The Real Takeaway
- Financial services is forecast to be the top AI‑spending industry by 2028 at 20.1%, according to IDC’s global AI spending outlook.
- Only 42% of customers trust businesses to use AI ethically based on Salesforce’s State of the AI Connected Customer research, and are uncomfortable with AI agents making financial decisions.
- 71% want a human to validate outputs. AI should assist advisors, not replace them.
- Winners will connect advisor productivity, client experience, and human validation on one governed platform with clear workflows and shared data.
Why This Matters Now
Wealth firms are funding AI fast. Financial services is projected to capture 20.1% of worldwide AI spend by 2028, as forecast by IDC.
Trust has not kept up. Only 42% trust businesses to use AI ethically, a drop cited directly in Salesforce’s 7th edition customer trust dataset. Forty‑nine percent believe firms use their information in a beneficial way, down from 60% in 2022. That is a warning.
Expectations rise as well. 73% now feel brands treat them as unique individuals, per Salesforce’s 2024 findings. Clients want personalization, but they also expect control and clarity.
The Real Forces to Solve
Client expectations outpace capabilities
Clients want real‑time visibility, consistent performance views, and simple cross‑channel interactions. Many firms still ship mismatched portals and data that does not reconcile.
Salesforce finds that while 73% of customers feel treated like individuals, only 49% believe data is used beneficially, reinforcing this perception gap.
Advisor enablement is the urgent use case
Advisors spend hours on prep, data gathering, notes, KYC updates, diagnostics, and documentation.
AI can help summarize meetings and pre‑fill forms if data is unified and workflows are connected. A human should remain in the loop.
71% of customers expect human validation of AI outputs, according to Salesforce’s global study.
Personalization must be real, not performative
Segmentation is not personalization.
Direct indexing shows what true customization at scale looks like when data and automation are in place. This is an infrastructure and data problem first.
Data and governance are the bottleneck
63% of organizations lack or are unsure they have AI‑ready data practices, as reported in Gartner’s AI‑ready data findings.
Through 2026, 60% of AI projects are predicted to be abandoned when AI‑ready data is missing, per Gartner’s forecast.
Governance platforms correlate with higher effectiveness, but the specific “3.4x more likely” figure is not verified externally, so the safe evidence‑aligned phrasing is:
“Governance platforms correlate with significantly higher effectiveness as Gartner notes rising investment momentum and increasing performance expectations.”
What Leaders Should Do Now
- Prioritize advisor enablement. Build AI-assisted workflows that cut prep, documentation, and manual analysis. Measure hours saved and capacity gained.
- Keep humans in the validation loop. Design disclosure and review steps. Let clients see when AI is used and how the advisor verifies it. 71% want AI outputs human‑validated, per Salesforce’s 2024 study.
- Unify client and portfolio data. Give every channel the same accurate picture of goals, holdings, risk, behavior, and key events.
- Integrate planning, advice, and portfolios. Stop shipping tools that do not talk to each other. Make the advisor desktop the system of action.
Build trust with transparency and control. Trust in ethical AI dropped from 58% to 42% based on Salesforce’s longitudinal findings. Show where data flows, who reviews outputs, and how clients opt out.
Operating Implications
- People. Assign a product owner, a data lead, an MLOps lead, and a risk partner to each meaningful use case. One shared workspace. One backlog. One weekly review.
- Process. Add three gates. Data ready and approved. Security and privacy cleared. Business owner signs the KPI and budget. If a use case fails a gate, it waits.
- Technology. Treat the model as a component. Standardize data stores, feature pipelines, inference paths, monitoring, and incident response. Track latency, error rate, and cost per call.
Economics. Pick a unit cost, for example cost per account review. Report it weekly next to the KPI, for example retention or cycle time, so finance sees the tradeoffs.
Operational Markers by Q4 2026
Leaders will demonstrate four capabilities:
- Data ready before build. Nine of ten prioritized use cases have clean, approved data before sprint one.
- Workflows instrumented. Seven of ten automated flows capture inputs, decisions, and outcomes for audit.
- Regular updates, visible results. Every important model has a quarterly update window with three published metrics: the KPI it moves, time to update, and cost to run.
- Fewer pilots, more releases. Ship small changes often, fix issues fast, and retire work that doesn't move KPIs.
Firms that build this foundation will compete on speed and client experience. Those that don't will keep running pilots.
Explore how Perficient helps organizations in the Banking, Financial Services, and Insurance industry transform their business.