CHATBOTS

Intercom voice-reply bot for accessibility-mode visitors

When a visitor in accessibility mode sends an Intercom message, this generates a clear text answer with OpenAI and returns it as a spoken ElevenLabs audio clip in the same…

CategoryChatbots
Enginesim
Difficultyintermediate
Triggerwebhook
Steps5
Setup~15 min

How it runs

The automated pipeline, trigger to output.

  • TriggerIntercom inbound message webhookIntercomIntercom
  • LogicBranch: only if accessibility_mode is on
  • ActionDraft plain-language answer with OpenAIOpenAI
  • ActionSynthesize spoken reply with ElevenLabsElevenLabsElevenLabs
  • OutputPost audio + transcript to Intercom conversationIntercomIntercom

What it does

Visitors who have turned on accessibility mode get spoken replies instead of text-only ones. Each inbound Intercom message is answered by an LLM, converted to natural speech with ElevenLabs, and the audio is posted straight back into the conversation so screen-reader and low-vision users can listen rather than read.

When to use it

Run this when your support inbox serves visitors with visual impairments or reading difficulties and you want voice answers without staffing a phone line. It assumes you tag accessibility-mode sessions with an Intercom custom attribute or contact tag.

How it works

  1. 1Intercom fires a conversation webhook on a new inbound user message.
  2. 2A logic step checks the contact's `accessibility_mode` attribute; non-flagged conversations are skipped so normal support flow is untouched.
  3. 3OpenAI drafts a concise, plain-language answer from the message text and conversation history.
  4. 4ElevenLabs synthesizes that answer into an MP3 using your chosen warm, slow-paced voice.
  5. 5The output step uploads the clip and posts it as an Intercom reply with a short text transcript fallback.

Set it up

What you configure once, before turning it on.

  1. 1
    Connect IntercomConversations, contacts, articles.
  2. 2
    Connect OpenAIModels, embeddings, files.
  3. 3
    Connect ElevenLabsText-to-speech, voice cloning.
  4. 4
    Set each agent's modelWe leave models unset so you pick the tier — fast + cheap, or top-quality.
  5. 5
    Tune it to your dataEdit the prompts, filters, and field mappings so it matches how your team works.
  6. 6
    Test, then turn it onRun once against a sample, confirm the output, then enable the trigger.

Run this workflow in your colony.

14-day trial. No DevOps. No Sales call. Provisioned in under a minute.