SuperAPI AI Directory

Synthflow docs: introduction

Synthflow is a platform for building voice and chat agents. Its documentation introduces a framework called BELL, comprising four steps: Build, Evaluate, Launch, and Learn, used to create, test, deploy, and improve voice and chat agents. Users can describe their needs to let Aurora generate a first version, or create a voice or chat agent from a prompt, enabling it to answer questions based on documents and websites, transfer calls, book meetings, call APIs, and remember contacts across conversations. Agents can be tested in the editor and simulated conversations scored, then launched by purchasing or importing numbers, bulk calling lists, event triggers, and other methods; they can also integrate with CRMs, calendars, and phone systems, and send call data to your own systems. In addition, the documentation covers monitoring call volume, outcomes, and quality, viewing transcripts and recordings, downloading call records, as well as developer APIs, account management, subaccounts, billing, and FAQs.

Open website

About this tool

The Synthflow documentation explains how to build voice and chat agents from scratch. Its core workflow is called the BELL framework, namely Build, Evaluate, Launch, and Learn, and every agent follows this cycle.

In the Build phase, users can create a voice or chat agent from a prompt, or first describe their needs to Aurora, which generates a first version. Agents can answer questions based on documents and websites provided by the user, perform actions such as transferring calls, booking meetings, and calling APIs, and can remember contacts across multiple turns of conversation. The documentation also mentions that Synthflow can be managed from Claude or Codex.

In the Evaluate phase, users can test agents before real customer contact, including talking or chatting directly in the editor, running simulated conversations and scoring them, and scoring calls according to custom criteria.

In the Launch phase, users can purchase or import phone numbers for agents and bind them, bulk call contact lists, run agents and actions when events occur, deploy chat agents on websites or in apps, and connect CRMs, calendars, and phone systems to send call data to their own systems.

In the Learn phase, users can track changes over time in call volume, outcomes, and quality, view transcripts and recordings for each call, and download call records to generate their own reports.

The documentation also lists popular integrations, such as answering calls in WhatsApp Business, syncing calls and customer context, calling contacts and recording outcomes, and billing subaccounts. For developers and account management, the documentation covers authentication and first requests, call limits and capacity, setting up subaccounts under your own brand, managing plans and invoices, as well as viewing updates, attending onboarding training and videos, watching demos and release notes, finding FAQs, and viewing real-time service availability.

Regarding specific pricing, platform compatibility, supported languages, or model details, the documentation does not clearly state them in the provided content.