Ця сторінка ще не доступна українською. Показано англійську версію.

Market insight

Integration decides whether AI scales.

In large organisations, AI meets existing telephony, CRM, collection and policy systems. Most barriers reported by enterprises are about connecting to those systems safely.

Key points

Integration with existing systems is the barrier enterprises name most often, ahead of the AI model itself.

Deployment at scale remains the exception in financial services and insurance, where security and compliance reviews add to the integration effort.

The work that decides whether automation scales sits before and after the conversation: data exchange, hand-back rules and controls.

Barriers named by enterprises

  • 77% of the 30 US banking executives surveyed cited integrating new technology with other systems as a major modernisation challenge; 67% cited data security and compliance. 37% already used generative AI in contact centres and a further 37% planned to in 2026.

    Deloitte Center for Financial Services, How banks can turn AI-assisted customer service into a business advantage, 30 June 2026. Small sample: 30 US banking executives (plus 100 customers); modernisation challenges in general, not barriers to AI-assisted service specifically.

  • Among less mature contact centres, 72% reported integration problems, 58% legacy-system constraints and 53% security or compliance concerns.

    Deloitte Digital, 2026 Global Contact Center Survey, 4 June 2026. Subgroup comparison.

  • 65% of executives said their organisations faced low operational efficiency in customer service; 16% of front-line agents were satisfied with their work (68% of supervisors and managers); 70% of agents reported that generative AI had reduced their overall workload.

    Capgemini Research Institute, Unleashing the value of customer service, 18 March 2025. Three samples, fieldwork November–December 2024: about 9,500 consumers; 506 agents and supervisors (315 agents); 1,002 executives. The percentages do not share a denominator; workload is self-reported.

  • Of more than 300 leaders, only 5% felt they were fully there in integrating marketing, sales and service.

    KPMG US, Total Experience Research 2026. Not specific to telephony.

How far deployment has scaled

  • 10% of financial-services organisations had deployed AI agents at scale.

    Capgemini, World Cloud Report – Financial Services 2026, 12 November 2025. 1,100 leaders in 14 markets, survey June–July 2025; not limited to contact centres.

  • 41% of the customer-facing AI use cases in the study cited had already been scaled; McKinsey argues that further progress requires orchestration of decisions, handovers and shared context.

    McKinsey, Rewiring customer experience for the agentic era, 14 July 2026. Covers customer-facing functions generally, not telephony alone.

  • 44% of insurers considered themselves AI leaders, but full redesign of insurance processes was rare and about 11% reported the strong data foundations and governance needed to scale AI.

    KPMG International, Unlocking AI Value in Insurance, 29 September 2026. Self-assessment; no total sample size published.

  • 67% of contact centres still relied on manual or limited automation.

    Benchmarking the 2026 State of Contact Center Maturity (400 leaders), as reported by CMSWire, February 2026. Secondary report.

  • In the United Kingdom 16% of firms had fully deployed “AI digital workers”, against about 45% in the United States; the barriers cited were trust, security, legacy systems and dispersed decision-making.

    RingCentral research, as reported by CX Today, 28 May 2026. Vendor research; secondary report.

Where the limits sit

  • Some organisations are on the way to automating as much as 70% of customer contact; the interviewee named organisations’ willingness to change as the factor that limits how far this goes.

    McKinsey, Beyond the bot: Building empathetic customer experiences with agentic AI (podcast transcript), 6 February 2026. Interview with a representative of a voice-AI vendor; not a representative survey.

Why integration dominates

Vee's interpretation. The analysis below is our own reading of the sources above, not a finding of the organisations cited.

The hard part of enterprise contact automation is rarely the conversation itself. It is the data the process needs before the call (who to contact, about what, under which consent), the data it must return afterwards (outcome, promise, appointment, complaint) and the rules for handing a case back to people with its history intact.

In large organisations each of these steps touches systems that predate the AI project – telephony, CRM, core banking, policy administration, collection platforms – and each connection must pass security and compliance review. The Deloitte and Capgemini figures suggest that this, rather than model capability, is where projects slow down.

Integration is also where operating knowledge accumulates. Once data flows, hand-back rules and controls are in place for one process, a further process or portfolio in the same organisation can reuse much of that work.

By sector

Banking and consumer finance: a small Deloitte survey of 30 US banking executives names integration with other systems and data security as the leading modernisation challenges, and Capgemini finds agents at scale in only one in ten financial-services organisations. Contact processes in this sector are typically linked to account, arrears and consent data held in core systems.

Insurance: KPMG’s research suggests that self-assessed AI leadership is far more common than strong data foundations, which limits how much of a claims or renewal process can be automated end to end.

Telecommunications, healthcare and energy: the research we hold does not measure integration barriers separately for these sectors. Governance findings for telecoms are set out under Regulation and trust.

What it means in operations

A contact process that cannot read and write the client’s systems can only hold conversations; one that can is able to complete cases. For buyers, the practical questions are who builds and maintains the connections, who is accountable when a hand-back fails, and how changes are approved and logged.

A managed-service model places part of that responsibility with the provider, but it does not remove the client’s own security, data-protection and change-management obligations.

What to watch

Whether the 37% of US banks that planned to introduce generative AI in contact centres during 2026 (Deloitte, June 2026) reach production in 2027.

Obligations for high-risk AI systems listed in Annex III of the EU AI Act, which legal commentary recorded in September 2026 expects to apply from 2 December 2027 after the Digital Omnibus amendment; relevant where AI is used to assess creditworthiness or access to services.

Forrester’s prediction that one in five companies will deploy fully agentic orchestration of customer journeys in 2027, which depends on the integration work described above. Forecast.

Vee in context

Vee’s Q3 2026 report (letter to shareholders, p. 7) cites the Deloitte banking finding and states that the capacity to use technology effectively in a real operating environment – integration with existing systems, security, running processes at scale and achieving measurable business results – is gaining importance. The letter adds (pp. 8–9) that developing international relationships required adapting technology, carrying out integrations and preparing processes, and that cooperation with international enterprises requires reliability, security, process quality and integration capability. It describes Land & Expand (p. 10) as building on experience, prepared integrations and knowledge of processes, with each further stage depending on operating results, contractual agreements and partners’ business decisions.

Market data on this page come from third-party publications and are attributed to their authors. They describe the market, not Vee’s results. Forecasts are the publishers’ own and are not forecasts by Vee S.A. Figures were checked against the publishers’ pages on the date shown.

Other market insights

The AI value gap

Spending on AI in customer service is rising, but few organisations report the return they expected. The gap sits between deploying a model and running a process that delivers a measurable result.

Read the insight

The human route

Automation is accepted when it solves the issue and when a person is reachable if it does not. Enterprises are redesigning human roles rather than removing them.

Read the insight

Contact data

Organisations record almost every call, yet analyse only a small fraction. Process data is where measurable improvement comes from.

Read the insight

Regulation and trust

In the European Union, transparency in interactions with AI is now a legal obligation. Regulated sectors expect controls, documentation and clear disclosure that a customer is speaking with an AI system.

Read the insight

Service economics

Budgets are moving towards AI, but the cost of automated resolution is not falling to zero. Buyers increasingly ask what a process achieves, not how many minutes it uses.

Read the insight