The Underlying Study Data
By Phil Hatch · Akholi · October 2026
Each open card links to the deep-dive analysis of one indicator: its specification, summary of findings, distributions, written responses where the indicator has them, and its associations with the four study measures. There are 216 deep-dive analyses in four parts.
Find an indicator
216 analysesShowing all 216 analyses. 71 are open now and 145 are to follow.
Engagement Outcomes and Their Correlates
The engagements in the study were performing poorly on every core study measure.
Is Canceled?
The cancellation flag recorded a study set mean (mean 0.4, median 0.0, n = 150 engagements).
Open the analysis →Customer Satisfaction %
The study set mean for customer satisfaction was below its median (mean 43.9, median 53.5, n = 150 engagements).
Open the analysis →Engagement indicator mean
The engagement indicator mean across the study set was just above its median (mean 4.0, median 3.8, n = 150 engagements).
Open the analysis →Age of Engagement (Months)
Engagement age, recorded in months for every engagement (mean 5.7, median 5.0, n = 150 engagements), held no association with any of the four study measures.
Open the analysis →Bank Type
Asset management was the most common institution type among the engagements (24 of 150 engagements, 16.00%), with Wealth management and private banking close behind (23 of 150 engagements, 15.33%).
Open the analysis →Corporate Functional or LOB Operational
Engagements split almost evenly between LOB Operational (76 of 150 engagements, 50.67%) and Corporate Functional (74 of 150 engagements, 49.33%) work.
Open the analysis →Engagement Phase
Build-Out (60 of 150 engagements, 40.00%) and Transition/Cut-Over (60 of 150 engagements, 40.00%) were the most common engagement phases, ahead of Production (30 of 150 engagements, 20.00%).
Open the analysis →Headcount
Engagement headcount had a mean close to its median across the study set (mean 8.4, median 8.0, n = 150 engagements).
Open the analysis →Outsourcing Type
The engagements split across the four outsourcing types, from CXM (42 of 150 engagements, 28.00%) to KPO (31 of 150 engagements, 20.67%).
Open the analysis →Use Case
Corporate and back-office functions was the most common process family (47 of 150 engagements, 31.33%), ahead of Technology engineering and operations (35 of 150 engagements, 23.33%).
Open the analysis →What type of a bank is the customer?
Wealth management and private banking was the institution type respondents named most often (140 of 848 respondents, 16.51%), followed by Asset management (132 of 848 respondents, 15.57%).
Open the analysis →What is the outsourcing type?
Respondents named ITO most often as the outsourcing type (244 of 849 respondents, 28.74%) and KPO least often (165 of 849 respondents, 19.43%).
Open the analysis →Is the outsourcing engagement Corporate Functional, or Line of Business Operational?
Respondents split close to evenly between LOB Operational (450 of 849 respondents, 53.00%) and Corporate Functional engagements (399 of 849 respondents, 47.00%).
Open the analysis →How long ago did this specific engagement start, in months?
Respondents reported a mean engagement age in months since the engagement started (mean 5.6, median 5.0, n = 804 respondents).
Open the analysis →What stage is the engagement in?
Engagement Canceled was the most common stage answer (309 of 810 respondents, 38.15%), followed by Development / buildout (131 of 810 respondents, 16.17%).
Open the analysis →How many months has the engagement been in its current stage?
Respondents reported a mean time in the current stage in months (mean 3.1, median 2.0, n = 794 respondents).
Open the analysis →Per the current plan, how many months from today until the engagement reaches Production/Delivery?
Respondents reported a mean expected time to Production/Delivery in months on the current plan (mean 3.9, median 4.0, n = 545 respondents).
Open the analysis →If the engagement has not been canceled, how likely is the customer to cancel the engagement?
Distressed answers on the financial institution’s likelihood to cancel came from a minority of answers (76 of 328 answers, 23.17%) and a minority of engagements (26 of 150 engagements).
Open the analysis →How many people are on the vendor team for this engagement today?
Vendor team size (mean 8.8, median 9.0, n = 789 respondents) showed no association with any of the four study measures.
Open the analysis →What was the peak vendor team headcount on this engagement to date?
Peak vendor team headcount (mean 10.5, median 10.0, n = 787 respondents) was associated with every study measure except customer satisfaction, which returned a tested null.
Open the analysis →Per the current plan, how many people will be on the vendor team 12 months from now?
The vendor team size respondents reported under the current plan for a year from now (mean 5.6, median 6.0, n = 339 respondents) was associated with all four study measures.
Open the analysis →How many AI agents are running in production on this engagement today?
At least half of respondents reported no AI agents running in production on their engagement (mean 1.2, median 0.0, n = 787 respondents).
Open the analysis →What is the pricing model for this engagement?
FTE-based pricing was the most common answer (140 of 347 respondents, 40.35%), ahead of Fixed price (61 of 347 respondents, 17.58%) and Transaction or outcome based pricing (58 of 347 respondents, 16.71%).
Open the analysis →What is the contract term of this engagement, in months?
Respondents reported multi-year contract terms at the median (mean 30.9, median 36.0, n = 339 respondents).
Open the analysis →What is the annual contract value of this engagement, in USD?
Respondents most often placed annual contract value in the second value band (106 of 248 respondents, 42.74%) or the third (102 of 248 respondents, 41.13%), and the lowest band was the least common answer (13 of 248 respondents, 5.24%).
Open the analysis →I know the KPIs the customer uses to measure success
Distressed ratings on knowing the KPIs the financial institution uses to measure success came from a minority of answers (226 of 787 answers, 28.72%).
Open the analysis →The customer is happy with the performance of the engagement at this stage
Distressed ratings on whether the financial institution was satisfied with the engagement’s performance at this stage came from a minority of respondents (89 of 379 answers, 23.48%).
Open the analysis →The engagement is meeting delivery dates successfully
Distressed ratings on meeting delivery dates came from a minority of answers (234 of 798 answers, 29.32%) and a minority of engagement means (30 of 150 engagements).
Open the analysis →The engagement is within budget
Distressed ratings on staying within budget came from a minority of answers (151 of 554 answers, 27.26%) and a minority of engagement means (31 of 150 engagements).
