what is bolnaxsyntarus fr?

This project connects a Bolna voice AI agent to Syntarus, a memory service, so a phone support agent can remember details between calls instead of treating every call as the first one. It does not build or replace the voice, phone, or agent parts of Bolna. It only adds a memory layer on top. Without this bridge, a caller has to repeat themselves every time, such as their preferred callback time or an earlier support issue. With the bridge running, a caller can say something like "remember that my preferred callback time is after 6 PM" during a call, and on a future call the agent can look that preference up and respond to it. The bridge works through two separate paths. During a live call, Bolna calls a recall endpoint on the bridge to fetch anything remembered about that caller. After the call ends, Bolna sends the finished call to a webhook on the bridge, which pulls out a privacy safe summary and stores it in Syntarus for next time. Setup involves creating a Syntarus project key, installing the Python requirements, filling in an environment file with API keys and two secret values you generate yourself, and running the bridge locally. Because Bolna needs a public web address to reach the bridge, a real deployment or a temporary tunnel is required for live testing. Configuring the Bolna side means pasting a provided tool template into the Bolna dashboard and adding specific webhook settings. Privacy is treated carefully. Caller identities are hashed before being sent to Syntarus, raw transcripts are not stored unless changed in settings, and the recall endpoint only reads data, it cannot write. The project notes that it is meant as an integration example rather than a guarantee that it meets any compliance standard, so a team would still need to check consent and data retention rules before using it with real customers. The project is released under the Apache 2.0 license.