Why AI Onboarding Projects Stall - and
How Banking Leaders Can Avoid It

Intro

Banking leaders are spending billions on artificial intelligence use cases driven by copilots and agentic AI, yet many customers still wait days or even weeks for investment accounts to be opened, assets to be transferred and commercial lending relationships to become operational. The challenge isn't a lack of AI capability. It's that many institutions start by deploying AI tools rather than redesigning the onboarding workflow those tools are meant to improve. Banks that modernize onboarding can improve advisor productivity, accelerate revenue and deliver a better customer experience—but only if they approach it strategically.

Key Insight: Successful onboarding initiatives don't begin with AI. They begin with understanding where operational friction prevents advisors, operations teams and customers from moving forward.

Need to know

• Start with operational bottlenecks, not AI capabilities. The highest-value use cases are usually hiding inside manual onboarding workflows.
• Automate administration while preserving human judgment. AI should remove repetitive work so advisors can spend more time advising clients.
• Deliver measurable value within 90 days. Successful institutions focus on one onboarding workflow before expanding AI more broadly.
• Measure advisor capacity, onboarding speed and error reduction, not simply lower costs.
• Business leaders should own onboarding transformation. Technology should support the operating model rather than define it.

Start with the workflow, not the technology

Many AI initiatives fail because organizations start with technology instead of identifying where onboarding creates unnecessary friction.

Successful institutions reverse that process: Citigroup recently disclosed that AI reduced the time required to review account-opening documents from more than an hour to roughly 15 minutes.

The greatest opportunities lie in high-value onboarding processes where documentation, KYC, account transfers and regulatory reviews create significant operational overhead. More importantly, onboarding is not just a customer experience issue, it's an advisor and operations challenge.

Key Insight: The primary users of onboarding AI are often advisors and operations teams, not just customers.

Actions to take

• Prioritize complex onboarding journeys such as wealth management, institutional accounts and commercial lending.
• Map the onboarding process from application through account activation before evaluating technology.
• Identify every manual touchpoint where employees repeatedly enter, review or reconcile information.
• Focus initial AI investment on one high-volume workflow rather than attempting enterprise-wide transformation.

Focus AI on the activities that create the greatest operational burden

Document digitization and multimodal extraction can capture client information—including names, addresses, tax details and entity structures, as well as investment positions such as tickers, CUSIPs and holdings directly from brokerage statements. That information can then populate ACAT transfers, DTCC forms and internal onboarding documentation automatically. This eliminates repetitive data entry across multiple systems. The same workflow can also incorporate validation, eligibility and fraud checks, including cross-document and database consistency checks, so that operations teams can focus their attention on exceptions.

It also reduces the risk of misclassifying family relationships, trusts, 529 plans and other ownership structures that create downstream KYC and compliance issues.

Why it matters: Every manual data entry task introduces delay, cost and the potential for human error.

Actions to take

• Deploy document digitization and multimodal extraction to capture client and account information automatically.
• Create standardized extraction rules that consistently map client information into onboarding systems.
• Automate population of KYC documentation, ACAT forms, DTCC documentation and internal onboarding workflows.
• Eliminate duplicate data entry across operations, compliance and advisor teams.

Design human review into the process from day one

Confidence scoring allows high-confidence information to flow automatically while routing ambiguous cases for human review. In identity verification, for example, facial matching across multiple documents can help determine which applications can proceed automatically and which require manual validation.

Human review remains essential for complex scenarios such as trust and estate structures, proprietary money market funds and regulatory exceptions. AI removes administrative work so advisors can spend more time advising clients, not replacing professional judgment.

It also reduces operational risk. A single incorrectly entered position, mislabeled entity or KYC error can create regulatory issues, financial losses and reputational damage.

Key Insight: AI should scale expertise, not automate away accountability.

Actions to take

• Implement confidence scoring across every extraction workflow.
• Route only exceptions and low-confidence cases to experienced employees.
• Define which onboarding decisions always require human approval.
• Use AI to reduce errors while maintaining advisor accountability for final decisions.

Deliver measurable results within 90 days

Many onboarding initiatives become year-long transformation programs that struggle to demonstrate business value.

Rather than attempting end-to-end modernization, leading institutions deliver one measurable capability within the first 90 days, such as automating KYC extraction or deploying a reusable information-extraction service.

Why it matters: Short implementation cycles build momentum while reducing implementation risk.

Actions to take

• Define one onboarding workflow to automate during the first 90 days.
• Deliver a functioning capability in a test environment before pursuing enterprise deployment.
• Build reusable extraction capabilities that can support future onboarding initiatives.
• Expand only after measurable operational improvements have been demonstrated.

Measure business outcomes - not just technology success

According to the Investment Adviser Association, SEC-registered investment advisers served 7.7% more clients in 2024 while assets under management grew 22.3% to record levels. Improving onboarding capacity is becoming essential to supporting that growth.

Success should be measured through faster onboarding, reduced manual effort, shorter time to fee revenue, greater advisor capacity and fewer onboarding errors.

Many institutions assume existing portfolio management platforms will eventually solve onboarding challenges. In practice, that's rarely their priority. Banks should also avoid waiting for portfolio management vendors to solve onboarding. Most remain focused on portfolio analytics rather than operational workflows.

Key Insight: AI projects succeed when they solve business problems, not when they showcase technology.

Actions to take

• Measure onboarding cycle time, advisor productivity, customer satisfaction and time to fee revenue.
• Track reductions in onboarding errors, regulatory exceptions and manual effort.
• Put business leaders in charge of onboarding modernization rather than central technology teams.
• Avoid relying on generic AI tools or waiting for existing platform vendors to develop onboarding capabilities.

The bottom line

Institutions that begin with operational bottlenecks, automate repetitive work, preserve human judgment and deliver measurable improvements within 90 days will create advantages that extend well beyond lower operating costs. They'll improve advisor productivity, reduce operational risk and deliver the faster onboarding experience customers increasingly expect.