midway. Six described the tool reporting a task or log entry as complete when it was not, in one case a batch of call summaries shown as logged that never reached the ticket.

Eight responses described different results for the same input on different runs or days, so that staff ran the job twice or read every output. Three described citations to sources that were wrong or did not exist, in one case a section of the financial institution's own guide for an endpoint renamed two versions earlier. Others named a narrow permitted scope, unpredictable cost from looping, a document that carried an instruction the agent followed, and logs too thin to reconstruct what the agent did.

Six negative responses to C216 described correcting most of the tool's output or not trusting it at all, in one case seven times out of ten on KYC work, with the lead aware and the financial institution not.

Of the 79 responses, 15 (18.99%, 95% CI 10.71% to 28.57%) were negative and 5 (6.33%) carried the strongest negative score of −10. Positive responses numbered 60 (75.95%), and 4 were neutral. The mean written score was 2.7 (95% CI 1.5 to 3.8, n = 79 responses), and the median was 5.0.

Of the 79 responses, 70 came from respondents who also rated the indicator. Among them, the written score rose with the rating (Spearman rho 0.55, 95% CI 0.37 to 0.71, p < 0.001, n = 70 responses). The written responses corroborated the rating. Of the 13 negative responses with a rating, 9 (69.23%) came from respondents whose rating was in the distressed range.

Respondents who wrote about the indicator and gave a rating numbered 70, with a mean rating of 5.1. The 732 respondents who rated the indicator without writing about it had a mean of 4.5. The difference, 0.6 (95% CI 0.0 to 1.2), excluded zero. Ratings in the distressed range made up 25.71% of the ratings from respondents who wrote (18 of 70) and 32.65% of the ratings from respondents who did not (239 of 732).

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 56.41, Holm p < 0.001, omega squared 0.470). Active engagements in Transition/Cut-Over rated highest (mean 5.5, n = 234 respondents) and canceled engagements lowest (mean 3.1, n = 305 respondents). Ratings also differed by respondent role (Welch F 7.27, Holm p = 0.008, omega squared 0.014). Portfolio directors rated highest (mean 5.4, n = 31 respondents) and individual contributors lowest (mean 4.4, n = 619 respondents). Ratings also differed by vendor type (Welch F 3.62, Holm p = 0.018, omega squared 0.017). Consulting firms rated highest (mean 4.9, n = 112 respondents) and BPO specialists lowest (mean 4.1, n = 186 respondents). Outsourcing model, institution type, process family and class of work showed no difference.

Written sentiment differed by engagement status and lifecycle stage across the 62 engagements with a written response (Kruskal-Wallis H 10.10, df 3, p = 0.018). Of the groups with 10 or more responses, responses from canceled engagements were the most negative (mean −0.4, 23 responses) and responses from active engagements in Build-Out the least negative (mean 4.3, 22 responses). Responses from active engagements in Production (8) 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 59, engagement managers wrote 16 and portfolio directors wrote 4 responses.

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 = −0.69, 95% CI −0.77 to −0.60, 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 = 0.64, 95% CI 0.50 to 0.74, Holm p < 0.001, n = 94 engagements). The association held after Holm correction in eleven 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 = 0.86, 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 = −0.67, 95% CI −0.70 to −0.63, Holm p < 0.001, n = 756 respondents). The association held after Holm correction in thirty-six of the 37 segment levels tested.

Two comparisons set the engagements with at least one negative response against all other engagements. Of the 135 other engagements, 88 had no written response addressing the indicator. These two comparisons therefore do not isolate written sentiment. Of the 15 engagements with a negative response, 9 (60.00%) were canceled, against 47 of the 135 other engagements (34.81%). The difference did not reach significance (Fisher exact p = 0.088). The comparison of customer satisfaction was not computed, because one of the two groups held too few active engagements.

Two correlations test written sentiment directly. Across the 62 engagements with a written response, a more negative engagement mean written score came with cancellation (point-biserial r −0.39, p = 0.002). Across 74 respondents, a more negative respondent mean written score came with higher intent to leave (Spearman rho −0.48, p < 0.001). Of the three 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 = 0.493).

Answers clustered within engagements (ICC 0.58). Clustering reduced the 802 respondent answers to the equivalent of about 227 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 0.86.

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, 18.99%, ranked 71st from the highest, against a median of 80.82%. Its mean written score, 2.7, ranked 71st 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 14 of the 79 responses (17.72%). C159, “Rate the consistency of the tools' outputs when given the same input and instructions more than once.” (10 responses), C164, “Rate how reliably the tools retained instructions and constraints given earlier in long tasks and sessions.” (9 responses), C172, “Rate how accurately the tools' own reports of success or failure matched the actual outcomes of their work.” (7 responses) and C174, “Rate how well the tools supported reliable testing, meaning correct behavior could be proven before work went live.” (6 responses) followed.

The most frequent theme was “Positive with an embedded negative”, in 18 responses (22.78%). The second was “Team or colleagues”, in 13 responses.

About a third of all ratings of the production readiness of the AI tools for bank-grade work fell in the distressed range (257 of 802 answers, 32.04%). 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 = 305 respondents).

The written evidence describes tools that respondents judged ready within limits they had learned and managed by hand: shorter runs, repeated instructions, second runs and reading every output. Where those limits were not managed, staff corrected most of the output. Written sentiment tracked the study measures: a more negative engagement mean came with cancellation, and a more negative respondent mean came with higher intent to leave.

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 (79 responses on 62 engagements). Intervals for written sentiment therefore used a bootstrap by engagement.