Rms pulled their own lists.
Daily Excel exports from the folio system, manual sorts, gut prioritisation. Roughly 6 hours of analyst-equivalent work per RM per week.
All case studies
Distribution & Cross-Sell Ops Agents in Production
A top Indian digital wealth platform.
2.4×
Cross-sell take rate (1.4% to 3.4%)
View outcome below₹95crDormant AUA reactivated in 90 days
View outcome below~6hrPer RM per week saved on list-pulling (~500 RM hours/week reclaimed)
View outcome below
- Client
- A top Indian digital wealth platform
- Industry
- Direct mutual funds, bonds, insurance, fixed deposits
- AUA
- ₹9,500cr
- Active investors
- 320K
- RMs
- 85
- ARPU
- ~₹4,200
- Agents deployed
- Distribution & Cross-Sell Ops Agents
- Deployment
- 8 weeks from kickoff to all workflows live
The situation
A digital engine, an RM bench, a thin layer in between.
The platform had built a strong digital acquisition engine and a real RM bench, but the layer in between was thin. RMs spent almost half their time pulling lists in Excel, deciding who to call from gut, and updating the CRM after the fact.
The top decile of RMs operated at a level the rest could only reach with years of experience, and no one could teach what they did differently. Meanwhile, dormant AUA was building up quietly, and cross-sell was happening through generic monthly campaigns at single-digit conversion.
Leadership wanted to do two things at once: lift the floor of RM performance, and stop bleeding dormant AUA.
Where the workflow broke down
Cross-Sell was generic broadcast.
Monthly campaign emails to the whole book at 1.4% take rate. No personalisation, no RM context.
Dormant accounts fell through cracks.
Investors who had not transacted in 6–18 months got the same reactivation email everyone else did. Conversion sat at 0.2%.
RM coaching was reactive.
The bench could not pick up what the top RMs knew, so the top decile carried the team.
The deployment
Eight weeks. 16 hours of platform engineering.
Eight weeks from kickoff to all three workflows live. Read-only across folio system, transaction history, KYC database, CRM, and product analytics.
Engineering effort on the platform side: 16 hours.
Three workflows.
One connected platform.
Trigger
Daily refresh, 6:30 a.m.
Data sources
Folio system, transaction history, KYC, risk profile, product holdings, bureau pull, prior outreach outcomes
How it runs
For every active investor, ranks the next-best product (top-up SIP, debt fund, insurance, bond) by propensity × likely revenue × RM fit. Segments by RM allocation. RM dashboard auto-populates with a 12-customer daily queue, with rationale per customer.
Human in the loop
RM works the queue. Can override or skip with a one-click reason that feeds back to the model.
Output
RM dashboard + dialer queue + Slack alert if a high-AUA customer hits a buy-signal during the day.
Before
Generic monthly broadcast to the whole book at 1.4% take rate.
After
Personalised daily queue at 3.4% take rate.
Trigger
Weekly, Monday morning
Data sources
Folio system, last transaction, communication history, market context
How it runs
Scores dormant accounts (no transaction 6–18 months). Identifies probable reason using behavioural signal mix. Recommends a reactivation pathway: voice (high-AUA + high probability), RM warm call (medium), app push + email (low). Voice agent dials with a personalised opening referencing the investor last holdings and current market context. Schedules an RM follow-up if interest is detected.
Human in the loop
RM picks up the warm conversations. Growth ops reviews a 3% voice call sample weekly for tone and compliance.
Output
Booked RM follow-ups, app re-engagement, reactivation events logged to CRM.
Before
Generic reactivation email at 0.2% click-through.
After
Voice agent at 9% callback intent rate. 3.6% transaction within 30 days on the top-decile dormant cohort.
Trigger
On-demand, during RM customer calls
Data sources
Full customer context, market data, product catalogue, compliance rules
How it runs
RM types a question inline in the workspace. The agent surfaces the top three portfolio recommendations with rationale and compliance flags. Optionally listens in on the call (with consent) and offers prompts in real time for higher-value conversations.
Human in the loop
The RM is always in the chair. The agent is the assistant.
Output
Inline workspace recommendations + post-call summary with next-action.
Before
RMs relied on 5+ years of experience plus Excel. The top decile carried the team.
After
Bottom-quartile RMs now make recommendations at top-decile quality. The floor lifted without changing headcount.
Outcomes
2.4×
Cross-Sell Take Rate
1.4% to 3.4%, sustained across three consecutive months.
Before1.4%
After3.4%
- Sustained across
- 3 consecutive months
Customer quote
“Our top RMs were already 10x. Now our whole team operates at that level. The floor lifted without us hiring a single person.”
Chief Distribution Officer
From kickoff to production
Weeks 1–2
Scoping with distribution and RM leadership. Three workflows confirmed.
Weeks 3–4
Folio, transaction, and CRM read-only integration. Hosted environment provisioned with private inference.
Weeks 5–6
Cross-sell ranking and RM concierge live.
Weeks 7–8
Dormant voice agent live. All three workflows in production.
Common questions
How does the RM Concierge work inside the existing RM workspace?
The agent sits inline in the RM workspace via embed. The RM types a question, the agent answers with top three portfolio recommendations, rationale, and compliance flags. No context switch.
What controls govern the voice workflow?
A 3% voice-call sample is reviewed weekly. The agent identifies itself, names the platform, and never makes recommendations on the call — only flags interest and books a warm RM follow-up.
Does data leave the platform environment?
No. Hosted with private inference inside the platform cloud account. Read-only by default.
How are RM overrides handled?
The RM can skip any queue entry with a one-click reason. Those signals feed back into the ranking model the next refresh.
How long does deployment take?
Eight weeks for this deployment. Typical wealth-platform engagements land in the six-to-ten week range depending on the number of workflows.
Can this extend to NRI or HNI segments?
Yes. Both are in scoping. NRI is voice-heavy; HNI extends the RM concierge with portfolio stress-test inputs.
