> For a complete page index, fetch https://docs.synthflow.ai/llms.txt. For full documentation content, fetch https://docs.synthflow.ai/llms-full.txt.

# Hamming

> Connect your Synthflow workspace to Hamming to test, monitor, and red-team voice agents by syncing agents, phone numbers, and calls with a Synthflow API key.

![Synthflow AI integration with Hamming](https://storage.googleapis.com/granular-changelog/doc-images/hamming_cover.svg)

[Hamming](https://hamming.ai) is an enterprise platform for testing, monitoring, and securing voice AI agents. Connect your Synthflow workspace with a [Synthflow API key](/authentication) and Hamming imports your voice agents, phone numbers, and call history, then keeps them synced so it can generate test scenarios, score outcomes, watch production quality, and run adversarial security tests.

This is a good fit when you run Synthflow agents in production and need repeatable quality assurance, continuous monitoring, and security coverage that spans the full lifecycle, from pre-launch testing to live traffic.

## What you can do

Hamming reads your synced agents and their prompts to drive three workflows against the same agents:

* **QA and testing**: Hamming auto-generates conversation scenarios from each agent's prompt, then runs simulated inbound and outbound calls at scale to exercise edge cases, interruptions, accents, and background noise. Outbound runs use the phone numbers it imports from Synthflow. Every call is graded on 50+ metrics spanning accuracy, conversation quality, voice performance, and task completion, so regressions surface before they reach callers. Your first test report is ready in minutes.
* **Production monitoring**: Hamming replays golden call sets as scheduled health checks to detect drift or outages, breaks down component latency across speech-to-text, LLM, and text-to-speech, and sends proactive alerts by email and Slack. Turning on **Import Calls** feeds your historical Synthflow [calls](/about-call-logs) and metrics into the monitoring dashboard.
* **Red-teaming**: Hamming runs security packs that probe for jailbreak attempts, prompt injection, and PII disclosure, and validates the guardrails your agents are supposed to enforce.

Keeping your agents and prompts synced is what lets Hamming generate relevant scenarios and scores, so leave **Import Voice Agents** enabled unless you have a reason not to.

## Connect Synthflow to Hamming

You connect from inside Hamming by adding a Synthflow API key. Nothing needs to be configured on the Synthflow side beyond creating the key: connect the provider, sync your agents, then run your first calls.

In Synthflow, open **Admin** → **Workspace Settings** → **API Keys**, click **Create new API key**, and copy the value. See [Authentication](/authentication) for details and key-management practices. Keep the key secret and store it securely; you paste it into Hamming only once.

In Hamming, open the provider import screen and click **Add API key** on the **Synthflow** card.

![Hamming Import from Provider screen with a Synthflow card](https://storage.googleapis.com/granular-changelog/doc-images/hamming_import_from_provider.png)

In the **Set your Synthflow API key** dialog, give the credential a **Name**, paste your Synthflow key into the **Key** field, choose your [data sync options](#data-sync-options), then click **Create**.

![Set your Synthflow API key dialog in Hamming with data sync options](https://storage.googleapis.com/granular-changelog/doc-images/hamming_synthflow_api_key.png)

Click **Done** to finish. Hamming imports the resources you selected and keeps them synced automatically as you change agents and prompts in Synthflow, so you do not re-import after every edit. Your agents are then ready for test runs, monitoring, and red-teaming.

### Data sync options

The dialog controls which resources Hamming imports from Synthflow and keeps updated:

| Option                   | Default          | What it does                                                                                                                                  |
| :----------------------- | :--------------- | :-------------------------------------------------------------------------------------------------------------------------------------------- |
| **Import Voice Agents**  | On               | Syncs your voice agents and their prompts so Hamming can generate relevant scenarios and scores.                                              |
| **Import Phone Numbers** | On               | Imports phone numbers and their agent assignments to make outbound test runs easier to set up.                                                |
| **Import Calls**         | Off              | Imports historical [call](/about-call-logs) data and metrics for the monitoring dashboard.                                                    |
| **Default Agent Group**  | No default group | Adds newly synced agents to the group you pick. Existing grouped agents are preserved. Use an agent group here, not a provider credential ID. |

## FAQ

Create it in Synthflow under **Admin** → **Workspace Settings** → **API Keys**, then paste it into Hamming's **Set your Synthflow API key** dialog. See [Authentication](/authentication) for the full steps.

No. Once connected, Hamming keeps your synced agents, prompts, and phone numbers updated automatically, so scenario and score generation reflect your latest changes without a manual re-import.

No. Hamming reads your agents, phone numbers, and calls through the API key to run its own tests and monitoring. It does not modify your Synthflow agents or their prompts.

[Simulations](/simulations) run scripted test conversations inside Synthflow to validate an agent before launch. Hamming is an external platform that layers large-scale scenario generation, scoring, production monitoring, and security red-teaming on top of the same agents.

Importing calls brings your historical Synthflow call data and metrics into Hamming's monitoring dashboard, so production quality tracking and health checks reflect real traffic. Leave it off if you only want pre-launch testing.

Yes. Each Synthflow API key maps to one workspace, so add a separate credential in Hamming for each workspace you want to test or monitor.