Open the analysis →The engagement is not achieving stated KPIs
Distressed answers on whether the engagement was achieving its stated KPIs came from a minority of answers (241 of 782 answers, 30.82%) and a minority of engagements (32 of 150 engagements).
Open the analysis →What is the customer’s approximate total asset size, in USD?
Respondents most often placed the financial institution’s total assets in the second-highest asset band (90 of 242 respondents, 37.19%) and least often in the highest band (35 of 242 respondents, 14.46%).
Open the analysis →Does the customer operate its own global capability centers?
A majority of respondents reported that the financial institution operated its own global capability centers (190 of 349 respondents, 54.44%), and the rest reported that it did not (159 of 349 respondents, 45.56%).
Open the analysis →How many months has your employer been serving this customer across all work, not just this engagement?
Respondents reported that their employer had served the financial institution for several years on average (mean 32.6, median 33.0, n = 354 respondents).
Open the analysis →How many agentic AI outsourcing engagements does your employer run concurrently with this customer today?
The number of agentic AI outsourcing engagements respondents reported running concurrently with the financial institution (mean 17.2, median 18.0, n = 254 respondents) showed no association with any of the four study measures.
Open the analysis →The customer is scaling their agentic AI outsourcing engagements
Ratings of whether the financial institution was scaling its agentic AI outsourcing engagements (mean 5.5, median 5.0, n = 364 respondents) showed no association with any of the four study measures.
Open the analysis →The customer is increasing the number of agentic AI outsourcing engagements
Agreement that the financial institution was increasing its number of agentic AI outsourcing engagements (mean 5.3, median 5.0, n = 247 respondents) showed tested nulls against all four study measures.
Open the analysis →The customer is shifting more work to their global capability centers
Higher ratings that the financial institution was shifting more work to its global capability centers came with weaker results on all four study measures.
Open the analysis →What type of company is your employer?
IT services major was the most common employer type among respondents (316 of 793 respondents, 39.85%), and Pure-play agentic AI firm was the least common (75 of 793 respondents, 9.46%).
Open the analysis →Approximately how many employees does your employer have worldwide?
The worldwide employee count respondents reported for their employer (mean 110,954.9, median 90,854.0, n = 538 respondents) showed no association with any of the four study measures.
Open the analysis →How many years has your employer been delivering agentic AI outsourcing?
Respondents reported the years their employer had delivered agentic AI outsourcing (mean 1.0, median 1.0, n = 571 respondents).
Open the analysis →How many agentic AI engagements does your employer run for banks today?
Respondents reported how many agentic AI engagements their employer ran for financial institutions, with the mean close to the median (mean 24.6, median 25.0, n = 328 respondents).
Open the analysis →The three remaining measures among the engagements not canceled: 94 active engagements
The three remaining measures moved together on the 94 active engagements.
Open the analysis →Shared structure across the four study measures: cancellation, customer satisfaction, engagement indicator mean, intent to leave
The four study measures moved together across the engagements studied (n = 150 engagements).
Open the analysis →The four study measures by engagement indicator mean quintile: 150 engagements in five bands of 30
Every study measure moved in order across the five engagement indicator mean quintiles set by the rule printed in Analysis Results.
Open the analysis →Reported return on the engagement: ROI %
Respondents reported a return on 17 of the 30 engagements in production.
Open the analysis →Indicator associations net of the engagement indicator mean: 100 component indicators against three measures
The analysis tested each component indicator against cancellation, customer satisfaction, and intent to leave, net of the leave-one-out engagement indicator mean, in 300 tests.
Open the analysis →The Five Failure Categories
The survey asked each respondent to rate 100 indicators of the engagement, 20 in each of five failure categories.
Compared to 5 years ago, is my employer more or less prepared to manage outsourcing engagements?
Ratings of how prepared the employer was to manage outsourcing engagements compared with five years earlier (mean 3.7, median 4.0, n = 786 respondents) were associated with all four study measures.
Open the analysis →Rate the quality of the customer’s data across the systems this engagement depends on
Ratings of the quality of the financial institution’s data across the systems the engagement depended on (mean 3.6, median 4.0, n = 795 respondents) were associated with all four study measures.
Open the analysis →Rate how accessible the customer’s data is to this engagement when and where the work requires it
Ratings of how accessible the financial institution’s data was to the engagement when and where the work required it (mean 3.5, median 3.0, n = 798 respondents) were associated with all four study measures.
Open the analysis →Rate how well the customer’s data is normalized, with consistent formats and definitions across source systems
Ratings of how well the financial institution’s data was normalized across source systems (mean 3.5, median 3.0, n = 809 respondents) were associated with all four study measures.
Open the analysis →Rate the maturity of the customer’s data governance over the data this engagement uses
Ratings of the maturity of the financial institution’s data governance over the data the engagement used (mean 3.6, median 3.0, n = 797 respondents) were associated with all four study measures.
Open the analysis →Rate how well the customer’s legacy systems support the integration this engagement requires
Ratings of how well the financial institution’s legacy systems supported the integration the engagement required (mean 3.5, median 3.0, n = 801 respondents) were associated with all four study measures.
Open the analysis →Rate the readiness of the customer’s technical infrastructure to support the agentic AI this engagement deploys
Ratings of the readiness of the financial institution’s technical infrastructure for the agentic AI the engagement deployed (mean 3.5, median 3.0, n = 803 respondents) were associated with all four study measures.
Open the analysis →Rate how completely the customer’s processes are mapped and documented to the depth agentic AI requires
Ratings of how completely the financial institution’s processes were mapped and documented to the depth agentic AI required (mean 3.5, median 3.0, n = 795 respondents) were associated with all four study measures.
Open the analysis →Rate the maturity of the customer’s processes in scope, meaning how standardized and repeatable they are
Ratings of the maturity of the financial institution’s processes in scope (mean 3.5, median 3.0, n = 789 respondents) were associated with all four study measures.
Open the analysis →Rate how consistently the customer performs the same processes across its business units and regions
Ratings of how consistently the financial institution performed the same processes across its business units and regions (mean 3.6, median 3.0, n = 801 respondents) were associated with all four study measures.
Open the analysis →Rate the customer’s willingness to redesign processes for agentic AI rather than automate them as they are
Ratings of the financial institution’s willingness to redesign processes for agentic AI rather than automate them as they were (mean 3.5, median 3.0, n = 804 respondents) were associated with all four study measures.
