What problem does it solve?
A wealth advisor is expected to know the firm's view on markets, sectors and asset classes, the terms of hundreds of products, tax and structuring notes for several jurisdictions, and the policies that govern what may be offered to whom. That knowledge lives in research libraries, product term sheets, intranet pages and email, and it changes often. Finding the right paragraph while a client waits is slow, so advisors lean on memory, ask a colleague or promise to call back.
The result is inconsistent answers, lost time before and during client conversations, and a real conduct risk when an advisor paraphrases an outdated view or a product rule from memory. Junior advisors and relationship managers in new markets are hit hardest, because they do not yet know where anything lives.
How does it work?
- Curate the corpus. Research notes, the house view, product documents, tax and structuring notes and policies are loaded into a knowledge base with an owner, a publication date and an audience per document.
- Ask in plain language. The advisor asks "what is our current view on European banks" or "can this structured note be sold to a client in Hong Kong" in chat or inside the collaboration tool.
- Retrieve and answer with citations. Hybrid search finds the relevant passages; the model answers only from them and links each statement to its source document and date.
- Respect entitlements. The assistant only retrieves documents the advisor may see, and it refuses questions it cannot answer from approved content instead of guessing.
- Learn from feedback. Thumbs down, unanswered questions and stale document hits go to the content owners, who fix the corpus rather than the prompt.
- Audience
- Employee facing
- Autonomy
- Assist
- Adoption
- Mainstream
- Channels
- Internal tools, Microsoft Teams
What is it worth?
Benchmarks are computed from the public deployments below: one data point per organization per KPI, with who made each claim.
| KPI | Median | Reported range | Data points | Claimed by |
|---|---|---|---|---|
| Interactions handled | Not pooled | at least 23 million | 1 | 1 organization |
Value drivers: Employee productivity, Compliance quality, Customer experience, Speed and cycle time.
Indicative value
A wealth manager with 500 client facing advisors
USD 460,000 to USD 8.6 million
Value of advisor time released from searching per year
How this is calculated
Formula: advisors * questionsPerWeek * minutesSaved / 60 * weeks * hourlyCost. The low scenario uses every low input, the high scenario every high input.
| Input | Low | High | Basis |
|---|---|---|---|
| Client facing advisors advisors, advisors | 500 | 500 | The reference firm. |
| Knowledge questions per advisor per week questionsPerWeek, questions per advisor per week | 5 | 15 | Editorial assumption, replace with your own search and help desk volumes. |
| Minutes saved per question minutesSaved, minutes per question | 3 | 10 | Conservative against the benchmark on this page (J.P. Morgan reports advisers find the right information up to 95% faster); most questions are short lookups. |
| Working weeks per year weeks, weeks per year | 46 | 46 | Editorial assumption. |
| Fully loaded advisor cost per hour hourlyCost, USD per hour | 80 | 150 | Editorial assumption, replace with your own fully loaded cost. |
What it leaves out: Time released is only value if advisors spend it with clients. The estimate leaves out the cost of running the assistant and curating content, and the harder to measure benefit of fewer answers given from outdated material.
Who already uses it?
6 public deployments, strongest evidence first. Grades: A regulator or audit, B the organization itself, C vendor case study, D anonymous or estimate.
Citi
United States · Wealth and asset management · 2025
Citi Wealth launched two AI tools built by its Data, Analytics and Innovation team. AskWealth is a generative AI assistant that gives service teams, advisors and managers answers across the wealth business, so that advisors can reach market insights and research when clients ask questions; after a launch in Asia it became available to Citi Wealth colleagues worldwide. Advisor Insights is a dashboard of timely messages about market moves, portfolios and events, including Chief Investment Office insights, piloted with Citigold and Citi Private Client advisors in North America with a wider rollout planned for Q4 2025 and Q1 2026. Citi says the tools will save hours of time but published no figures.
No outcome disclosed.
Bank of America
United States · Wealth and asset management · 2024
Merrill and Bank of America Private Bank teams use ask MERRILL and ask PRIVATE BANK, built on the technology behind Erica, to curate the information they need for clients. For more complex requests, the chat can connect teams with experts at the bank. The bank reports more than 23 million interactions with the two tools in 2024.
