How AI Agents Coordinate with Intercom

Intercom positions itself as the "AI-first" support platform with Fin as its in-house AI agent. But many teams running AI voice or third-party AI chat agents still rely on Intercom for ticket management and customer messaging. Coordinating SIMBA-style external AI with Intercom requires understanding where each tool fits and how they share data.

TL;DR

• Intercom owns the customer conversation history; SIMBA owns the call audio.
• Sync structured call summaries into Intercom as conversation records or notes.
• Use Intercom's customer attributes to inform the AI agent's context.
• Don't duplicate functionality — pick which tool handles which surface.

How they fit together

A typical setup:

• Phone calls handled by SIMBA voice agent.
• In-app chat handled by Intercom (with or without Fin).
• Email handled by Intercom.
• Customer history lives in Intercom as the unified record.
• Call transcripts and audio live in SIMBA, linked from Intercom.

Each tool does what it's best at. The integration ties them together.

What goes from SIMBA to Intercom

For each call:

Conversation record. Create or update an Intercom conversation tied to the customer.

Internal note. A 2-3 sentence summary of what happened.

Tags. Intent, resolution status, AI vs human, sentiment.

Attribute updates. Update customer attributes if the call surfaced new info ("preferred_callback_time", "last_issue_topic").

Link to recording. A URL pointing to the SIMBA call detail (so the human can listen if needed).

Customer attribute sync

Intercom's customer attributes can inform the AI:

SIMBA reads: before each call, fetch Intercom customer attributes (lifecycle stage, plan, last interaction). Inject relevant ones into the prompt.

SIMBA writes: after the call, update attributes (last_call_date, last_call_intent, satisfaction_indicator).

This makes the AI context-aware about the customer's history without storing it all locally.

Setting up the integration

Three pieces:

1. Intercom API access. Generate an access token with permissions for Conversations, Contacts, and Tags.

2. Webhook from SIMBA. Configure SIMBA to POST call summaries to an endpoint you own.

3. Bridge service. A small server that receives SIMBA webhooks, formats the data for Intercom, and posts to Intercom's API.

Most AI platforms ship with prebuilt Intercom integrations. If not, the bridge service is a few hundred lines.

Routing in Intercom

Intercom Workflows (or Operator Rules) can route AI-handled conversations:

AI resolved, no follow-up needed: Auto-close.

AI resolved, customer wanted callback: Assign to next available agent with priority.

AI escalated, urgent: Page on-call.

AI escalated, complex: Route to specialist queue.

AI tagged "negative sentiment": Flag for QA review.

These rules turn structured AI data into actionable workflow.

Fin and SIMBA together

If you're running Fin for chat and SIMBA for voice:

• Each handles its own channel.
• Both should write to the same customer record.
• Both should share the same knowledge base if possible.
• Cross-channel context: customer chats with Fin, then calls SIMBA — SIMBA should know Fin had the prior interaction.

This requires either:

• A single source of truth (Intercom holds all conversations).
• A shared event bus where both AIs write.

For most teams, Intercom-as-source-of-truth is simpler.

Conversation continuity

A subtle case: customer chats with Fin at 2pm; calls SIMBA at 4pm about the same issue. SIMBA should pick up where Fin left off.

Pattern:

1. SIMBA looks up the caller in Intercom.
2. Pulls recent conversations (last 7 days).
3. Summarizes and injects into the system prompt.
4. The agent acknowledges the context: "I see you chatted with us earlier about your billing — let's pick up there."

This is rare but high-impact when it happens.

Common Intercom + AI mistakes

Duplicate conversations. Same call creates two Intercom records due to webhook retry. Use idempotency keys.

Tag explosion. Every AI call adds 5 new tags. After a year, tag list is 500+. Use a controlled vocabulary.

Chat vs voice routing confusion. AI agent confused about which channel it's on. Be explicit in the prompt.

Customer attribute overwrites. AI overwrites attribute set by a human. Use additive updates or designated AI-only attributes.

Reporting

Useful Intercom reports for AI-handled cohorts:

• AI-resolved conversations per week
• Median time to AI resolution
• Escalation rate from AI
• Customer satisfaction by AI vs human
• Repeat contact rate after AI resolution

Set up these as saved views.

Migration considerations

If you're moving from Intercom-only to Intercom + external AI:

• Don't try to replace Fin and add SIMBA in the same project.
• Pilot SIMBA on voice (where Fin doesn't compete).
• Add chat handling later if Fin doesn't meet your needs.
• Keep customer attributes consistent across both AIs.

For the broader integration pattern, see how AI agents coordinate with helpdesks like Zendesk — same idea, different tool.

Related reading

• The Definitive Guide to AI Customer Support in 2026
• Building a Tier-1 AI Support Agent Step by Step
• Why "Human-in-the-Loop" Beats "Fully Autonomous" for Most Teams
• How to Calculate ROI for AI Customer Support
• How AI Support Agents Should Handle Account Verification

FAQ

Should I use Fin or external AI for chat? Depends on customization needs. Fin is convenient if you're all-Intercom. External AI gives more control.

Will SIMBA's voice transcripts show up in Intercom? Summaries, yes. Full audio lives in SIMBA with a link from Intercom.

Can SIMBA call Intercom's API directly during a call? Yes — useful for live customer attribute lookups.

What about Intercom's mobile SDK? SIMBA voice agents can be embedded in mobile apps via WebRTC, separate from Intercom's chat SDK.

Can I migrate from Intercom to a different helpdesk? Yes — most modern helpdesks have similar API surfaces. The integration logic is portable.