Open the analysis →Rate the agentic AI knowledge and skills of the customer staff working with this engagement
Ratings of the agentic AI knowledge and skills of the financial institution staff working with the engagement (mean 3.5, median 3.0, n = 793 respondents) were associated with all four study measures.
Open the analysis →Rate the effectiveness of the customer’s management of its side of this engagement
Ratings of how effectively the financial institution managed its side of the engagement (mean 3.6, median 4.0, n = 811 respondents) were associated with all four study measures.
Open the analysis →Rate the availability of the customer’s subject matter experts when this engagement needs them
Ratings of the availability of the financial institution’s subject matter experts when the engagement needed them (mean 3.5, median 3.0, n = 793 respondents) were associated with all four study measures.
Open the analysis →Rate the customer organization’s readiness to absorb the changes to daily work that agentic AI introduces
Ratings of the financial institution’s readiness to absorb the changes to daily work that agentic AI introduced (mean 3.5, median 4.0, n = 794 respondents) were associated with all four study measures.
Open the analysis →Rate the customer’s ability to provision the system access and identities that AI agents require
Ratings of the financial institution’s ability to provision the system access and identities that AI agents required (mean 3.5, median 3.0, n = 801 respondents) were associated with all four study measures.
Open the analysis →Rate how quickly the customer responds when this engagement raises issues that need customer action
Ratings of how quickly the financial institution responded when the engagement raised issues that needed its action (mean 3.5, median 3.0, n = 789 respondents) were associated with all four study measures.
Open the analysis →Rate how well the customer’s change control process handles the changes this engagement requires
Ratings of how well the financial institution’s change control process handled the changes the engagement required (mean 3.6, median 3.0, n = 800 respondents) were associated with all four study measures.
Open the analysis →Rate the customer’s capacity to govern and oversee the risks of AI agents doing this work
Ratings of the financial institution’s capacity to govern and oversee the risks of AI agents doing the work (mean 3.5, median 3.0, n = 797 respondents) were associated with all four study measures.
Open the analysis →Rate how clearly ownership and accountability for AI are assigned inside the customer organization
Ratings of how clearly ownership and accountability for AI were assigned inside the financial institution (mean 3.5, median 3.0, n = 802 respondents) were associated with all four study measures.
Open the analysis →Rate the stability of the customer’s business case and ROI definition for this engagement
Ratings of the stability of the financial institution’s business case and definition of return for the engagement (mean 3.5, median 3.0, n = 794 respondents) were associated with all four study measures.
Open the analysis →Rate the overall effectiveness of your employer’s top leadership team
Ratings of the overall effectiveness of the employer’s top leadership team (mean 3.6, median 4.0, n = 792 respondents) were associated with all four study measures.
Open the analysis →Rate how well your employer is transforming its commercial model as agentic AI reduces the billable human effort it sells today
Ratings of how well the employer was transforming its commercial model as agentic AI reduced the billable human effort it sold (mean 3.7, median 4.0, n = 812 respondents) were associated with all four study measures.
Open the analysis →Rate how well your employer’s internal management processes have been redesigned for agentic AI delivery rather than traditional staffed delivery
Ratings of how well the employer’s internal management processes had been redesigned for agentic AI delivery (mean 3.7, median 4.0, n = 799 respondents) were associated with all four study measures.
Open the analysis →Rate the effectiveness of your employer’s process and leadership development programs
Ratings of the effectiveness of the employer’s process and leadership development programs (mean 3.7, median 4.0, n = 803 respondents) were associated with all four study measures.
Open the analysis →Rate your employer’s ability to manage agentic AI outsourcing engagements end to end
Ratings of the employer’s ability to manage agentic AI outsourcing engagements end to end (mean 3.7, median 4.0, n = 787 respondents) were associated with all four study measures.
Page to followRate the depth of your employer’s knowledge of the financial industry
Ratings of the depth of the employer’s knowledge of the financial industry (mean 3.6, median 4.0, n = 802 respondents) were associated with all four study measures.
Page to followRate your employer’s ability to manage agentic AI outsourcing engagements for global banks in particular
Ratings of the employer’s ability to manage agentic AI outsourcing engagements for global financial institutions (mean 3.6, median 4.0, n = 779 respondents) were associated with all four study measures.
Page to followRate your employer’s current supply of qualified agentic AI talent against what its engagements need
Ratings of the employer’s supply of qualified agentic AI talent against what its engagements needed (mean 3.7, median 4.0, n = 802 respondents) were associated with all four study measures.
Page to followRate the adequacy of your employer’s investment in training its workforce
Ratings of the adequacy of the employer’s investment in training its workforce (mean 3.7, median 4.0, n = 789 respondents) were associated with all four study measures.
Page to followRate how well your employer’s training programs build practical agentic AI skills
Ratings of how well the employer’s training programs built practical agentic AI skills (mean 3.7, median 4.0, n = 804 respondents) were associated with all four study measures.
Page to followRate how well your employer’s training programs build financial industry knowledge
Ratings of how well the employer’s training programs built financial industry knowledge (mean 3.7, median 4.0, n = 788 respondents) were associated with all four study measures.
Page to followRate the maturity of your employer’s playbook for taking agentic use cases from concept to governed production
Ratings of the maturity of the employer’s playbook for taking agentic use cases from concept to governed production (mean 3.7, median 4.0, n = 807 respondents) were associated with all four study measures.
Page to followRate your employer’s quality assurance capability for agents whose outputs are not identical every run
Ratings of the employer’s quality assurance capability for agents whose outputs varied from run to run (mean 3.7, median 4.0, n = 794 respondents) were associated with all four study measures.
Page to followRate your employer’s ability to price agentic work profitably when billable hours no longer measure the value delivered
Ratings of the employer’s ability to price agentic work profitably once billable hours no longer measured the value delivered (mean 3.6, median 4.0, n = 795 respondents) were associated with all four study measures.
Page to followRate how well the technical platforms your employer builds or licenses are optimized for agentic AI outsourcing
Ratings of how well the technical platforms the employer built or licensed were optimized for agentic AI outsourcing (mean 3.7, median 4.0, n = 794 respondents) were associated with all four study measures.
