Synthflow AI

@synthflow list all AI assistants in my workspace and show me their current language and voice settings

Based on the workspace configuration, here are your active assistants: 'Customer Support Bot' (English US, voice: 'Sarah - Professional'), 'Sales Qualifier' (English UK, voice: 'James - Friendly'), and 'Appointment Scheduler' (Spanish, voice: 'Maria - Warm'). Each assistant is currently active and has been configured with distinct prompt templates. The Customer Support Bot handles an average of 45 calls per day, while the Sales Qualifier processes about 20 daily interactions.

This example demonstrates the MCP's ability to retrieve and synthesize assistant metadata across a workspace. Requires API key with read permissions on assistants. The actual response structure depends on how many assistants exist in your account — this illustrative output assumes a typical small-team setup with 3 configured bots.

@synthflow create a new assistant named 'Product Demo Bot' with English language, professional female voice, and a prompt that focuses on explaining our SaaS features

Successfully created assistant 'Product Demo Bot' with ID 'ast_9k2m4n6p'. Configuration: Language set to English (US), voice assigned as 'Emily - Professional Female', and custom prompt initialized to guide feature explanations. The assistant is now active and can be assigned to phone numbers or integrated into your call routing. You'll need to separately configure any knowledge bases or specific product documentation you want this assistant to reference during calls.

This write operation showcases the MCP's assistant creation capability. The prompt, voice, and language parameters are required; knowledge base linking is a separate step. Be aware that creating assistants counts against your Synthflow plan limits, and the assistant will immediately become billable once it starts handling calls.

@synthflow review my current team routing setup and suggest which assistants should handle after-hours calls versus business hours

Your workspace has two teams configured: 'Business Hours Team' (routes to Customer Support Bot and Sales Qualifier) and 'General Routing' (routes to all three assistants). For after-hours optimization, I recommend creating a dedicated 'After Hours Team' that routes exclusively to the Appointment Scheduler, since it's designed for async booking rather than real-time support. The Customer Support Bot could serve as a fallback, but you'd want to update its prompt to set expectations about response times for complex inquiries submitted outside 9-5.

This synthesis example combines the MCP's team-fetching capability with AI reasoning about call routing strategy. The MCP provides the raw team and assistant data; the AI interprets it in context of the user's operational question. Implementing the suggested changes would require follow-up @synthflow commands to create the new team and modify routing rules.

When Synthflow wins for distributed customer support teams

A 12-person support team covering three time zones needs to route inbound calls to the right assistant based on language and business hours. Synthflow's team creation and phone number tools let you build routing logic that assigns calls to region-specific voice assistants without writing telephony code. Each assistant can have its own knowledge base and prompt tuning. This works if your call volume is under 500 per day and your routing rules are stable—beyond that threshold, you'll spend more time managing assistant configs than you save on dev hours. If you're already running a contact center platform, Synthflow is overkill. But if you're a SaaS company adding phone support for the first time and your team uses Switchy for shared context, this MCP lets you script the entire assistant lifecycle from one workspace.

When this MCP handles short-lived voice campaigns

A 6-person marketing team launches a 3-week product launch campaign with a dedicated phone line for demo requests. They need to spin up a custom voice assistant, load it with campaign FAQs, then tear it down when the promo ends. Synthflow's create and delete assistant tools make this a 10-minute scripted task instead of a vendor support ticket. The team can version the assistant prompt in Switchy, test it with sample calls, and hand off the phone number to the sales team without touching the Synthflow dashboard. This scenario breaks down if you need real-time call transcription or CRM integration—those require custom webhooks Synthflow doesn't expose through the MCP. But for ephemeral voice endpoints tied to a specific campaign calendar, the MCP gives you full lifecycle control without leaving your shared workspace.

When Synthflow's knowledge base tools fit documentation workflows

A 4-person product team updates their voice assistant's knowledge base every sprint when new features ship. The knowledge base tool lets them retrieve the current KB, diff it against the new docs, and push updates without manual copy-paste into the Synthflow UI. This works if your docs are under 50 pages and you're updating once or twice a month—more frequent changes mean you're better off with a docs-as-code pipeline that bypasses Synthflow entirely. The MCP doesn't support bulk KB uploads or version rollback, so if you need audit trails or A/B testing on assistant responses, you'll hit limits fast. But for small teams syncing product docs to a single voice assistant on a predictable cadence, the get and update flow saves 20 minutes per release and keeps the source of truth in your Switchy workspace.