PRODUCTION · SLACK BOT + ANALYST

"Where can we place this?" — asked in Slack, answered from the book.

Scout is a Slack bot that answers carrier-placement questions in plain English. An agent types @Scout homeowners Houston TX and gets a ranked list of the carriers we already write in that line and area — strategic partners first, sourced straight from our book of business in the data lake. Ask it something more open-ended and a separate tier takes over: Scout writes and runs its own SQL against the book of business or producer-licensing data, asking before it guesses on anything genuinely ambiguous. Usage — who's asking, how often, what's coming back empty — is tracked on an Entra-gated dashboard.

Node · Slack Bolt — AWS Athena · Ennabl · Atlas — Staff-only · YP Slack workspace — Usage dashboard · Entra SSO
CAPABILITIES
01 · Natural-language queries
💬
Internal · Slack
Ask like a human, not like a form

Type it however you'd say it. Scout extracts the line of business, city, and state from casual phrasing — @Scout workers comp near Billings Montana works as well as @Scout homeowners Houston TX. Works as a mention in any channel or as a direct DM.

Parser LLM extracts LOB · city · state · confidence
Surfaces @mentions in channels · direct messages
Lines Personal auto · home · workers comp · BOP · umbrella · more
Fallback Tells you what it heard, asks for a nudge if unclear
Internal · Slack workspace
02 · Ranked from our book
🎯
Internal · Ennabl
Real placements, not a generic carrier list

Every recommendation is grounded in our own book of business — queried live from the Ennabl data lake, filtered by the requested line and a 50-mile radius. Ranks strategic partners first (up to four), plus one non-strategic option so the agent sees the whole picture.

Source Ennabl data lake · queried via AWS Athena
Filter Line of business · 50-mile radius from zip
Ranking Strategic partners first (≤4) · +1 non-strategic
Fallback Local carrier-tier map if warehouse is down
Internal · Slack workspace
03 · Three tiers, one bot
🧠
Internal · Role-aware
Agents, leadership, and a full data analyst — same bot, different depth

Everyone gets the same placement lookup, but leadership sees deeper detail on identical queries. A separate Tier II tier goes further still: open-ended questions — written premium by LOB, retention trends, a carrier's appointment status — get routed to an agent loop that writes and runs its own SQL against the book of business or producer-licensing data, iterating until it has a real answer and charting results where the shape allows it. It asks before it guesses when a question is genuinely ambiguous. Tier II and Leadership are independent — someone can have one, the other, or both.

Agent view Ranked list with strategic-partner badges
Leadership + policy counts · written premium · commission context
Tier II NL→SQL agent loop over Ennabl/Athena + Atlas, SQL validated before it runs
Gating Leadership + Tier II allowlists — independent axes, every query logged
Internal · Slack workspace
Platform
Node Slack Bolt Claude API
Data
AWS Athena Ennabl Atlas (Postgres) zipcodes
Access
Slack signing secret Leadership whitelist Tier II allowlist
Infrastructure
Railway GitHub
Dashboard
Entra SSO Usage analytics

A staff-only bot in the YP Slack workspace. Every placement question stays inside the tools we already work in — no new chat window to learn. A separate, Entra-gated usage dashboard exists for tracking adoption and query history, open to a small allowlist.