Sales enablement AI that answers every rep’s fund question from approved content only, each sentence cited to its page
Reps get a cited answer in seconds, only from approved, current content, in any language.
See one case, screen by screen ↓Sales enablement AI gives sales reps and wholesalers the answers they need to talk to clients, where sales enablement software mostly gives them content to search. Sales Enablement AI from Botminds does that for asset managers with seven agents: they answer a rep’s fund question in seconds, in the rep’s own language, only from documents compliance approved and that are still current, with every sentence cited to its page. Questions without an approved answer go to a product specialist with a drafted reply, and compliance signs it before any rep sees it.
AI in asset management: a buyer's guide to agentic AI for investment managers →
Updated
A client asks a sales rep for one number: the fund’s duration against its benchmark, a share class’s ongoing charges, whether the fund holds AT1 bonds. The rep searches the content portal, a shared drive and last month’s email, then messages a product specialist and waits hours for a reply. Factsheets change monthly, and KIDs, prospectuses and commentaries on their own clocks, so the copy a rep finds on a laptop is often superseded. If the client is in Milan or Frankfurt, the rep translates the answer too.
The firm answers for every word. FINRA’s AI guidance is plain: firms are responsible for their communications whether a person or AI wrote them, and chatbot communications must be supervised. Yet the answers reps give are rarely seen by compliance: a portfolio manager’s desk note, a forecast or a fund not registered in the client’s country slips into an email. And when a factsheet is corrected, nobody can list the reps who already quoted the wrong figure.
Estimated split of the demo firm's specialist time, about 1,000 hours a month — 20,000 rep questions, one in four sent to a specialist at about 12 minutes each — in 8-hour days, by hand and with the solution.
Seven agents read, answer, check and translate only approved content; specialists write, compliance signs.
Wednesday, October 21, 2026, 9:42 a.m. in Milan. Giulia Ferrante, a sales rep at Quillfield Capital, has a call at 10:30 with Paolo Venturi, fund selector at Banca Corvara Private. He emailed yesterday about duration and AT1 bonds. Here is how she works with the agents, screen by screen, in the working solution.
Giulia types “Qual è la duration del Quillfield Euro Short Duration Fund rispetto al benchmark?” Her client is already in context from the CRM: Banca Corvara Private, professional, Italy. The agents work in view: the Question agent finds the fund, share class I EUR Acc and that the fund is registered for sale in Italy; the Answer agent searches 412 candidates for this fund, audience and country and uses 4; two are not allowed — an internal desk note and a superseded factsheet.
On screen: “Searched approved, current documents only.”
“As of September 30, 2026 the fund’s modified duration is 1.8 years, against 2.1 years for the benchmark: 0.3 years shorter.” Each sentence is cited to its page — the September 30 factsheet, the Q3 commentary, the Italian KID’s risk indicator of 2 out of 7 — and the required wording for professional clients in Italy, DISC-IT-PRO-04, is attached as approved. A chip reads “OK to share · professional clients · Italy”.
Hover a sentence in Italian and its English pair lights up.
The Guard agent checked 5 sentences against their pages: 4 kept, 1 removed. Each kept sentence lists what was checked — the numbers match page 1 exactly, the as-of date is stated, the approval is current today. Under it sits the approved shelf: the factsheet MKT-2026-0912 valid to November 13, 2026, the commentary, the disclosure and the KID, each with its approval ID, page and expiry, and the two documents the agents were not allowed to use.
Removed: “The team expects to lengthen duration by year-end” — “From an internal desk note that is not approved for clients, and it is a forecast.”
Her follow-up: does the fund hold AT1 (CoCo) bonds? The client’s mandate excludes them. The Answer agent finds that the prospectus allows up to 10% in contingent convertibles, but no approved statement of what the fund holds today. It names who knows — Marcus Hale, Fixed Income — and offers one next step. Giulia clicks Ask a specialist: ESC-2026-4471, due 1:45 p.m.
“I will not answer from memory or from internal data.”
The Escalation agent routed the question to Marcus and drafted the reply from the approved holdings disclosure of September 30 and the prospectus: the fund held no AT1 bonds. It also found two similar open questions, ESC-2026-4462 in German and ESC-2026-4466 in French. Marcus adds bank senior bonds at 14.2% of net assets from the same page and submits it for compliance.
Service level: 4 business hours, 8 a.m. to 6 p.m. Milan; past it, Daniel Whitcombe is told.
