Zero-code launch path

Launch the agent workflow first. Add custom engineering when it earns its place.

Put a voice or multimodal agent on the web or phone, define how it behaves, and keep a direct path to human handoff, artifacts, widgets, SDKs, and APIs.

Build an agent widget Run a live demo
  • No custom backend required for the first launch
  • Web, telephony, and human escalation paths
  • Extend with SDKs and APIs without starting over
About 36 minutesGuided first-test workflow
Widget or dashboardChoose the fastest useful entry point
APIs remain openCustomize only where the product needs it
What goes live

More than a prompt and a microphone.

A useful agent needs an entry point, operating rules, escalation, and follow-up artifacts. MediaSFU brings those pieces into the same workflow.

Agent behavior

Configure prompts, providers, fallback behavior, and availability without building an orchestration backend.

Phone and web entry

Meet customers in a browser, through a call button, or on a managed phone workflow.

Human handoff

Escalate sensitive or high-value conversations into a human-operated workflow.

Useful artifacts

Retain transcripts, notes, summaries, and call outcomes for the team that follows up.

First-test plan

Move from idea to a testable workflow in one focused session.

The times are planning guides for the MediaSFU setup steps. Provider onboarding, phone-number provisioning, and organization approvals can add time.

See the workflow live
  1. 01
    Choose the entry point

    Start with a web agent, call button, managed number, or operator-led workflow.

    3 min
  2. 02
    Shape agent behavior

    Set the greeting, provider credentials, fallback behavior, escalation rules, and availability.

    8 min
  3. 03
    Connect the conversation path

    Use web voice or connect telephony, then confirm the AI providers and routing you want.

    10 min
  4. 04
    Publish the experience

    Launch from the dashboard or place the generated widget in your site or product shell.

    5 min
  5. 05
    Test real outcomes

    Run support, sales, and handoff scenarios; then review transcripts and call artifacts.

    10 min
Build depth

Use the lightest surface that proves the outcome.

Each level is a valid destination. Move deeper only when your UI, policy, or integration requirements demand it.

01
Launch

Dashboard + widget

Best for proving the workflow and getting the first experience in front of users.

Open Widget Studio
02
Operate

Guided agent tools

Add telephony, campaign behavior, provider choices, and team operating controls.

Explore AI agents
03
Extend

SDK + API control

Own the product UI, tenant policies, events, integrations, and automation depth.

Review developer paths
Strong starting points

Begin with one conversation your team already understands.

A narrow, measurable first workflow makes behavior, handoff, and conversion quality easier to evaluate.

01

AI receptionist

Answer common questions, route calls, collect intent, and hand off when the conversation needs a person.

02

Product onboarding

Place a voice or multimodal guide inside the product so users can ask for help in context.

03

Lead qualification

Capture requirements, answer product questions, and send structured outcomes to the sales workflow.

Decision questions

Know what you are choosing before you publish.

Use these answers to decide whether the widget path is enough or whether your first version already needs developer control.

Can a non-technical team launch the first agent?

Yes. The dashboard and widget paths cover the first setup, launch, and operating workflow. Engineering becomes useful when you want a deeply custom product surface or automation model.

Does zero-code mean we are locked out of APIs later?

No. The same product can progress from a managed widget to SDK components and headless API control without treating the first launch as throwaway work.

Can the agent work on phone calls and the web?

Yes. MediaSFU supports browser-based agent experiences and telephony workflows, with provider and routing choices configured for the use case.

What should we validate before production?

Test fallback behavior, human escalation, latency, transcript quality, provider limits, consent requirements, and the exact outcomes your team needs after each conversation.

Where do we estimate usage cost?

Use the developer pricing page and agent workload model to estimate the MediaSFU portion, then include telephony and selected AI provider charges in the full operating budget.

Ready to test the workflow?

Launch the first useful agent, then learn from real conversations.