RIAs Embrace AI, But Clear Strategies Still Lag, New Reports Find

RIA AI adoption is climbing, but strategy is lagging

Artificial intelligence is in the room. But without a clear plan, most firms are stuck in pilot mode and leaving results on the table.

Two new studies make the point. Charles Schwab's "Advisor AI in Action" surveyed 533 advisors across its network with a median firm AUM of $375m. F2 Strategy's "Wealthtech Outlook Report" gathered input from senior leaders across 85 RIAs, wealth managers, broker-dealers, and asset managers representing over $61tn in AUM.

The takeaway: 63% of advisors are using AI in some form. About 10% are building AI into longer-term plans, 30% are experimenting, and the remaining 60% sit in the messy middle. Adoption is up. Strategy is thin.

Why progress stalls

• Unclear ownership: no single leader accountable for outcomes, risk, and scale.
• Pilots without a path: proofs-of-concept that never move to production.
• Data friction: scattered sources, unclear access, and weak labeling/governance.
• Compliance hesitation: ad hoc reviews slow everything down.
• Vendor sprawl: overlapping tools and rising cost with little consolidation.
• Skills gap: advisors dabble, but teams lack repeatable workflows and prompts.

A 90-day plan for executives

• Pick three high-ROI use cases with low risk: client communications, meeting notes/CRM updates, and research summarization.
• Appoint an AI product owner with a clear mandate and budget.
• Set guardrails: approved tools, data access, human review points, and logging.
• Baseline metrics: time per task, cycle time, error rates, and compliance flags.
• Pilot with 10-20 users, weekly check-ins, and a go/no-go decision at day 45.
• If the pilot clears thresholds (time saved >25%, quality equal or better), move to phase two and expand to 25-40% of target users.

Operating model that scales

• Stand up a small AI Center of Enablement (risk, data, product, compliance).
• Run product squads by use case (e.g., Comms Copilot, Research Copilot).
• Create a prompt library with version control and monthly quality reviews.
• Build light MLOps/monitoring: usage, quality scores, drift, and incident tracking.
• Integrate with CRM and document systems to remove copy/paste work.

Governance that clears compliance

• Adopt a simple risk framework aligned to the NIST AI Risk Management Framework.
• Default to human-in-the-loop for client-facing outputs.
• Recordkeeping: prompts, model versions, data sources, and final outputs.
• Third-party due diligence: data retention, SOC 2, model transparency, and SLAs.
• Red-team sensitive use cases (advice generation, suitability, personalization).

Use cases producing value right now

• Client email and message drafting with firm-approved tone and disclosures.
• Meeting transcription, action items, and automatic CRM updates.
• Research and market summary generation with linked sources.
• RFP and due diligence questionnaire drafting from firm templates.
• Compliance pre-checks on communications before manual review.
• Code assistants for analytics/data teams to speed report updates.

Metrics that matter

• Time saved per task and cycle time reduction.
• Advisor and client satisfaction (CSAT/NPS) deltas.
• Error/rework rate and compliance exception rate.
• Adoption and active usage by team and use case.
• Attribution: revenue protected or won due to faster turnaround.

Common pitfalls to avoid

• Buying tools before defining problems and metrics.
• Skipping data hygiene; garbage in equals noisy outputs.
• Letting pilots run forever without success criteria.
• Ignoring IP and privacy in prompts and datasets.
• Underfunding change management and training.

Budget and resourcing

• Start with a small, cross-functional team (3-6 people) for 90 days.
• Budget for licenses, a secure workspace, and change management first; custom builds later.
• Consolidate overlapping vendors quarterly to keep costs in check.

Upskill the team

Advisors and ops teams need practical workflows, not theory. If you're building repeatable skills by role, this curated library can help: AI courses by job. For finance-focused tool options, see this roundup: AI tools for finance.

Bottom line

AI use is widespread, but depth is uneven. The firms that will pull ahead are treating AI like a product: tight scope, fast pilots, clear guardrails, and measurable outcomes. Pick the work that moves the needle, prove it in weeks, then scale with discipline.