Case Study · 10 / 10 RegTech · Market Intel Save as PDF ↓
Market intelligence agent · 5-week build

Eyes that don't blink.

How a Bangalore RegTech firm cut market scanning from 12 hours a day to 90 seconds — and won a ₹1.4 Cr tender they would otherwise have missed.

ClientTrustline RegTech LocationBangalore, India PublishedMay 2026
Industry
RegTech · KYC · AML · Banks
Scale
85 banks & NBFCs · Series B
Team
140 employees · ₹62 Cr ARR
Engagement
5 weeks · 47 sources
Outcomes · 90 days post-launch
12h → 90s
Analyst scanning load, from 12 hours / day to 90 seconds
94%
Signal-to-noise of items reviewed, up from 38%
₹4.7Cr
Pipeline value from agent-sourced tenders, first 90 days
§ 01 · The company

Compliance infrastructure for 85 banks & NBFCs.

The company

Trustline RegTech is a Bangalore-based compliance platform serving 85 banks and NBFCs across India with KYC, AML, and transaction-monitoring infrastructure. Series B, 140 employees, ₹62 Cr ARR.

Their protagonist: Aakash Mehrotra, Head of Strategy — the person every CXO pings on Sunday night asking "did you see what [competitor] just launched?" His recurring nightmare: finding out from a customer first.

§ 02 · The problem

12 analyst hours a day. And signals still slipping.

Scanning load

Two analysts on Aakash's team spent roughly 6 hours each, every weekday, scanning the same 40-odd sources: GeM tender portal, RBI and SEBI notification pages, IRDAI circulars, 14 competitor websites, 22 LinkedIn company pages, plus a rotating list of industry publications. Despite that, signals kept slipping. A direct competitor quietly cut their per-API price by 22% in March; Trustline found out 9 days later when a customer used it as leverage in a renewal.

Breaking pointApril

A ₹3.2 Cr PSU bank tender for a KYC overhaul closed before anyone on the team saw it — because it was filed under a GeM category nobody was watching. The CEO's response was short and unprintable. Clients had begun flagging RBI circulars to customer success before Trustline's own compliance leads had seen them.

₹3.2 Cr tender closed unseen

Signal-to-noise was brutal: of roughly 180 items reviewed per day, maybe 8 actually mattered.

§ 03 · The solution

One configurable agent. 47 sources.

Architecture

Vynara built a configurable market intelligence agent that ingests signals from 47 sources: tender portals (GeM, state eProcurement), regulator feeds (RBI, SEBI, IRDAI, MeitY), 18 competitor websites with diff detection on pricing and product pages, 22 LinkedIn company pages, 3 news APIs, and 6 industry publications.

Each signal runs through a Claude-based classifier that tags it by topic, urgency tier, and owning team, then routes it to the right Slack channel with a 2-line summary and a source link. Daily 8 AM digest for leadership; pricing changes, tender deadlines, and regulator alerts push instantly. Built in 5 weeks. Humans still own what to bid for, what to respond to, and what to ignore.

Push instantly when
competitor pricing diff · tender deadline < 7d · regulator alert
§ 04 · How it works

40 tabs at 9 AM, replaced by 4 Slack channels.

Before · 40 tabs 12 hrs / day · 2 analysts
  1. Analyst opens 40 tabs at 9 AM
  2. Manually scans each source for 6 hours
  3. Copies interesting items into a shared Google Doc
  4. Slacks each item to what they guess is the right team
  5. Other team often misses the ping
  6. CEO finds out from a customer first
After · 47 sources 1.5 hrs / day
  1. Agent ingests 47 sources every 15–60 min
  2. LLM classifies, tags, scores each signal
  3. Urgent items push to #intel-urgent
  4. Routine items flow into team-specific channels
  5. 8 AM digest summarises the last 24 hours for leadership
  6. Analyst spends 90 min reviewing, tuning, escalating
§ 05 · The numbers

Six metrics. 90 days.

Analyst hours / day on scanning−87.5%
Before12.0 hrs
After1.5 hrs
Detection lag · competitor pricing−97%
Before4.2 days
After3.1 hours
Relevant tenders surfaced / month+63%
Before8
After13
Signal-to-noise of items reviewed+56 pts
Before38%
After94%
Missed regulatory updates / quarter−100%
Before7
After0
Pipeline from agent-sourced tenders+₹4.7Cr
Before₹0
After (90 days)₹4.7 Cr
§ 06 · The timeline

Five weeks. 47 sources online.

Phase
W1
W2
W3
W4
W5
Week 1Source audit · access · GeM + regulator prototypes
AUDIT
Week 247 connectors live · Postgres state · dedup logic
CONNECT
Week 3LLM classification · routing rules · urgency tiers
CLASSIFY
Week 4Slack integration · daily digest · dashboard
DELIVER
Week 5False-positive tuning (38% → 6%) · handover · runbook
LIVE
§ 07 · What they said

Browser tabs with legs do analysis now.

"

On a Sunday at 2 AM the bot caught that a competitor had slashed their video-KYC pricing by 30%. By Monday standup we had a counter on the table. Six months ago we'd have heard about it from a churned customer.

AM
Aakash Mehrotra
Head of Strategy · Trustline RegTech
"

My analysts stopped feeling like browser tabs with legs. They actually do analysis now.

VI
Vinod Iyer
COO · Trustline RegTech
§ 08 · The takeaway

The pattern is portable.

Any B2B firm operating across tender portals, regulator feeds, and competitor moves can rebuild this in 4–6 weeks.

The trick isn't the LLM — it's the source list and the routing logic. Vynara ships that part.

One-liner We built a market intel agent that cut a RegTech firm's scanning load by 87% and surfaced ₹4.7 Cr in tender pipeline in 90 days.
Send us your source list →