- Interactions handled: at least 23 million, calendar year 2024
"In 2024, there were more than 23 million interactions with ask MERRILL and ask PRIVATE BANK, an increase of 1 million over 2023, helping employees more proactively connect with clients about timely and relevant opportunities."
Claimed by: organization
Morgan Stanley
United States · Wealth and asset management · 2023
Morgan Stanley Wealth Management fully rolled out the AI @ Morgan Stanley Assistant in September 2023, a generative AI chatbot that gives Financial Advisors quick access to the firm's intellectual capital. The rollout followed the firm's March 2023 announcement of OpenAI as its strategic partner. In its June 2024 release the firm said that 98% of Financial Advisor teams had adopted it. It was followed by AI @ Morgan Stanley Debrief, which drafts meeting notes and follow up emails with client consent.
No outcome disclosed.
JPMorgan Chase
United States · Wealth and asset management · 2025
J.P. Morgan's private client advisers use an internal generative AI tool, Coach AI, to find research and content for client conversations more quickly. The firm's asset and wealth management chief executive credited its AI tools, which pull clients' trading patterns and anticipate their questions, with helping advisers respond to clients during the April 2025 market sell off. Its asset and wealth management chief information officer said advisers find the right information up to 95% faster. The firm's aim of growing adviser client books by 50% over three to five years is a target, not a result.
- Search time reduction: up to 95%, time to find information for a client conversation
"Our advisers are finding the right information up to 95% faster - which means they spend less time searching and more time engaging in meaningful conversations with clients"
Claimed by: organization
UBS
Switzerland · Wealth and asset management · 2024
UBS built two domain specific assistants, together called UBS Red, on Azure AI Search and Azure OpenAI Service to give client advisors fast, multilingual access to the bank's investment advice and product content during client work. UBS digitized about 60,000 investment advice and product documents into a queryable knowledge base, which it says saves considerable time in meeting preparation and research. Within 10 months the wider Azure OpenAI footprint reached key wealth, banking and operations divisions in the Switzerland, Hong Kong and Singapore booking centres. No usage or time saving figure specific to UBS Red is published.
No outcome disclosed.
Yes Bank
India · Banking · 2024
Yes Bank developed RM Assist (Ask Genie), an internal chatbot on Azure OpenAI that gives relationship managers answers to customer queries from the bank's repository of product and policy documents and helps them prepare product pitches. Microsoft's page, which also names Power Apps Copilot, presents it as a way to improve first time resolution of customer interactions. The benefits are described as expectations; no live usage or outcome figures are published.
No outcome disclosed.
How do you implement it?
A model agnostic playbook: what to prepare, the order to build in, and what goes wrong.
Data you need
- Research library and house view with publication dates and authors
- Product term sheets, key information documents and product governance rules per market
- Tax, structuring and policy notes with an owner and a review date
- Entitlement data that maps advisors to booking centres, licences and client segments
Systems to integrate
- Research and content management systems
- Document repositories such as SharePoint
- Identity and access management for document level permissions
- Collaboration tools such as Microsoft Teams, or the advisor desktop
Complexity: Low
The model work is simple; the effort is in the content. Research, product and policy documents need owners, review dates and audience tags, and entitlements must follow the advisor's booking centre and licence.
- 1
Start with one corpus and one audience
Pick the content advisors search most (usually the house view and product documentation) for one booking centre. Measure what they ask in the first weeks before adding more sources.
- 2
Fix the content before the model
Remove superseded documents, tag every document with an owner, audience and expiry date, and agree who updates it. An assistant that retrieves a superseded document gives a wrong answer with a valid looking citation.
- 3
Make citations mandatory
Every answer shows its source passages and dates, and the assistant refuses when retrieval finds nothing relevant. Advisors must be able to check an answer in one click.
- 4
Enforce entitlements at retrieval time
Filter documents by the advisor's permissions before they reach the model, so a restricted research note or a product not approved in that market never appears in an answer.
- 5
Build a test set from real questions
Collect a few hundred real advisor questions with approved answers and run them on every content or model change, including questions that must be refused.
- 6
Roll out with training and feedback loops
Train advisors that the assistant informs and they decide, publish usage and feedback per desk, and route every thumbs down to the content owner.
