Sherlock Calls
for ClickHouse + Vapi
ClickHouse stores and queries high-volume event and operational data at scale. Vapi runs your AI voice agents on any telephony stack. When you need to investigate across both, the evidence is split between two dashboards neither of which knows the other exists. Sherlock Calls bridges them — no code, no exports, no manual joins. Ask once from Slack and get a sourced answer in under 5 seconds.
TL;DR — What beta users get access to
- 1
Sherlock Calls connects to ClickHouse, Vapi simultaneously — read-only, no code changes, no webhooks — and lets you query both with a single Slack message.
- 2
Ask questions that neither ClickHouse nor Vapi can answer alone. ClickHouse holds the raw source of truth — but querying it for business questions requires SQL access and data engineering resources. Vapi shows agent sessions — not how they map to telephony errors or CRM results. Sherlock deduces the complete picture from both.
- 3
No dashboard switching, no manual joins, no fog of uncertainty — ask in Slack and receive a sourced answer with evidence from every connected provider in under 5 seconds. The game is afoot.
<5s
Answer to any database + voice AI query
2
Connected platforms, 1 Slack question
0
Code changes or webhooks required
The Investigation Gap
What's invisible when you use ClickHouse + Vapi without Sherlock
Each platform shows you its own data. But the questions that matter most live in the gaps between them.
Vapi AI agents run without the ClickHouse application context that would change their answers
ClickHouse holds the business records, product state, and event history that would transform your Vapi AI agent's response from generic to precise. But Vapi agents call without access to ClickHouse — so they give the same answer regardless of what the data says.
Vapi AI conversation outcomes and ClickHouse downstream business results are never correlated
ClickHouse holds what happened after the Vapi AI conversation — did the record update? Did the issue resolve? Did the customer return? Whether Vapi AI conversations actually produce the right ClickHouse outcomes is a question that requires joining both datasets.
ClickHouse data patterns that predict Vapi AI escalation are invisible
Customers with specific ClickHouse application states — certain usage patterns, recent errors, specific record conditions — may escalate Vapi AI conversations at much higher rates. That predictive signal exists in the data but is invisible without a cross-system query.
Cross-Provider Questions
What teams ask Sherlock about ClickHouse + Vapi
Questions that would take hours to answer manually — answered in under 5 seconds from Slack.
- SC“Which Vapi AI conversation outcomes correlate with specific ClickHouse record states at the time of the call?”
- SC“Show me ClickHouse tables most frequently updated in the 24 hours following a Vapi AI conversation”
- SC“Find Vapi AI conversations that ended in escalation for customers with specific ClickHouse application states”
- SC“Which ClickHouse data patterns best predict whether a Vapi AI conversation will succeed or escalate?”
- SC“What's the ClickHouse account health for customers who had multiple Vapi AI conversation failures this month?”
Beta Setup
Connect ClickHouse + Vapi to Sherlock in 2 minutes
No code, no webhooks, no new dashboards. Beta users get direct onboarding support.
- 1
Connect ClickHouse
Add your ClickHouse credentials to Sherlock Calls. Read-only access — no code changes, no webhooks, no ClickHouse configuration required.
- 2
Connect Vapi
Add your Vapi credentials. Sherlock indexes all AI agent sessions, call transcripts, and tool call logs automatically.
- 3
Ask your first cross-provider question. The game is afoot.
Type any question about your combined ClickHouse + Vapi stack in Slack. Sherlock queries all connected platforms in parallel, correlates the evidence, and returns a sourced answer in under 5 seconds.
FAQ
Common questions about Sherlock + ClickHouse + Vapi
How does Sherlock Calls connect ClickHouse and Vapi data?
- Sherlock uses read-only API access to both platforms simultaneously. When you ask a question, it queries ClickHouse, Vapi in parallel, correlates the results by timestamp and shared identifiers, and produces a single sourced answer — the same way a good detective correlates evidence from multiple witnesses.
Do I need to set up any data pipelines between ClickHouse and Vapi?
- No. Sherlock Calls is entirely pull-based — it queries both APIs on demand when you ask a question. There are no webhooks, no ETL pipelines, no data warehouses, and no code changes required in any of the connected platforms.
What kinds of questions can I ask about my ClickHouse + Vapi stack?
- You can investigate anything that spans both platforms — query volume and event ingestion rate, conversation success rate and latency, cross-platform costs, handoff patterns, and performance comparisons. Sherlock translates your plain-English question into the right API calls and returns the deduced answer.
Is my ClickHouse and Vapi data stored by Sherlock?
- No. Sherlock Calls queries your data in real time and returns results directly to Slack — nothing is stored, indexed, or replicated in any Sherlock database. All data remains in ClickHouse and Vapi and is accessed only during an active investigation.
How long does it take to set up the ClickHouse + Vapi integration?
- Elementary — typically under 5 minutes total. Connect each platform with read-only credentials, install the Sherlock Calls Slack app, and ask your first question. No engineering, no dashboards, no onboarding calls required.
Apply for early access to Sherlock + ClickHouse + Vapi
We're accepting a select group of beta users to shape the ClickHouse + Vapi combination. Tell us about your stack and we'll reach out personally if you're a fit.
Explore individual integrations