Skip to content

Xplorr Slack Bot: AI-Powered Cloud Cost Queries in Slack

The Xplorr Slack bot lets your team query cloud costs, check recommendations, and manage budgets directly from Slack. No need to open a dashboard or write API calls — just ask a question in natural language.

The bot connects to the same Xplorr APIs that power the dashboard and MCP server. It has 16 tools covering cost queries, optimization, budgets, alerts, and account management. When you send a message, the bot figures out which tools to call, runs them, and returns a formatted response.

Examples of what you can ask:

  • “What did we spend on AWS last week?”
  • “Show me the top 5 most expensive services this month”
  • “Are there any cost anomalies?”
  • “What optimization recommendations do we have?”
  • “How much would we save by right-sizing our EC2 instances?”
  • “Create a $10,000 monthly budget for the production account”
  • “What’s the forecast for this month’s total spend?”
  • “Compare this month vs last month”
  • “List all active alerts”
  • “Trigger a sync for our Azure account”

The bot uses an agentic loop:

  1. Your message arrives via the Slack Events API
  2. The bot maps your Slack workspace to your Xplorr organization
  3. It sends your message to GPT-4o with the 16 available tools defined as function calls
  4. GPT-4o selects which tools to call based on your question
  5. The bot executes those tool calls against the Xplorr API
  6. Results go back to GPT-4o to generate a natural language response
  7. If more data is needed, the loop repeats (up to 5 iterations)
  8. The final response is posted to your Slack channel in formatted Slack markdown

For multi-step questions like “What are my top services and how do they compare to last month?”, the bot chains multiple tool calls in a single conversation turn.

CategoryTools
Cost Analysisget_cost_summary, get_cost_trend, get_top_services, get_cost_forecast, get_mom_comparison, compare_periods, get_top_resources
Optimizationget_recommendations, get_savings_summary, get_anomalies, get_ai_recommendations, update_recommendation_status
Budgetslist_budgets, create_budget
Alertslist_alerts, create_alert
Accountslist_cloud_accounts

The bot confirms before executing destructive actions (deleting budgets or alerts). Recommendations can be applied, dismissed, or snoozed directly via interactive buttons — see Approval Workflows.

The Xplorr Slack bot is configured during onboarding. Here’s what happens:

  1. Install the Slack app. An admin in your Slack workspace installs the Xplorr Slack app. This grants the bot permission to read messages where it’s mentioned and post responses.

  2. Connect to your Xplorr org. The Xplorr team links your Slack workspace ID to your Xplorr organization. This ensures the bot queries the right data.

  3. Invite the bot to channels. Add @Xplorr to any channel where you want to use it. The bot only responds when directly mentioned or in DMs.

To get started, reach out at console.xplorr.io or email support. Setup takes about 5 minutes.

Mention it in a channel:

@Xplorr What did we spend on AWS this week?

DM it directly:

What are my top 10 most expensive services this month?

Multi-step requests work:

@Xplorr Show me the cost trend for the last 30 days,
highlight any anomalies, and list our top recommendations.

The bot responds with rich Slack Block Kit messages — structured sections, progress bars, action buttons, and formatted tables. In Microsoft Teams, responses use Adaptive Cards with the same level of detail. Dollar amounts are formatted as USD.

Beyond on-demand queries, the bot proactively posts to your channels — no one needs to ask:

  • Daily cost digest — Every morning, a smart summary of yesterday’s spend, top services, anomalies, recommendations, budget status, and forecast. See Daily Cost Digest.
  • Anomaly alerts — Real-time notifications when Xplorr detects a cost spike, with AI-generated explanations and interactive action buttons.
  • Budget alerts — When a budget threshold is crossed, with progress bars and projected impact.
  • Alert rule triggers — When custom alert rules fire (daily spend, service cost, forecast thresholds).
  • Weekly executive summary — A polished leadership-ready report every Monday with AI-generated insights and budget scorecards. See Weekly Executive Summary.
  • Approval requests — When a team member requests to apply a high-value recommendation. See Approval Workflows.

All proactive notifications use rich Block Kit formatting with interactive buttons. Configure them in Settings → Integrations → Channels or see Channel Routing for team-based routing.

Slack BotTeams BotMCP Server
InterfaceSlack (any device)Microsoft TeamsClaude Desktop / Claude Code / Claude.ai
Best forTeam-wide queries, alerts, approvalsSame as Slack, for Teams-first orgsDeep individual analysis, scripting, automation
Auth modelWorkspace-levelAzure AD tenant-levelToken per user
AI modelGPT-4oGPT-4oClaude (whichever model you use)
Tools181826
Proactive alertsYes (Block Kit)Yes (Adaptive Cards)No (reactive only)
Interactive actionsApply/dismiss/approve buttonsApply/dismiss/approve buttonsNo
SetupWorkspace admin installs appAzure AD admin consentsUser adds MCP config

Use all three: Slack/Teams for team visibility, alerts, and approvals; MCP for deep analysis and automation.

  • Not mentioning the bot. In channels, the bot only responds to @Xplorr mentions. It doesn’t read all messages.
  • Asking about accounts that aren’t connected. If you ask about Azure costs but only have AWS connected, the bot returns empty results. Check connected accounts with “List my cloud accounts.”
  • Expecting real-time data. Cost data syncs every 6 hours. The bot returns the most recent synced data, which may be up to 6 hours old.

Can multiple people use the bot at the same time? Yes. The bot handles concurrent requests. Each message is processed independently.

Does the bot store conversation history? No. Each message is a standalone request. The bot doesn’t maintain conversation context between messages.

Can I restrict who can use the bot? Currently, anyone in the Slack workspace who can see the bot can use it. Role-based access control for the bot is on the roadmap.

What if the bot is slow to respond? Complex queries that require multiple tool calls can take 5-10 seconds. If it consistently takes longer, check that your cloud accounts are syncing successfully.