What it does

This is a webhook server that forwards voice queries from Google Assistant to the SUSI AI API and returns the responses. You create an Actions on Google project, hook it to API.AI (now Dialogflow), deploy this Node code to Heroku, and point the fulfillment URL at it. The README is essentially a 10-step deployment checklist with screenshots.

The interesting bit

The project is a thin adapter layer — the actual intelligence lives entirely in SUSI’s remote API. What’s notable is the vintage: the instructions still reference “API.AI” (rebranded to Dialogflow in 2017) and treat Heroku’s free tier as the default hosting path, which Heroku discontinued in 2022. It’s a time capsule of mid-2010s chatbot plumbing.

Key highlights

• Single webhook endpoint (/webhook) handles all intent fulfillment
• Automatic Heroku deploys via GitHub integration
• SUSI API responses are passed through without transformation
• Includes screenshot-heavy setup guide for Actions on Google console
• Demo video linked for testing on physical Google Assistant devices

Caveats

• Setup instructions reference deprecated services (API.AI branding, Heroku free dynos)
• No code visible in README; actual implementation is in the repo files, not shown
• Testing restricted to the Google account that created the Actions project

Verdict

Worth a look if you’re maintaining a legacy SUSI deployment or studying how pre-LLM voice assistants were stitched together. Skip it if you want modern conversational AI — this is plumbing, not intelligence, and the pipes are showing their age.