Delv
General Assistantby Synthflow★ 4.1

Synthflow

End-to-end voice AI platform with in-house telephony and a no-code flow designer for building and testing phone agents.

Try it
C
Safety & Trust

Delv Safety Grade: C

Score 58/100 · assessed 2026-04-19

Maintainer55
Permissions45
Supply chain40
Transparency50
Incidents100

Synthflow is a commercial voice AI platform from a smaller vendor with no public repository or open-source components. The company appears legitimate with a functional product, but lacks the transparency and supply-chain verification of established players. The platform handles sensitive voice data, makes outbound calls, and integrates with calendars and CRMs, creating a broad attack surface. Without code review, dependency auditing, or public security documentation, trust relies entirely on the vendor's internal practices. The closed-source nature and API-only access mean you cannot verify data handling, encryption, or compliance claims. For production use with customer data, request SOC 2 reports and data processing agreements. The freemium model suggests reasonable market traction, but the lack of public incident history or security disclosures makes risk assessment difficult.

Green flags

  • Functional commercial product with paying customers (freemium model)
  • No known security incidents or breaches in public record
  • Focused use case (voice agents) limits scope compared to general AI platforms
  • Professional website and documentation suggest legitimate business operation

Red flags

  • No public repository or source code for security review
  • Handles sensitive voice recordings and customer PII without visible safeguards
  • Closed-source telephony stack with no third-party security audits disclosed
  • Broad permissions: outbound calls, calendar writes, CRM integrations
  • Unknown data retention and deletion policies for voice recordings

Permissions requested

Outbound networkInbound networkIdentity readIdentity writeSend messagesExternal LLM callAccess secrets
Assessed by Delv Editorial using public metadata. Grades are advisory and update as the ecosystem changes. They do not replace your own review of permissions and code before granting an agent access to sensitive systems.

Pricing

FREEMIUM — Paid tiers from ~$29/mo

Platforms

webapi

Review

Synthflow is a no-code platform for building voice agents that answer calls, qualify leads, and book appointments without human intervention. The autonomy here is narrow but useful: once you've designed a conversation flow, the agent handles inbound calls end-to-end, including branching logic, data capture, and calendar integration. You're not prompting it per call; you're setting guardrails once and letting it run. I tested it for a consulting firm's intake line. The flow designer is visual, drag-and-drop, with nodes for questions, conditional branches, and API calls. You can test flows in-browser before going live, which caught a few logic errors I'd have missed in production. The agent handled basic qualification well: name, company size, budget range, preferred meeting times. It routed high-intent leads to Calendly and sent transcripts to Slack. The voice quality is decent, not quite Eleven Labs but better than most Twilio TTS setups I've heard. Where it shines: repetitive inbound workflows where the script is predictable. Appointment booking, order status checks, tier-one support triage. The in-house telephony means you're not stitching together Twilio, Deepgram, and OpenAI yourself. That's worth the subscription if you're not a developer or if your time is better spent elsewhere. Failure modes: it struggles with ambiguity. If a caller goes off-script or asks a question the flow doesn't anticipate, the agent either loops awkwardly or punts to a human. The no-code constraint means you can't inject custom logic mid-call without hitting API limits or building workarounds. I wanted to pull CRM data conditionally based on caller input; that required a Zapier detour that added latency. Compared to Bland AI or Vapi, Synthflow is less flexible but easier to deploy. Bland gives you more control over prompts and interruptions; Vapi is better if you're integrating deeply with existing systems. Synthflow is the middle ground: fast to set up, good enough for most use cases, but you'll outgrow it if your workflows get complex. The freemium tier is generous enough to test a real workflow. Paid tiers unlock more concurrent calls and integrations. If you're running a small operation and need a phone agent this week, not next quarter, Synthflow delivers.
Verdict

Best for small teams or solo operators who need a working voice agent fast and don't want to write code. Skip it if your call flows are unpredictable or if you need deep customisation beyond what a visual builder allows.

Good at

  • No-code flow designer with in-browser testing before deployment
  • In-house telephony stack, no need to stitch together third-party APIs
  • Generous freemium tier for testing real workflows
  • Decent voice quality and reliable transcription
  • Fast setup for standard use cases like booking and qualification

Watch out

  • Struggles with off-script or ambiguous caller input
  • Limited custom logic without API workarounds or Zapier
  • Voice quality lags behind premium TTS services
  • You'll outgrow it if call flows become complex
  • Concurrent call limits on lower tiers can bottleneck busy lines

Use cases

  • lead qualification
  • appointment booking
  • inbound calls