Salesforce · Financial Services Cloud · Agentforce
Salesforce AI data readiness for financial services firms
For RIAs, broker-dealers and trust companies on Financial Services Cloud that want Agentforce, Prompt Builder or tools like Claude to give advisors answers they can trust, without exposing client data they should not.
Short answer
AI in Salesforce is only as good as the household, account and activity data under it. Vantage Point audits your Financial Services Cloud org, fixes the records and permissions AI will read, and sets up the grounding data and audit trail your compliance team will ask for. The assessment costs $7,500 and takes 2 weeks.
Why data comes first
Every Salesforce AI for financial services use case, from meeting prep to service summaries, reads the same client records. Agentforce and Einstein features answer from the records in your org. If a client sits in two households, if account balances are a week old, or if notes are stored as free text on the wrong record, an AI summary will repeat those errors to an advisor with full confidence.
Permissions matter as much as quality. An agent that can read every note in the org can surface a note a junior associate was never meant to see. Readiness means the AI sees the same data the user is allowed to see, and nothing more.
AI inside Salesforce records, with or without Agentforce
You do not need Agentforce licenses to put AI to work on client records. Once the data is ready, three routes are open:
| Route | Good for |
|---|---|
| Prompt Builder and Einstein features | Summaries and drafts generated from fields on the household, account or case record, shown inside the page layout advisors already use. |
| RecordIQ, our Salesforce component | Firms not ready for an Agentforce program. RecordIQ sits on any record page, sends the record and its related data to the model you choose, and runs without an Agentforce or Einstein license while respecting sharing rules and field-level security. |
| Agentforce | Agents that take actions, such as updating a case or booking a review meeting, under rules you set. Start with our free Agentforce Discovery and POC Accelerator. |
| Claude or ChatGPT connected to Salesforce | Firms that already license Claude or ChatGPT and want advisors to query CRM data from there, with access limited to what each user can see in Salesforce. See Claude or ChatGPT Team rollout. |
The data governance work is the same for all three. That is why we do it first, and why our Data Foundations accelerator comes with prebuilt household checks, duplicate rules and a field inventory for Financial Services Cloud, so the assessment starts in week one rather than week three.
What we check
| Area | What good looks like |
|---|---|
| Households and relationships | One household per family, primary member set, no orphan accounts. Relationship groups match how advisors think about clients. |
| Financial accounts and holdings | Custodian data current and matched to the right household. See custodian data integration. |
| Activity and notes | Meeting notes on the right record, with dates and owners, in fields an AI can read. |
| Sharing and field security | Role hierarchy, sharing rules and field-level security tested against real users, including what an agent can reach. |
| Consent and communication preferences | Opt-outs and contact preferences stored where automation and agents check them. |
| Data 360 | Whether you need it now, and which sources to unify first if you do. |
| Supervision and logging | Prompts, responses and agent actions logged and retained in a way your compliance team can review. |
What you get
- A scored readiness report covering data quality, permissions, consent and logging, with the record counts behind each score.
- A fix list ranked by risk and effort, with the items we can fix during the assessment marked done.
- Two or three first use cases that your data can support today, such as meeting prep briefs or service case summaries.
- A compliance pack describing what the AI can read, where outputs are stored and how supervision works, for your CCO.
Supervision and recordkeeping
FINRA's Regulatory Notice 24-09 reminds member firms that existing rules, including supervision and recordkeeping, apply to generative AI the same way they apply to any other technology. The SEC's books and records rules apply to advisor communications whatever tool drafted them.
We design the setup so AI drafts are reviewed before they reach a client, prompts and outputs are retained, and access follows the same role model as the rest of your org. For a deeper build, see AI compliance and supervision.
Start with the Data Foundations accelerator
Most banks, lenders, insurers and wealth firms don't need to buy Agentforce on day one. They need records an AI model can trust. Our Data Foundations Discovery and POC Accelerator is where that starts, and the discovery is complimentary.
- Who it's for. Financial services teams on Salesforce, including those that own or are evaluating MuleSoft, Informatica or Data 360.
- How long. Three sessions over about four weeks, or four complimentary hours.
- What you get. A data health scorecard, an integration architecture map, an MDM and golden-record strategy, a Data 360 readiness assessment, a phased 90-day roadmap and a 12 to 15 page findings report.
- The POC. We fix one priority issue on your data: connect one system, deduplicate one entity type or unify one segment.
Once the records are clean, you can run AI inside them in the way that fits your budget: Prompt Builder summaries, our RecordIQ components, Claude or ChatGPT through a governed connector, or Agentforce when you're ready. See the Data Foundations accelerator for the full scope.
Pricing
| Package | Price | Timeline |
|---|---|---|
| AI Readiness Assessment | $7,500 | 2 weeks |
| Salesforce Org Assessment | $5,000, credited to follow-on work | Quoted |
| Data 360 Foundation | $15,000 | Quoted |
| Data 360 Growth | $32,000 | Quoted |
| AI Supervision and Recordkeeping | $25,000 | 4 to 6 weeks |
Proof
Financial services only. Most of our clients are wealth managers, banks, insurers and asset managers. We have built Financial Services Cloud orgs, custodian feeds and AI pilots for them, so we know where the data usually breaks.
Talk to a client. Tell us about your firm and we will introduce you to a client running Financial Services Cloud. Ask for a client reference.
Frequently asked questions
What does AI data readiness mean in Salesforce?
It means the records, permissions and logging an AI feature depends on are accurate and controlled. In Financial Services Cloud that usually means clean households, current account data, notes on the right records, tested sharing rules and a log of prompts and outputs.
Do we need Data 360 before using Agentforce?
Not always. Many first use cases can run on data already in your org. Data 360 makes sense when you need to unify several sources, such as custodian, planning and marketing data, before an agent can answer well.
How long does an AI readiness assessment take?
Our AI Readiness Assessment takes 2 weeks and costs $7,500. Fixes found during the assessment are quoted separately or credited into a follow-on package.
Can an AI agent see data a user should not?
It can if permissions are loose. We test sharing rules and field security against real user profiles and the agent user, so the AI respects the same limits as the advisor.
How do we supervise AI output for FINRA and SEC purposes?
Keep a human review step before anything reaches a client, retain prompts and outputs, and treat AI drafts as communications under your existing policies. We document the setup for your CCO.
Which AI use cases should a wealth manager start with?
Internal ones with low client risk: meeting prep briefs, call summaries for the advisor, service case summaries and data quality checks. Client-facing drafting comes after supervision is in place.
Can we use AI on Salesforce records without Agentforce?
Yes. Prompt Builder and Einstein features can summarize and draft from record data, and Claude or ChatGPT can be connected to Salesforce with access limited to each user's permissions. The data readiness work is the same either way.
Is this only for wealth managers?
No. We run the same data readiness work for banks, credit unions, lenders and insurers on Salesforce. The checks change by line of business, such as households for wealth firms and borrower and loan records for lenders, but the method is the same.
Do we need Agentforce to get value from AI in Salesforce?
No. Once the data is clean, Prompt Builder, governed connectors to Claude or ChatGPT, and in-record components can summarize and draft inside the record. Agentforce becomes an option when you want agents that take actions.
Updated November 11, 2026.