Use cases and examples

This page shows common patterns for deploying Quant AI Slack bots. Each example builds on the concepts from Agents and routing.

Internal knowledge base assistant

Goal: A bot that answers questions from internal documentation stored in a vector database.

Setup:

Create a vector database collection and ingest your internal docs (policies, runbooks, FAQs)

Create a Slack bot with a system prompt like:

You are an internal knowledge assistant for [Company].
Answer questions using the documentation in your knowledge base.
Always cite the source document when providing answers.
If you cannot find a relevant answer, say so clearly.

Assign the vector database collection in the Knowledge Base section
No sub-agents needed — the bot handles everything directly

How it works in Slack:

User: @Bot What's our policy on remote work?

Bot: According to the Employee Handbook (section 4.2), remote work
is available for all full-time employees with manager approval...

Multi-department support hub

Goal: A single Slack bot that routes questions to the right department.

Setup:

Create specialist agents:
- IT Support Agent — system prompt for technical troubleshooting, tools for querying ticket systems
- HR Agent — system prompt for HR policies, vector database with policy documents
- Facilities Agent — system prompt for office/building requests
Create a Slack bot and select all three agents as sub-agents
Enable keyword monitoring with keywords like password, vpn, leave, desk

How it works in Slack:

User: @Bot I can't connect to the VPN from home

Bot: ⏳ Routing to IT Support...
Bot: 🤖 IT Support → checking known issues...

Bot: Here are the steps to resolve VPN connectivity issues:
1. Ensure you're using the latest VPN client (v4.2+)...

The router agent evaluates the question and delegates to IT Support. If someone asks about leave policies, it routes to the HR agent instead.

Customer support triage

Goal: A Slack bot that helps your support team by looking up customer data and suggesting responses.

Setup:

Create agents:
- Customer Lookup Agent — custom tools that query your CRM and ticket system
- Knowledge Base Agent — vector database with product documentation and known issues

Create a Slack bot with a system prompt like:

You help support agents resolve customer issues.
When given a customer name or ticket number, look up their details.
Suggest responses based on our knowledge base and past resolutions.

Select both agents as sub-agents

How it works in Slack:

Support agent: @Bot Customer Jane Smith is reporting slow page loads on their site

Bot: ⏳ Looking up customer...
Bot: 🤖 customer-lookup → querying CRM...
Bot: 🤖 knowledge-base → searching known issues...

Bot: **Customer: Jane Smith** (Acme Corp)
- Plan: Enterprise
- Sites: 3 active, primary site seeing 2.3s avg TTFB (up from 0.8s)

Likely cause: Their recent deployment at 14:32 included unoptimised
images. Suggested response: [draft response with steps]

Compliance review bot

Goal: A bot that reviews content against regulatory requirements before publication.

Setup:

Create a Slack bot with:
- System prompt defining compliance review criteria for your industry
- A skill for compliance checking methodology
- Vector database collections containing regulatory documents and guidelines
No sub-agents needed for a focused compliance bot

How it works in Slack:

User: @Bot Please review this press release for compliance:
[pastes content]

Bot: ⏳ Reviewing against compliance guidelines...

Bot: **Compliance Review Results:**
✅ No financial forward-looking statements detected
⚠️ Line 3: "guaranteed results" — avoid absolute guarantees per
   marketing guidelines section 2.4
✅ Privacy disclosures present
❌ Missing required disclaimer for regulated product mentions

Automated reporting

Goal: A bot that generates reports on demand using data from external systems.

Setup:

Create custom tools (edge functions) that query your analytics platforms, databases, or APIs

Create a Slack bot with tools for data retrieval and a system prompt like:

You are a reporting assistant. When asked for a report, use your
tools to fetch current data and present it in a clear, formatted
summary. Include relevant trends and highlights.

How it works in Slack:

User: /ask Give me this week's site performance summary

Bot: ⏳ Fetching performance data...
Bot: ✓ CDN metrics loaded · fetching origin stats...

Bot: **Weekly Performance Summary (Mar 10–17)**

| Metric | This week | Last week | Change |
|--------|-----------|-----------|--------|
| Requests | 2.4M | 2.1M | +14% |
| Cache hit rate | 94.2% | 93.8% | +0.4% |
| Avg TTFB | 42ms | 45ms | -7% |
| Error rate | 0.02% | 0.03% | -33% |

Highlights: Traffic spike on Wednesday (product launch).
Cache hit rate improved after page rule optimisation.

Development and CI/CD assistant

Goal: A bot in your engineering Slack channel that helps with deployments, code review, and incident response.

Setup:

Create specialist agents:
- Deploy Agent — tools for checking deployment status, triggering builds
- Code Review Agent — skills for code review methodology, web search for documentation
- Incident Agent — tools for querying monitoring systems, vector database with runbooks
Create a Slack bot with all three as sub-agents
Restrict to your engineering channels

How it works in Slack:

Engineer: @Bot What's the status of the latest deploy to production?

Bot: ⏳ Routing to deploy-agent...
Bot: 🤖 deploy-agent → checking deployment status...

Bot: **Latest Production Deployment**
- Commit: abc1234 "Add user preferences API"
- Status: ✅ Deployed successfully at 14:23 UTC
- Deploy time: 2m 34s
- Health checks: All passing

Tips for effective bots

Start simple — Begin with a single agent and add sub-agents as you identify distinct use cases
Write clear agent descriptions — The router uses these descriptions to decide where to route questions. Specific descriptions lead to better routing
Use keyword monitoring sparingly — Only enable it when the bot genuinely needs to respond to ambient channel messages. @mentions and DMs cover most use cases
Set appropriate session TTLs — Short TTLs (1—7 days) for transactional bots; longer TTLs (14—30 days) for ongoing projects
Test in a private channel first — Create a test channel to verify routing and tool execution before deploying to team channels

Next steps

Get started — Create your first bot
Agents and routing — Understand the routing architecture
Configuration reference — All bot settings