analysis is not working it is getting breaked
# support
g
structring data not working also sucess analysis
v
Try resetting your assistant's
analysisPlan
to the default prompts and schema to rule out configuration issues.
For example, set your analysis plan like this:
Copy code
json
{
  "summaryPrompt": "",
  "structuredDataPrompt": "",
  "successEvaluationPrompt": ""
}
This will use the default analysis logic. If analysis still fails, check the call dashboard for error details or review your API logs for more information. [See documentation for details](https://docs.vapi.ai/assistants/call-analysis) Source: - [Call analysis documentation](https://docs.vapi.ai/assistants/call-analysis)
k
This usually stem from misconfigured prompts, JSON schemas, or unclear evaluation criteria, kindly review and correct these settings to fix the problem.
v
@User I am also facing same issue since this morning.
z
How this has happened? as it worked fine this morning..
s
The main issue is that while the analysis plan contains system messages in each section, it's missing user messages. Anthropic's API requires at least one user message in addition to system messages to properly function. Solution We need to add a user message to each of the enabled plan sections. @Vinita @zurpol @ghostdragon can you guys update your analysis plan and include the relevant user message in
messages
array of summary, successEval and structure data plan until this is fixed.
Copy code
json
//template
{
  "analysisPlan": {
    "summaryPlan": {
      "enabled": true,
      "timeoutSeconds": 5,
      "messages": [
        {
          "role": "system",
           //  replace with your instructions here if required 
          "content": "You are an expert note-taker. You will be given a transcript of a call. Summarize the call in 2-3 sentences. DO NOT return anything except the summary."
        },
        {
          "role": "user",
          "content": "Here is the transcript:\n\n{{transcript}}\n\n. Here is the ended reason of the call:\n\n{{endedReason}}\n\n"
        }
      ],
      "useAssistantLlm": false,
      "customLlmUrl": null
    },
    "structuredDataPlan": {
      "enabled": true,
      "timeoutSeconds": 5,
      "schema": {
                   //  replace with your schema here if required 
        "type": "object",
        "properties": {
          "mainTopic": {
            "type": "string",
            "description": "The main topic discussed in the call"
          },
          "customerIntent": {
            "type": "string",
            "description": "The primary goal or intent of the customer during the call"
          },
          "actionItems": {
            "type": "array",
            "items": {
              "type": "string"
            },
            "description": "Action items identified during the call"
          },
          "customerSentiment": {
            "type": "string",
            "enum": ["positive", "neutral", "negative"],
            "description": "The overall sentiment of the customer during the call"
          }
        },
        "required": ["mainTopic", "customerIntent", "actionItems", "customerSentiment"]
      },
      "messages": [
        {
          "role": "system",
          "content": "You are an expert data extractor. You will be given a transcript of a call. Extract structured data per the JSON Schema. DO NOT return anything except the structured data.\n\nJson Schema:\n{{schema}}\n\nOnly respond with the JSON."
        },
        {
          "role": "user",
          "content": "Here is the transcript:\n\n{{transcript}}\n\n. Here is the ended reason of the call:\n\n{{endedReason}}\n\n"
        }
      ],
      "useAssistantLlm": false,
      "customLlmUrl": null
    },
    "successEvaluationPlan": {
      "enabled": true,
      "timeoutSeconds": 5,
      "rubric": "PassFail",
      "messages": [
        {
          "role": "system", 
          "content": "You are an expert call evaluator. You will be given a transcript of a call and the system prompt of the AI participant. Determine if the call was successful based on the objectives inferred from the system prompt. DO NOT return anything except the result.\n\nRubric:\n{{rubric}}\n\nOnly respond with the result."
        },
        {
          "role": "user",
          "content": "Here is the transcript:\n\n{{transcript}}\n\n"
        },
        {
          "role": "user",
                      //  replace with your instructions here if required 
          "content": "Here was the system prompt of the call:\n\n{{systemPrompt}}\n\n. Here is the ended reason of the call:\n\n{{endedReason}}\n\n"
        }
      ],
      "useAssistantLlm": false,
      "customLlmUrl": null
    }
  }
}
Most likely because of our changes, things went wrong.
g
ahmm
confusing
some time it works sometime dont
o
@Shubham Bajaj that worked we've just tried it. but when we republish the agent via the UI it loses those user messages. if we don't touch it, it seems to stay working.
s
You just need to replace the schema plan, and analysis plan where required and then using this update your assistant analysis plan.
Yeah issue on our side, please use API.
g
tell me one thing can we re run only analysis part somehow ???