Renée opens the reply with its sources and approves it with reason code A2 · Accurate after edits, because Marcus changed the agent’s draft. It becomes approved answer MKT-2026-0957 for professional and institutional clients in every registered country, valid until the October 31 holdings disclosure is approved. Marcus wrote it, so he cannot approve it.
“Signed as Renée Okafor · Compliance Reviewer · 10:52 a.m. Milan · recorded on the audit trail.”
The Library agent adds the answer to the approved library and the Language agent writes the Italian, German and French. Giulia, Lukas Brenner, Chloé Dubois and Sofía Marín, who asked the same question, are told. In Giulia’s conversation it arrives cited to the holdings disclosure, the prospectus and the disclosure for professional clients in Italy, with 3 of 3 sentences kept by the Guard agent.
Every rep who asks next gets it in seconds.
Marketing withdrew the Global High Yield factsheet MKT-2026-0903 on October 20: it printed the effective duration as 3.9 years, and the corrected MKT-2026-0915 says 3.4. The Library agent compared both field by field, found 37 answers to 29 reps since October 12 that used the old figure, and drafted a correction for each rep in their language. 12 of those answers were copied into an email or logged to the CRM, so those client records get a CRM note.
“Nothing goes out until you approve.”
From 4,486 answers between October 12 and 16, the Log Review agent drew 120 — 60 chosen by risk, 60 at random — and flagged 7 in three patterns, among them two reps who rephrased a refused forecast question to get a view on rate cuts. Renée decides each item as OK or a finding with a reason code, then signs the week; the review, its findings and the sample join the evidence pack.
Signing records that the sample was reviewed under FINRA Rule 3110 supervision procedures.
From October 1 to 21, 13,613 questions came from 571 of 600 reps in 11 languages, and 89.2% were answered from approved content. Insights groups what the library cannot answer yet: AT1 holdings by bond fund is the biggest gap, 63 questions in 4 languages. Daniel can commission a monthly approved answer per bond fund from the gap itself; the Escalation agent can draft all 12 for specialists to edit.
“Specialist replies are on time 96.4% of the time; Private Markets is slowest at 4.1 hours.”
The same morning, seen by five of the people who carry it — what their week looked like, and what it looks like now.
Seven agents read, answer, check and translate only approved content; specialists write, compliance signs.
Reads each question: its language, the fund and share class, the client’s category and country from the CRM, and whether the fund is registered for that country and audience.
Searches only approved, current documents permitted for this audience and country, and drafts in the rep’s language, one citation per sentence. With no approved source, it says so and offers a specialist.
Checks every sentence against its page — numbers, as-of dates, approval current today, audience and country, no forecast — and attaches the required wording.
Writes the English beside answers in other languages, back-translates to check the meaning, and reads sources in their original language.
Routes questions without an approved answer to the right product specialist with a drafted, cited reply, tracks the service level and merges duplicate questions across languages.
Watches the content library for new versions, expiries and withdrawals, proposes changes, and recalls every answer that cited a corrected document, with a correction drafted in each rep’s language.
Draws the weekly risk-based sample of the answer log, groups patterns and drafts findings for compliance.
Shares, writes, approves. The agents propose; a named person decides.
Reps ask about any fund in their own language. On every other screen, the agent that owns it speaks first and answers questions about it, with the page or the record it came from.
Posso proporre il Private Credit Fund IV a un cliente al dettaglio?
No. Quillfield Private Credit Fund IV is offered to professional investors only and cannot be offered to a retail client in Italy. The offering memorandum says so on page 2.
Why was the Global High Yield factsheet withdrawn?
Marketing withdrew MKT-2026-0903 on October 20: it printed the effective duration as 3.9 years. The corrected factsheet MKT-2026-0915 says 3.4 years, and Renée Okafor approved it at 5:48 p.m. 37 answers to 29 reps used the old figure; opening the recall.
How many of these reached a client?
12 of the 37 answers were copied into an email or logged to the CRM, so a client may have seen 3.9 years. When you approve, those 12 client records get a CRM note with the corrected figure, queued while the CRM is delayed.
Draft the AT1 answer for all 12 bond funds
Plan: I draft one answer per bond fund from its September 30 holdings disclosure, every figure cited to its page. Marcus Hale edits them as the specialist, compliance approves each one, and the Library agent adds them, valid until the October disclosures. Nothing reaches a rep before compliance signs.