Guardrails
- Answers only from approved, dated content, with a refusal when nothing relevant is retrieved
- Citation of the source document and date on every answer
- Document level permissions applied before retrieval, per booking centre and licence
- No personalized recommendations; the assistant informs, the advisor advises
- PII masking so client names and account data are not sent to the model unless required
KPIs to instrument
- Weekly active advisors as a share of licensed advisors
- Share of questions answered with a citation versus refused
- Answer accuracy on a monthly human reviewed sample
- Median time to answer compared with the previous search process
- Feedback rate and top unanswered topics
Human in the loop
The advisor decides what, if anything, to tell a client and stays responsible for the advice. Content owners review flagged answers weekly and research or product governance approves any new corpus before it is added.
Common failure modes
- Stale house view
- The assistant quotes last quarter's view because the old note was never retired. Expire documents automatically and prefer the newest version at retrieval.
- Answers that look like advice
- Advisors paste a fluent answer into a client email without checking it. Train on the assist posture, keep citations visible and log what is copied.
- Entitlement leakage
- A restricted or wrong market document appears in an answer. Filter by permission before retrieval, not after generation.
- Adoption stalls after launch
- Advisors try it, get a poor answer and stop. Seed the corpus with the most searched content and publish improvements.
What are the risks and rules?
EU AI Act
Limited risk (transparency)
Article 50(1) requires that people who interact directly with an AI system are informed of it, unless this is obvious from the context, as it usually is for an internal assistant labelled as AI; Article 50(2) requires providers of systems that generate text to mark the output as AI generated in a machine readable way. Helping advisors find information is not an Annex III use and not a prohibited practice under Article 5. It would become high risk only if the system were used to evaluate the creditworthiness of clients (point 5(b)) or to evaluate or make decisions about advisors (point 4(b)). If the assistant were opened to clients, they would have to be told they are dealing with AI.
Guidance
- ESMA public statement on the use of AI in the provision of retail investment services (European Securities and Markets Authority, Europe). Decisions remain the responsibility of the management body whether taken by people or AI tools, including third party AI used by staff; MiFID II organisational, conduct and record keeping requirements apply.
- Artificial Intelligence (AI) Model Risk Management (information paper) (Monetary Authority of Singapore, Asia Pacific). Good practices observed in MAS's mid 2024 thematic review of banks' AI and generative AI model risk management, covering governance and oversight, key risk management systems and processes, and development and deployment.
- Principles to Promote Fairness, Ethics, Accountability and Transparency (FEAT) in the Use of Artificial Intelligence and Data Analytics in Singapore's Financial Sector (Monetary Authority of Singapore, Asia Pacific). Foundational principles for firms offering financial products and services on the responsible use of AI and data analytics, including internal governance, accountability and transparency.
Controls to put in place
- Entry in the AI inventory with an accountable business owner and a content owner per corpus
- Document ownership, audience tags and expiry dates enforced in the knowledge base
- Retrieval and answer logs kept for supervision and investigation
- Regression test set run on every model or content change
- Clear written rule that the assistant supports but does not replace the advisor's judgment
Frequently asked questions
- How many advisors actually use these assistants?
- Where firms publish figures, adoption is high. Morgan Stanley said in June 2024 that 98% of its Financial Advisor teams had adopted its assistant, and Bank of America reports more than 23 million interactions with ask MERRILL and ask PRIVATE BANK in 2024.
- How do you stop the assistant from giving wrong answers to clients?
- Ground it only in approved, dated content, show the source with every answer and make it refuse when nothing relevant is found. The advisor, not the assistant, talks to the client and checks the source first.
- Is this the same as enterprise knowledge search?
- It uses the same retrieval pattern, but the corpus and controls are specific to advice: research and house view, product governance per market, and entitlements by booking centre and licence. Those controls matter because the answers feed conversations with clients.
How to cite this page
Blits.ai AI Use Case Library, "AI knowledge assistant for wealth advisors and relationship managers", last verified 26 September 2026, https://www.blits.ai/ai-use-cases/wealth-advisor-knowledge-assistant. Licensed under CC BY 4.0. Method: how we verify use cases.
Changelog
- 27 September 2026: First published