Page to followRate your employer’s readiness to take accountability when an agent it operates makes a costly error
Ratings of the employer’s readiness to take accountability when an agent it operated made a costly error (mean 3.7, median 4.0, n = 794 respondents) were associated with all four study measures.
Page to followRate the depth of your employer’s knowledge of agentic AI
Ratings of the depth of the employer’s knowledge of agentic AI (mean 3.6, median 3.0, n = 807 respondents) were associated with all four study measures.
Page to followRate the depth of your employer’s experience running agentic AI systems in live production, beyond pilots
Ratings of the depth of the employer’s experience running agentic AI systems in live production (mean 3.6, median 4.0, n = 794 respondents) were associated with all four study measures.
Page to followRate your employer’s ability to operate large numbers of agents in production, including monitoring and exception handling
Ratings of the employer’s ability to operate large numbers of agents in production (mean 3.7, median 4.0, n = 798 respondents) were associated with all four study measures.
Page to followRate your employer’s overall qualification to deliver elite agentic AI outsourcing services to banks
Ratings of the employer’s overall qualification to deliver agentic AI outsourcing services to financial institutions (mean 3.6, median 4.0, n = 803 respondents) were associated with all four study measures.
Page to followRate how effectively the customer and vendor jointly manage this engagement now that agents perform part of the work
Ratings of how effectively the financial institution and the vendor jointly managed the engagement once agents performed part of the work (mean 3.7, median 4.0, n = 789 respondents) were associated with all four study measures.
Page to followRate how clearly the two firms have settled who decides what agents may do without human approval
Ratings of how clearly the two firms had settled who decides what agents may do without human approval (mean 3.6, median 3.0, n = 808 respondents) were associated with all four study measures.
Page to followRate how clearly ownership of each workflow step is assigned when agents, vendor staff, and customer staff work in the same process
Ratings of how clearly ownership of each workflow step was assigned when agents, vendor staff, and financial institution staff worked in the same process (mean 3.7, median 4.0, n = 808 respondents) were associated with all four study measures.
Page to followRate how clearly accountability lands on one firm when an agent’s action causes a problem in this engagement
Ratings of how clearly accountability landed on one firm when an agent’s action produced a problem (mean 3.7, median 3.0, n = 799 respondents) were associated with all four study measures.
Page to followRate how well the two firms maintain a single shared view of the risks that agent work introduces
Ratings of how well the two firms maintained a single shared view of the risks that agent work introduced (mean 3.6, median 4.0, n = 793 respondents) were associated with all four study measures.
Page to followRate how effectively the two firms manage issues arising from agent work as one team rather than through two separate processes
Ratings of how effectively the two firms managed issues arising from agent work as one team (mean 3.7, median 4.0, n = 799 respondents) were associated with all four study measures.
Page to followRate how well the speed of joint issue resolution matches the speed at which agent-caused problems can spread
Ratings of how well the speed of joint issue resolution matched the speed at which problems from agent actions could spread (mean 3.7, median 4.0, n = 809 respondents) were associated with all four study measures.
Page to followRate how quickly the two firms together can stop an agent that is behaving badly in live operations
Ratings of how quickly the two firms together could stop an agent behaving badly in live operations (mean 3.6, median 3.0, n = 804 respondents) were associated with all four study measures.
Page to followRate how well the joint change control process keeps pace with the speed of agent behavior changes
Ratings of how well the joint change control process kept pace with changes in agent behavior (mean 3.7, median 3.5, n = 798 respondents) were associated with all four study measures.
Page to followRate how well agent updates made by one firm become visible to the other firm before they take effect
Ratings of how well agent updates made by one firm became visible to the other firm before they took effect (mean 3.6, median 4.0, n = 803 respondents) were associated with all four study measures.
Page to followRate how well the two firms jointly govern changes agents make to their own behavior between formal releases
Ratings of how well the two firms jointly governed changes agents made to their own behavior between formal releases (mean 3.7, median 4.0, n = 806 respondents) were associated with all four study measures.
Page to followRate how well human review points are placed across the two firms at the moments where agent errors would matter most
Ratings of how well human review points were placed across the two firms at the moments where agent errors would matter most (mean 3.6, median 4.0, n = 797 respondents) were associated with all four study measures.
Page to followRate how well joint human review capacity keeps up with the volume of output the agents produce
Ratings of how well joint human review capacity kept up with the volume of output the agents produced (mean 3.7, median 4.0, n = 805 respondents) were associated with all four study measures.
Page to followRate how much genuine scrutiny agent outputs receive at human review points rather than routine approval
Ratings of the depth of scrutiny agent outputs received at human review points rather than routine approval (mean 3.6, median 4.0, n = 794 respondents) were associated with all four study measures.
Page to followRate how reliably the joint quality assurance loop catches agent errors before they reach live operations
Ratings of how reliably the joint quality assurance loop caught agent errors before they reached live operations (mean 3.7, median 4.0, n = 787 respondents) were associated with all four study measures.
Page to followRate how well the quality measures both firms use capture the kinds of errors agents actually make
Ratings of how well the quality measures both firms used captured the kinds of errors agents made (mean 3.6, median 3.0, n = 792 respondents) were associated with all four study measures.
Page to followRate how fully each firm can see what the other firm’s agents actually did when questions arise
Ratings of how fully each firm could see what the other firm’s agents had done when questions arose (mean 3.7, median 3.0, n = 780 respondents) were associated with all four study measures.
Page to followRate how closely the speed of joint decision making matches the speed at which the agents act
Ratings of how closely the speed of joint decision making matched the speed at which the agents acted (mean 3.7, median 4.0, n = 786 respondents) were associated with all four study measures.
Page to followRate how well today’s joint governance model fits work performed by agents rather than work performed by people
Ratings of how well the current joint governance model fit work performed by agents rather than by people (mean 3.6, median 4.0, n = 803 respondents) were associated with all four study measures.
Page to followRate how ready the joint management model is for the level of agent autonomy you expect in this engagement within two years
Ratings of how ready the joint management model was for the level of agent autonomy respondents expected within two years (mean 3.6, median 3.0, n = 808 respondents) were associated with all four study measures.