13 screens from the working solution, on its sample data. Pick one to see it large.
Rules R-01 to R-10 and the thresholds the agents work to are settings, each versioned and owned. A change is a proposal, tested on a golden set of 250 questions in 11 languages and approved by compliance — never by the person who proposed it.
| Setting | Default | Choose from |
|---|---|---|
| Guard support needed to answer | 90% | Below this the whole answer is blocked |
| Confidence to answer without a specialist | 85% | Below this the question goes to a specialist |
| Numbers must match their page | Exactly | No rounding beyond the page’s own |
| Oldest monthly data allowed | 45 days | After month-end, until the new factsheet is approved |
| Similarity to propose merging escalations | 0.95 | Across languages |
| Weekly review sample | 120 answers | 60 by risk, 60 at random |
| Specialist service level | 4 business hours | Desk hours, 8 a.m. to 6 p.m. Milan on business days; past it, the Head of Sales Enablement is told |
| Retention of the answer log and audit trail | 7 years | Advisers Act Rule 204-2 asks for 5; drafts never sent are kept 90 days; legal hold available |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (Forrester sales enablement research — 47% of organizations estimate reps spend 20 or more hours a month on content, 62% say reps look in six or more places; SEC Marketing Rule enforcement since November 2022, including press release 2024-121). Context: FINRA Advertising Regulation FAQs B.4 and D.8 (May 10, 2024), FINRA Rule 2210, FINRA Regulatory Notice 24-09. People, firms, clients and funds named on this page — including Quillfield Capital and Banca Corvara Private — are fictional: characters and sample data in the working solution.
Sales enablement AI gives sales reps and wholesalers the answers they need to talk to clients, where sales enablement software mostly gives them content to search. Sales Enablement AI from Botminds does that for asset managers with seven agents: they answer a rep’s fund question in seconds, in the rep’s own language, only from documents compliance approved and that are still current, with every sentence cited to its page. Questions without an approved answer go to a product specialist with a drafted reply, and compliance signs it before any rep sees it.
Sales enablement in asset management starts with the client: a rep asks in the conversation, with the client in context from the CRM. The Question agent finds the fund, share class and the client’s category and country; the Answer agent drafts from approved documents for that audience; the Guard agent checks every sentence and attaches the required wording; the Language agent writes the English beside. Under every answer sits the approved shelf: the documents used, each with its approval ID and expiry.
Yes. Only documents with an approval ID, status Approved and today within their validity can be used; expired, superseded and internal documents are never used, and the rep is offered the current version instead. In the working solution the Guard agent removed a sentence taken from a portfolio manager’s desk note before the rep saw the answer.
It works to your rules: performance only from a factsheet with the standard periods, net of fees and the past-performance warning; no forecasts or suitability advice; professional-only documents never answered for retail clients; and the approved wording for the client’s country and audience attached exactly. Every answer is kept with its sources and checks, and compliance reviews a weekly sample of the answer log; FINRA’s AI guidance expects chatbot communications to be supervised. Your compliance team decides how these fit your FINRA and SEC obligations.
The Library agent finds every answer that cited the old document, drafts a correction in each rep’s language and lists the client records that received the old figure. Nothing goes out until compliance approves; then each rep gets the correction by email and in their conversation, and a note is added to each client record in the CRM.
Yes. Reps ask in their own language — the working solution shows 11, including Italian, German, French, Spanish and Japanese — and get the answer in that language with the English beside it. Sources are shown in their original language with the English reading, and approved local disclosures are used as written, never translated by an agent.
Yes. The rep decides what to send a client, product specialists write new answers, and compliance approves every specialist reply, library change, correction notice, setting and agent version with a reason code. The agents never send anything to a client, never change the library and never approve; nobody approves their own work.
The Agentic Solution Engine builds and deploys this AI sales enablement platform from your requirements and documents — your approved library, approval register, CRM and fund data — and it goes live once every quality gate has passed, including a golden set of test questions in your reps’ languages. We will walk you through it on your own fund documents first.
Every consultant profile current each quarter, every figure tied to its source.
1-min demoPitch Book CreationFinals decks the same day, with every number stamped to its source.
1-min demoInstitutional Client OnboardingWon mandates funded in weeks, not months, with every signer and bank detail proven.
1-min demoWe’ll run Sales Enablement AI on your own approved library and a sample of the questions your reps ask.