How much could agentic AI reduce KYC processing time in banks?

According to a McKinsey & Company report covered by Mint on 27 October 2026, agentic AI could reduce the time banks spend on know-your-customer processes by 30 to 70 per cent.

Why is human oversight so costly in AI-driven banking tasks?

McKinsey's report states that human oversight accounts for 70 to 75 per cent of variable running costs per agentic task in banking, while token costs make up only 20 to 25 per cent.

What share of bank employees typically work on KYC and AML?

The McKinsey report says banks often assign 10 to 15 per cent of all employees to know-your-customer and anti-money-laundering activities alone.

← All news

McKinsey: Agentic AI May Cut Bank KYC Time by 70%

Invisipilot Team · 27 Oct 2026, 16:30 IST

Written with AI assistance from the cited sources

Agentic AI could reduce the time banks spend on know-your-customer (KYC) compliance by 30 to 70 per cent, according to a McKinsey & Company report published on 27 October 2026. The report, covered by Mint, finds that banks often dedicate 10 to 15 per cent of their entire workforce to KYC and anti-money-laundering tasks. Despite the promise, high human-oversight costs and weak governance frameworks are slowing adoption across financial institutions globally.

Key points

McKinsey says agentic AI could reduce bank KYC processing time by 30 to 70 per cent, according to a report covered by Mint on 27 October 2026.
Banks typically assign 10 to 15 per cent of all employees to KYC and anti-money-laundering tasks, the McKinsey report states.
Human oversight accounts for 70 to 75 per cent of variable running costs per AI agent task in banking, with token costs making up only 20 to 25 per cent, per McKinsey.
One bank's AI evaluations showed consistency ranging from 72 to 100 per cent; on the weakest test, roughly one in four runs returned a different answer to the same input, the report says.
McKinsey estimates optimisation measures could cut agentic AI operational costs by more than half, but warns that reducing quality checks in regulated banking requires compliance consideration.
Takeaway: Agentic AI is already in production in financial-crime prevention, but governance gaps and high oversight costs remain the primary barriers to wider adoption, according to McKinsey.

Source: Mint, 27 Oct 2026

What the McKinsey Report Found

A McKinsey & Company report, covered by Mint on 27 October 2026, says agentic AI — autonomous AI systems that can carry out multi-step tasks with limited human direction — has moved from experimental use to live production in financial-crime prevention faster than most banks anticipated. The report focuses on know-your-customer (KYC) processes, which verify the identity of bank customers, and anti-money-laundering (AML) controls.

According to the report, banks commonly assign 10 to 15 per cent of all employees to KYC and AML work alone, making these functions a significant cost centre and a natural target for automation.

The Numbers Behind the Promise and the Problem

McKinsey says agentic AI could cut KYC processing time by 30 to 70 per cent. However, the cost structure of running AI agents in banking is less favourable than it might appear. The report states that token costs — the direct computing charges for running an AI model — account for only 20 to 25 per cent of variable running costs per task, while human oversight accounts for 70 to 75 per cent. A further 60 per cent of the total cost of an agentic task is tied to refining answers rather than generating the first response.

One unnamed bank cited in the report ran ten evaluations of its customer onboarding process, setting minimum thresholds of 80 per cent accuracy and 70 per cent consistency. Measured consistency across published evaluations ranged from 72 to 100 per cent. On the weakest evaluation, roughly one run in four returned a different answer to the same input, the report notes.

Governance Gaps and the Limits of Optimisation

Despite evaluations being relatively low-cost — they do not require labelled data, the report says — many financial institutions still lack adequate metrics to assess how AI agents reason and behave. McKinsey suggests that optimisation measures, such as using less expensive AI tools for simpler tasks, eliminating duplicate operations, reducing data transfers between agents and shifting to sample-based quality reviews, could cut operational costs by more than half.

However, the report cautions that reducing the frequency of quality checks in regulated banking functions involves compliance and risk management decisions, not just technical ones. As banks expand autonomous AI use, reliable human supervision becomes harder to maintain. "The reason that having a human in the loop is no longer sufficient is supervisory," the report states.

Why It Matters for Job Seekers

If you work in banking compliance or KYC roles, consider building familiarity with AI-assisted workflows, as the report signals these functions are being restructured around agentic tools.
If you are applying for roles in AI governance or risk management at financial institutions, the McKinsey findings highlight a documented shortage of oversight frameworks, pointing to growing demand for professionals who can design and audit AI evaluation processes.
If you are a technology professional targeting the financial sector, the report's emphasis on cost optimisation and agent evaluation suggests that skills in AI quality assurance and compliance-aware system design are increasingly valued.
If you are job-seeking in India's banking or fintech sector, this report reflects a global shift that Indian institutions are also navigating, making AI literacy in regulated environments a practical differentiator.

Audience note: This story is relevant to professionals in banking, compliance, AI governance and financial technology worldwide, including India.

Frequently asked questions

How much could agentic AI reduce KYC processing time in banks?: According to a McKinsey & Company report covered by Mint on 27 October 2026, agentic AI could reduce the time banks spend on know-your-customer processes by 30 to 70 per cent.
Why is human oversight so costly in AI-driven banking tasks?: McKinsey's report states that human oversight accounts for 70 to 75 per cent of variable running costs per agentic task in banking, while token costs make up only 20 to 25 per cent.
What share of bank employees typically work on KYC and AML?: The McKinsey report says banks often assign 10 to 15 per cent of all employees to know-your-customer and anti-money-laundering activities alone.