Page to followRate the factual reliability of the AI tools’ outputs on this engagement before any human review
Ratings of the factual reliability of the AI tools’ outputs before any human review (mean 4.6, median 5.0, n = 796 respondents) were associated with all four study measures.
Page to followRate the accuracy of citations, references, and quotations the tools produced against the real sources
Ratings of the accuracy of the citations, references, and quotations the tools produced (mean 4.5, median 5.0, n = 791 respondents) were associated with all four study measures.
Page to followRate the consistency of the tools’ outputs when given the same input and instructions more than once
Ratings of the consistency of the tools’ outputs when given the same input and instructions more than once (mean 4.6, median 5.0, n = 800 respondents) were associated with all four study measures.
Page to followRate the end-to-end reliability of the tools on multi-step tasks, from first step to finished output
Ratings of the end-to-end reliability of the tools on multi-step tasks (mean 4.5, median 5.0, n = 782 respondents) were associated with all four study measures.
Page to followRate the predictability of the tools’ token consumption for a given volume of work
Ratings of the predictability of the tools’ token consumption for a given volume of work (mean 4.6, median 4.0, n = 795 respondents) were associated with all four study measures.
Page to followRate the efficiency of the tools’ token use, considering retries, loops, and reprocessing needed to complete work
Ratings of the efficiency of the tools’ token use, counting retries, loops, and reprocessing (mean 4.6, median 5.0, n = 788 respondents) were associated with all four study measures.
Page to followRate how completely the tools kept each client’s and each session’s information isolated from every other client and session
Ratings of how completely the tools kept each financial institution’s and each session’s information isolated (mean 4.6, median 5.0, n = 791 respondents) were associated with all four study measures.
Page to followRate how reliably the tools retained instructions and constraints given earlier in long tasks and sessions
Ratings of how reliably the tools retained instructions and constraints given earlier in long tasks and sessions (mean 4.6, median 5.0, n = 796 respondents) were associated with all four study measures.
Page to followRate how completely the tools captured material facts located anywhere in long documents, including mid-document content
Ratings of how completely the tools captured material facts located anywhere in long documents (mean 4.6, median 5.0, n = 805 respondents) were associated with all four study measures.
Page to followRate the integrity of the tools’ persistent memory against contamination by incorrect or malicious content encountered during work
Ratings of the integrity of the tools’ persistent memory against contamination by incorrect or malicious content (mean 4.6, median 4.0, n = 806 respondents) were associated with all four study measures.
Page to followRate the tools’ resistance to hostile instructions embedded in the content they read and process
Ratings of the tools’ resistance to hostile instructions embedded in the content they read and processed (mean 4.6, median 5.0, n = 812 respondents) were associated with all four study measures.
Page to followRate how consistently the agents took only intended and authorized actions when operating connected tools and systems
Ratings of how consistently the agents took only intended and authorized actions when operating connected tools and systems (mean 4.6, median 5.0, n = 789 respondents) were associated with all four study measures.
Page to followRate how safely the tools handled sensitive data, with no exposure through outputs, logs, or connected systems
Ratings of how safely the tools handled sensitive data (mean 4.5, median 4.0, n = 800 respondents) were associated with all four study measures.
Page to followRate how reliably the tools’ built-in guardrails and safety controls held during production use
Ratings of how reliably the tools’ built-in guardrails and safety controls held during production use (mean 4.7, median 5.0, n = 787 respondents) were associated with all four study measures.
Page to followRate the overall production readiness of the AI tools for bank-grade work as experienced on this engagement
Ratings of the overall production readiness of the AI tools for financial-industry work on the engagement (mean 4.5, median 5.0, n = 802 respondents) were associated with all four study measures.
Page to followRate how accurately the tools’ own reports of success or failure matched the actual outcomes of their work
Ratings of how accurately the tools’ own reports of success or failure matched the actual outcomes of their work (mean 4.6, median 5.0, n = 786 respondents) were associated with all four study measures.
Page to followRate the tools’ ability to explain and evidence how each output was produced, at the depth an audit requires
Ratings of the tools’ ability to explain and evidence how each output was produced, at the depth an audit required (mean 4.6, median 5.0, n = 800 respondents) were associated with all four study measures.
Page to followRate how well the tools supported reliable testing, meaning correct behavior could be proven before work went live
Ratings of how well the tools supported testing that could establish correct behavior before work went live (mean 4.6, median 5.0, n = 792 respondents) were associated with all four study measures.
Page to followRate the stability of the tools’ behavior over the engagement when nothing changed on the engagement side
Ratings of the stability of the tools’ behavior over the engagement when nothing changed on the engagement side (mean 4.6, median 5.0, n = 797 respondents) were associated with all four study measures.
Page to followRate the quality of the tools’ logs, traces, and usage telemetry for detecting and diagnosing failures
Ratings of the quality of the tools’ logs, traces, and usage telemetry for detecting and diagnosing failures (mean 4.6, median 5.0, n = 796 respondents) were associated with all four study measures.
Page to followRate how well the agents on this engagement are protected from breaking silently when a third party changes something without warning
Ratings of how well the agents were protected from breaking silently when a third party changed something without warning (mean 4.6, median 5.0, n = 781 respondents) were associated with all four study measures.
Page to followRate the quality of the fallback the engagement has ready when a critical third-party feed, platform, or model stops working
Ratings of the quality of the fallback the engagement had ready when a critical third-party feed, platform, or model stopped working (mean 4.6, median 5.0, n = 802 respondents) were associated with all four study measures.
Page to followRate the security posture of the third-party tools, platforms, and sub-processors the engagement’s agents connect to
Ratings of the security posture of the third-party tools, platforms, and sub-processors the agents connected to (mean 4.6, median 5.0, n = 788 respondents) were associated with all four study measures.
Page to followRate the quality of advance notice third-party providers give before changes that can affect the agents on this engagement
Ratings of the quality of advance notice third-party providers gave before changes that could affect the agents (mean 4.6, median 5.0, n = 801 respondents) were associated with all four study measures.
Page to followRate how reliably word of an upcoming third-party change reaches the people who operate the affected agents in time to act
Ratings of how reliably word of an upcoming third-party change reached the people operating the affected agents in time to act (mean 4.6, median 5.0, n = 798 respondents) were associated with all four study measures.
