Sherlock Calls
for Rasa
Rasa powers open-source and enterprise AI agents with the CALM framework. Sherlock gives your team the investigation layer — ask any question about conversation outcomes, NLU performance, and voice AI results without engineering support.
TL;DR — What this integration will do
- 1
Sherlock Calls connects to your Rasa deployment via JWT or token authentication — no code changes, no webhooks — giving your team plain-English access to all conversation and NLU analytics from Slack in under 2 minutes.
- 2
Ask anything about your Rasa agents from Slack: 'Which intents have the lowest NLU confidence this week?' 'How is the CALM framework performing on voice calls?' — answered in under 5 seconds.
- 3
Sherlock correlates Rasa conversation and NLU data with your telephony and CRM systems to give operations teams actionable intelligence without digging through raw logs.
50M+
Rasa framework downloads
$83.8M
Total funding raised
2025
CALM multimodal voice AI launch
About Rasa
What is Rasa?
Rasa is the world's most widely downloaded open-source conversational AI platform — with 50M+ developer downloads and $83.8M raised ($30M Series C, February 2024, led by StepStone Group and PayPal Ventures). Trusted by Adobe, American Express, Deutsche Telekom, Dell, and T-Mobile, Rasa's CALM (Conversational AI with Language Models) framework, launched in 2025, adds multimodal voice AI while preserving enterprise-grade deterministic, compliance-ready conversational logic.
About Sherlock Calls
What is Sherlock Calls?
Sherlock Calls is an AI call detective that lives in your Slack workspace. It connects to voice providers, CRMs, and ad platforms, then answers any question about your calls in plain English — in under 5 seconds, with no code, no new dashboards, and no SDK required. Think of it as a senior analyst always on call, inside the tools your team already uses.
Planned Setup
Connect Rasa to Sherlock in 2 minutes — when it launches
No code, no webhooks, no new dashboards. Here's how setup will work.
- 1
Connect Rasa via JWT or token auth
Configure JWT or token authentication for Sherlock Calls in your Rasa deployment — read-only access to all conversation, NLU, and analytics data. No code changes required.
- 2
Add Sherlock to Slack
Install the Sherlock Calls Slack app to your workspace in one click. Your operations and AI teams can start asking questions from any Slack channel immediately.
- 3
Investigate your first NLU performance issue
Ask 'Which Rasa intents have the lowest confidence scores this week?' — Sherlock queries your NLU analytics and returns a sourced, ranked answer in under 5 seconds.
Future Use Cases
What teams will ask Sherlock about Rasa
- SC“Which Rasa intents have the lowest NLU confidence this week?”
- SC“How is the CALM framework performing on voice calls vs. chat this month?”
- SC“Find conversations where the agent entered an unexpected fallback state”
- SC“What are the top 5 entity extraction failures across our Rasa agents today?”
- SC“How many voice AI calls were contained without a human escalation this week?”
Planned Features
What you'll be able to do with Rasa
NLU Accuracy Monitoring
Track Rasa NLU intent recognition accuracy, entity extraction, and confidence score distributions. Surface training data gaps and low-confidence intents from Slack.
CALM Framework Analytics
Monitor Rasa CALM framework performance — track how LLM-powered conversation management handles complex, multi-turn interactions and voice AI calls.
Voice AI Investigation
Query Rasa multimodal voice AI call outcomes — containment, escalations, and speech recognition accuracy — since the CALM voice AI launch in 2025.
Escalation Root Cause Analysis
Identify which Rasa intents, stories, and conversation flows trigger the most escalations. Correlate with conversation transcripts for actionable training insights.
Conversation Flow Debugging
Investigate conversation flows that fail, loop, or produce unexpected outcomes. Debug Rasa agents faster without engineering support by querying from Slack.
Enterprise and Open-Source Support
Sherlock supports both Rasa Open Source and Rasa Enterprise deployments — on-premises or cloud — via standard JWT/token authentication.
AI Tools
5 specialized tools, zero configuration
Sherlock will pick the right tool automatically. You just ask in plain English — the AI will know which APIs to call and how to correlate the results.
FAQ
Common questions about Sherlock + Rasa
How does Sherlock Calls connect to Rasa?
- Sherlock Calls connects to your Rasa deployment via JWT or token authentication with read-only access. Supports both Rasa Open Source and Rasa Enterprise — on-premises or cloud — with no code changes required.
Can Sherlock Calls analyze Rasa NLU performance?
- Yes. Sherlock queries Rasa NLU data to surface intent recognition accuracy, confidence distributions, and entity extraction failures — all from a plain English Slack question.
Does Sherlock support Rasa's CALM framework?
- Yes. Sherlock can monitor CALM framework conversation management performance — including multi-turn interaction success rates and LLM-driven dialog outcomes.
Can Sherlock investigate Rasa voice AI calls?
- Yes. Sherlock supports Rasa's multimodal voice AI launched in 2025, allowing you to query voice call outcomes, containment rates, and escalations from Slack.
How is Sherlock different from Rasa's built-in analytics?
- Rasa provides raw analytics APIs and platform dashboards primarily designed for engineers. Sherlock adds a natural language investigation layer so your entire team — operations, CX, and product — can query conversation data from Slack without engineering support.
Be first when Rasa launches
Leave your details and we'll notify you the moment the Rasa integration goes live — plus early access and priority onboarding.
Also works with