Page to followRate the engagement’s monitoring for unannounced changes in the third-party APIs, data feeds, and models its agents depend on
Ratings of the engagement’s monitoring for unannounced changes in the third-party APIs, data feeds, and models its agents depended on (mean 4.6, median 5.0, n = 794 respondents) were associated with all four study measures.
Page to followRate how well the contracts with third-party providers deal with provider changes that can break the agents on this engagement
Ratings of how well the contracts with third-party providers dealt with provider changes that could break the agents (mean 4.7, median 5.0, n = 799 respondents) were associated with all four study measures.
Page to followRate how well the engagement absorbs model version deprecations and retirements forced by model providers
Ratings of how well the engagement absorbed model version deprecations and retirements forced by model providers (mean 4.6, median 5.0, n = 793 respondents) were associated with all four study measures.
Page to followRate the availability of true machine interfaces, rather than screens built for humans, across the third-party systems the agents must use
Ratings of the availability of true machine interfaces across the third-party systems the agents had to use (mean 4.6, median 5.0, n = 802 respondents) were associated with all four study measures.
Page to followRate the stability of the third-party APIs and data structures the agents depend on, from one provider release to the next
Ratings of the stability of the third-party APIs and data structures the agents depended on, from one provider release to the next (mean 4.7, median 5.0, n = 796 respondents) were associated with all four study measures.
Page to followRate the quality of the data arriving from third-party feeds, judged by whether agents can use it without human cleanup
Ratings of the quality of the data arriving from third-party feeds, judged by whether agents could use it without human cleanup (mean 4.6, median 5.0, n = 795 respondents) were associated with all four study measures.
Page to followRate how well third-party rate limits and quotas hold up under the volume the agents on this engagement generate
Ratings of how well third-party rate limits and quotas held up under the volume the agents generated (mean 4.6, median 5.0, n = 805 respondents) were associated with all four study measures.
Page to followRate the engagement’s ability to still perform by hand the manual workarounds that once absorbed third-party quirks and breakage
Ratings of the engagement’s ability to still perform by hand the workarounds that once absorbed third-party quirks and breakage (mean 4.7, median 5.0, n = 808 respondents) were associated with all four study measures.
Page to followRate how completely the engagement knows where humans used to quietly fix third-party data and interface problems before agents took over
Ratings of how completely the engagement knew where people had quietly fixed third-party data and interface problems before agents took over (mean 4.6, median 5.0, n = 792 respondents) were associated with all four study measures.
Page to followRate how deliberately the quiet human fixes for third-party quirks were replaced with explicit checks when agents took over the work
Ratings of how deliberately the quiet human fixes for third-party quirks were replaced with explicit checks once agents took over (mean 4.6, median 5.0, n = 796 respondents) were associated with all four study measures.
Page to followRate how well the engagement manages its dependence on a small number of model and cloud providers
Ratings of how well the engagement managed its dependence on a small number of model and cloud providers (mean 4.6, median 5.0, n = 802 respondents) were associated with all four study measures.
Page to followRate how well this engagement could keep operating if it had to switch its primary model or cloud provider
Ratings of how well the engagement could keep operating if it had to switch its primary model or cloud provider (mean 4.6, median 5.0, n = 811 respondents) were associated with all four study measures.
Page to followRate how well third-party licenses on this engagement cover work done by agents rather than named human users
Ratings of how well third-party licenses covered work done by agents rather than named human users (mean 4.6, median 5.0, n = 799 respondents) were associated with all four study measures.
Page to followRate how completely the engagement has mapped every third and fourth party its agents depend on, down the full chain
Ratings of how completely the engagement had mapped every third and fourth party its agents depended on (mean 4.6, median 5.0, n = 791 respondents) were associated with all four study measures.
Page to followRate how well the engagement manages agent dependence on gated external rails such as payment networks and regulatory submission systems
Ratings of how well the engagement managed agent dependence on gated external rails such as payment networks and regulatory submission systems (mean 4.6, median 5.0, n = 794 respondents) were associated with all four study measures.
Page to followThe Delivery Workforce
The delivery workforce of the study set was early in its career and new to agentic AI, and it learned the agents on live engagements.
CME Respondent Count
The number of Customer Management Executive respondents on each engagement (mean 0.0, median 0.0, n = 150 engagements) showed tested nulls against all four study measures.
Page to followIC Respondent Count
The Individual Contributor respondent count recorded for each engagement (mean 4.4, median 4.0, n = 150 engagements) showed tested nulls against all four study measures.
Page to followPD Respondent Count
The Portfolio Director respondent count recorded for each engagement (mean 0.2, median 0.0, n = 150 engagements) showed no association with any of the four study measures.
Page to followTotal Respondents
The number of respondents per engagement (mean 5.7, median 6.0, n = 150 engagements) showed tested nulls against all four study measures.
Page to followWhat is your role on this engagement?
Individual Contributors made up most of the respondents (667 of 850 respondents, 78.47%), and Customer Management Executives were the fewest (2 of 850 respondents, 0.24%).
Page to followHow many years of experience do you have in your career?
Years of career experience (mean 3.9, median 2.0, n = 797 respondents) showed tested nulls against all four study measures.
Page to followHow many years of experience do you have with agentic AI?
Years of experience with agentic AI (mean 0.4, median 0.0, n = 793 respondents) showed tested nulls against all four study measures.
Page to followHow many years have you been working in the outsourcing industry?
Years worked in the outsourcing industry (mean 2.6, median 1.0, n = 806 respondents) showed tested nulls against all four study measures.
Page to followHow many years have you been working within the financial industry?
Years working within the financial industry (mean 2.1, median 1.0, n = 789 respondents) showed tested nulls against all four study measures.
Page to followHow many agentic AI outsourcing engagements have you worked on across your career, including this one?
The number of agentic AI outsourcing engagements respondents had worked on across their careers (mean 1.8, median 2.0, n = 799 respondents) showed no association with any of the four study measures.
Page to followHow well do you know the financial industry?
Respondents’ ratings of how well they knew the financial industry (mean 4.8, median 5.0, n = 796 respondents) were associated with all four study measures.
Page to followHow well do you know agentic AI specific to the financial industry?
Respondents’ ratings of how well they knew agentic AI specific to the financial industry (mean 4.9, median 5.0, n = 797 respondents) were associated with all four study measures.
Page to followRate your overall agentic AI expertise
Respondents’ ratings of their own overall agentic AI expertise (mean 4.9, median 5.0, n = 800 respondents) were associated with all four study measures.
Page to followWhat professional certifications do you have?
Respondents most often reported no professional certification (236 of 793 respondents, 29.76%), followed by Six Sigma Green Belt (127 of 793 respondents, 16.02%).
Page to followWhat AI certifications do you have?
Most respondents reported holding no AI certification (437 of 788 respondents, 55.46%).
Page to followWhere did you gain your AI knowledge?
Respondents named the sources of their AI knowledge at similar rates, from Vendor and partner training as the most common (140 of 800 respondents, 17.50%) to Employer training program as the least common (125 of 800 respondents, 15.63%).
Page to followHow happy are you with your current job?
Respondents’ ratings of how satisfied they were with their current job (mean 4.2, median 4.0, n = 780 respondents) were associated with all four study measures.
Page to followHow likely are you to leave your current job in the next 12 months?
Every figure on this item was read on the goodness view (each figure is 11 minus the raw answer), so a higher figure marks a respondent less likely to leave.
Page to followHow happy is your extended team with the current engagement?
Respondents’ ratings of how satisfied their extended team was with the current engagement (mean 4.3, median 4.0, n = 788 respondents) were associated with all four study measures.
Page to followHow likely are members of your current team to leave their jobs in the next 12 months?
Ratings of how likely the respondent’s team was to leave within twelve months were read on the goodness view (mean 4.3, median 4.0, n = 799 respondents), so a higher figure marks a team less likely to leave.
Page to followCompared to 5 years ago, I am happier with the work I am doing now
Agreement that the respondent was more satisfied with the work than five years earlier (mean 4.9, median 5.0, n = 804 respondents) was associated with all four study measures.
Page to followCompared to 5 years ago, I am more likely to seek a new job
Answers on whether respondents were more likely to seek a new job than five years earlier (mean 4.9, median 5.0, n = 801 respondents) were read on the goodness view (each figure is 11 minus the raw answer), so a higher figure marks less inclination to seek one.
Page to followCompared to 5 years ago, the team I work with seems to be happier
Agreement that the respondent’s team seemed more satisfied than five years earlier (mean 4.9, median 5.0, n = 797 respondents) was associated with all four study measures.
Page to followCompared to 5 years ago, the team I work with is more likely to seek a new job
Answers on whether the respondent’s team was more likely to seek a new job than five years earlier (mean 4.9, median 5.0, n = 799 respondents) were read on the goodness view (each figure is 11 minus the raw answer), so a higher figure marks less reported job-seeking.
Page to followI want to continue working on projects in the global financial industry
Agreement that the respondent wanted to continue working on projects in the global financial industry (mean 4.9, median 5.0, n = 784 respondents) was associated with all four study measures.
Page to followI would prefer to work directly for a bank in one of their global capability centers
Agreement that the respondent would prefer to work directly for a financial institution in one of its global capability centers (mean 6.0, median 6.0, n = 799 respondents) was associated with all four study measures.
Page to followWorking directly for a bank would be better than working for an outsourcing vendor
Agreement that working directly for a financial institution would be better than working for an outsourcing vendor (mean 6.1, median 6.0, n = 802 respondents) was associated with all four study measures.
Page to followI trust the customer
Agreement that the respondent trusted the financial institution (mean 4.9, median 5.0, n = 799 respondents) was associated with all four study measures.
Page to followThe customer knows and cares about me
Agreement that the financial institution knew and cared about the respondent (mean 4.9, median 5.0, n = 789 respondents) was associated with all four study measures.
Page to followI talk with peers inside the customer regularly
Agreement that the respondent talked regularly with peers inside the financial institution (mean 5.5, median 6.0, n = 809 respondents) showed tested nulls against all four study measures.
Page to followI talk with engagement management inside the customer regularly
Agreement that the respondent talked regularly with engagement management inside the financial institution (mean 5.5, median 6.0, n = 790 respondents) showed tested nulls against all four study measures.
Page to followI talk with the customer’s leadership team regularly
Agreement that the respondent talked regularly with the financial institution’s leadership team (mean 5.6, median 6.0, n = 333 respondents) showed tested nulls against all four study measures.
Page to followI know how the customer feels about my team
Agreement that the respondent knew how the financial institution felt about the respondent’s team (mean 4.9, median 5.0, n = 798 respondents) was associated with all four study measures.
Page to followI know how the customer feels about the engagement I am on
Agreement that the respondent knew how the financial institution felt about the engagement (mean 4.9, median 5.0, n = 791 respondents) was associated with all four study measures.
Page to followMy manager is effective
Agreement that the respondent’s manager was effective (mean 4.8, median 5.0, n = 802 respondents) was associated with all four study measures.
Page to followMy manager knows the financial industry well
Agreement that the respondent’s manager knew the financial industry well (mean 4.8, median 5.0, n = 800 respondents) was associated with all four study measures.
Page to followMy manager knows agentic AI well
Agreement that the respondent’s manager knew agentic AI well (mean 4.7, median 5.0, n = 803 respondents) was associated with all four study measures.
Page to followMy manager cares about me
Agreement that the respondent’s manager cared about the respondent (mean 4.7, median 5.0, n = 804 respondents) was associated with all four study measures.
Page to followMy manager cares about the larger team
Agreement that the respondent’s manager cared about the larger team (mean 4.7, median 5.0, n = 797 respondents) was associated with all four study measures.
Page to followMy manager knows how to manage agentic AI outsourcing engagements
Agreement that the respondent’s manager knew how to manage agentic AI outsourcing engagements (mean 4.8, median 5.0, n = 800 respondents) was associated with all four study measures.
Page to followMy manager knows how to manage agentic AI outsourcing engagements for global banks
Agreement that the respondent’s manager knew how to manage agentic AI outsourcing engagements for global financial institutions (mean 4.8, median 5.0, n = 791 respondents) was associated with all four study measures.
Page to followMy employer cares about me
Agreement that the employer cared about the respondent (mean 5.0, median 5.0, n = 786 respondents) was associated with all four study measures.
Page to followMy employer cares about the team
Agreement that the employer cared about the team (mean 4.9, median 5.0, n = 802 respondents) was associated with all four study measures.
Page to followMy employer’s top leadership team cares about my wellbeing and development
Agreement that the employer’s top leadership team cared about the respondent’s wellbeing and development (mean 4.9, median 5.0, n = 812 respondents) was associated with all four study measures.
Page to followI trust my employer
Agreement that the respondent trusted their employer (mean 4.9, median 5.0, n = 799 respondents) was associated with all four study measures.
Page to followI trust my manager
Agreement that the respondent trusted their manager (mean 4.7, median 5.0, n = 807 respondents) was associated with all four study measures.
Page to followI am being used effectively by my employer
Agreement that the employer used the respondent effectively (mean 4.9, median 5.0, n = 808 respondents) was associated with all four study measures.
Page to followYears of experience by reported intent to leave: C007 by C019
Respondents likely to leave, under the rule printed in Analysis Results, reported fewer years of experience (mean 3.8, n = 409 respondents) than respondents not likely to leave (mean 4.4, n = 337 respondents).
Page to followManual Alteration by the Delivery Team
The delivery team sustained the engagements by altering their inputs, processes and outputs by hand.
The dashboard and performance reports given to the customer are accurate
Agreement that the dashboards and performance reports given to the financial institution were accurate (mean 4.8, median 5.0, n = 794 respondents) was associated with all four study measures.
Page to followThe vendor manipulates the dashboard and performance reports given to the customer
Ratings of whether the vendor altered the dashboards and performance reports given to the financial institution were read on the goodness view, so a higher figure marks less reported reporting alteration (mean 4.7, median 5.0, n = 786 respondents).
Page to followCompared to 5 years ago, is manipulation of the dashboards and performance reports given to the customer more or less common now?
Answers on whether reporting alteration was more or less common than five years earlier were read on the goodness view (each figure is 11 minus the raw answer), so a higher figure marks it seen as less common (mean 4.8, median 5.0, n = 795 respondents).
Page to followI fully understand how these laws, regulations, and similar affect the outsourcing engagement
Agreement that the respondent fully understood how the governing laws and regulations affected the engagement (mean 4.8, median 5.0, n = 800 respondents) was associated with all four study measures.
Page to followMy management team fully understands these laws, regulations, and similar, and how they affect the engagement
Agreement that the respondent’s management team fully understood the governing laws and regulations and how they affected the engagement (mean 4.9, median 5.0, n = 516 respondents) was associated with all four study measures.
Page to followThe customer’s team actively discusses these laws, regulations, and similar with our team
Agreement that the financial institution’s team actively discussed the governing laws and regulations with the vendor team (mean 4.8, median 5.0, n = 801 respondents) was associated with all four study measures.
Page to followThe training available to me effectively teaches me about these laws, regulations, and similar
Agreement that the training available to the respondent effectively taught the governing laws and regulations (mean 3.6, median 4.0, n = 798 respondents) was associated with all four study measures.
Page to followI have altered client-provided data
Respondents divided almost evenly on whether they had manually altered financial-institution-provided data, with Yes (421 of 839 respondents, 50.18%) and No (418 of 839 respondents, 49.82%) close to parity.
Page to followI discussed the data alteration with internal management
Most respondents who answered reported internal disclosure of the manual alteration of financial-institution-provided data (292 of 401 respondents, 72.82%), and the rest reported none (109 of 401 respondents, 27.18%).
Page to followI or my management discussed data alteration with the client
Nearly every respondent who answered reported no financial institution disclosure of the manual alteration of financial-institution-provided data (363 of 366 respondents, 99.18%), and a few reported one (3 of 366 respondents, 0.82%).
Page to followMy greater team regularly alters client provided data
Agreement that the respondent’s wider team regularly altered financial-institution-provided data (mean 5.4, median 5.0, n = 773 respondents) was associated with all four study measures.
Page to followI have altered client processes
Most respondents reported that they had manually altered financial institution processes (490 of 841 respondents, 58.26%), and the rest reported that they had not (351 of 841 respondents, 41.74%).
Page to followI discussed the process alteration with internal management
Most respondents who answered reported internal disclosure of the manual alteration of financial institution processes (341 of 459 respondents, 74.29%), and the rest reported none (118 of 459 respondents, 25.71%).
Page to followI or my management discussed process alteration with the client
Nearly every respondent who answered reported no financial institution disclosure of the manual alteration of financial institution processes (405 of 414 respondents, 97.83%), and a few reported one (9 of 414 respondents, 2.17%).
Page to followMy greater team regularly alters client processes
Agreement that the respondent’s wider team regularly altered financial institution processes (mean 5.7, median 6.0, n = 779 respondents) was associated with all four study measures.
Page to followI have altered AI outputs
Most respondents reported that they had not manually altered AI outputs (550 of 828 respondents, 66.43%), and the rest reported that they had (278 of 828 respondents, 33.57%).
Page to followI discussed the AI output alteration with internal management
Most respondents who answered reported internal disclosure of the manual alteration of AI outputs (150 of 253 respondents, 59.29%), and the rest reported none (103 of 253 respondents, 40.71%).
Page to followI or my management discussed AI output alteration with the client
Nearly every respondent who answered reported no financial institution disclosure of the manual alteration of AI outputs (223 of 224 respondents, 99.55%), and a single respondent reported one (1 of 224 respondents, 0.45%).
Page to followMy greater team regularly alters AI outputs
Agreement that the respondent’s wider team regularly altered AI outputs (mean 4.4, median 4.0, n = 665 respondents) was associated with all four study measures.
Page to followThe three alteration items by respondent role: C199, C205, C211 by C005
Across all respondents who answered, reported manual alteration was most common for financial institution processes (58.26% of 841), followed by financial-institution-provided data (50.18% of 839) and AI outputs (33.57% of 828).
Page to followThe three alteration items as reported by managers and directors: C199, C205, C211 by C005
Engagement Manager respondents reported 14 manual alterations across the three items, and Portfolio Director respondents reported 0, against 1,175 from Individual Contributor respondents.
Page to follow