@Shubham Bajaj
like i call api and it do analysis again
s
@ghostdragon Regrettably, you are unable to re-run the analysis plan through our platform. However, you can use the above shared messaging format for different parts of the analysis plan and then pass it to your LLM of your choice to get the same results from it.
z
@Shubham Bajaj Has this been fixed now, please?
p
hey @zurpol, yes the issue has been fixed, and it will deployed to the production by end of today.
g
@Shubham Bajaj it is fixed?
p
Yes, it is fixed.
Make sure to update your assistant by pressing publish button.
g
i dont have to do that api stuff which @Shubham Bajaj told me to do
p
Yup, that's right. You don't need to do any API stuff. Just go to your assitant and press on publish button and it will work.
g
test not working
please help me out i need to do this by today
v
If your test is still not working after pressing the "Publish" button in your assistant UI, try the following: - Double-check that you have published the latest version of your assistant. - If the issue persists, please contact Vapi support with details of your test failure, as there is no documented way to re-run call analysis via the API or further troubleshoot from the UI alone. This ensures the backend fix is applied, and support can assist if problems continue. Sources: - [Vapi FAQ](https://docs.vapi.ai/faq) - [Vapi Test Suites Documentation](https://docs.vapi.ai/test/test-suites)
Could you please describe what happens when you run your test after republishing—are you seeing an error message or unexpected behavior? This will help me assist you further.
If you've already republished your assistant and are still seeing "N/A" for extracted data, the best next step is to contact Vapi support with your assistant ID and a screenshot of the issue. This will help the team investigate any lingering backend or configuration problems specific to your account. You can reach support via the [Vapi support page](https://docs.vapi.ai/support). Tip: Double-check that your assistant's analysis plan is enabled and properly configured in the dashboard before reaching out. Sources: - [Vapi Call Analysis Documentation](https://docs.vapi.ai/assistants/call-analysis) - [Vapi Support](https://docs.vapi.ai/support)
If your test is still not working after republishing your assistant, you can check the status and results of your test suite run using the API:
Copy code
bash
curl -X GET "https://api.vapi.ai/test-suite/<testSuiteId>/run/<id>" \
  -H "Authorization: Bearer <your_token>"
Replace
<testSuiteId>
,
<id>
, and
<your_token>
with your actual values. This will help you see if there are any errors or issues with your test run. If problems persist, consider reaching out to [Vapi support](https://docs.vapi.ai/support) for further help. Sources: - [Get Test Suite Run](https://docs.vapi.ai/api-reference/test-suite-runs/test-suite-run-controller-find-one?explorer=true) - [Vapi Support](https://docs.vapi.ai/support)
To resolve the call analysis issue, go to your assistant in the Vapi dashboard and press the "Publish" button. This will apply the latest fix and should restore structured data and success evaluation in your tests. If you have already published and the problem persists, please reach out to Vapi support for urgent assistance: [support@vapi.ai](support@vapi.ai) or via [Discord](https://discord.com/invite/pUFNcf2WmH). Sources: - [Vapi Support Documentation](https://docs.vapi.ai/support) - [Vapi Test Suites Documentation](https://docs.vapi.ai/test/test-suites)
Please try this: Go to your assistant in the Vapi dashboard, press the Publish button again, and then re-run your test. This should trigger the updated call analysis and structured data extraction. If you still see "N/A" for extracted data, double-check that your assistant's
analysisPlan
includes a valid structured data prompt and schema. If the issue persists, please share your assistant's configuration or contact support for urgent help. Source: - [Vapi Call Analysis Documentation](https://docs.vapi.ai/assistants/call-analysis)
Thanks for letting us know. Could you please share what error or unexpected behavior you see when running the test after republishing your assistant? This will help us diagnose the issue more effectively. Source: - [Test Suites Documentation](https://docs.vapi.ai/test/test-suites)
p
checking if this is resolved/solved for you?
g
can i give assitant id ?
v
Yes, you can provide your assistant ID to Vapi support for further help. To do this, visit the [Vapi support page](https://docs.vapi.ai/support) and submit your assistant ID along with a description or screenshot of the issue. This will help the support team investigate and resolve your problem as quickly as possible. Sources: - [Vapi Support](https://docs.vapi.ai/support) - [Vapi Call Analysis Documentation](https://docs.vapi.ai/assistants/call-analysis)
Yes, you can provide your assistant ID to Vapi support for further help. Please send your assistant ID (and a screenshot of the issue, if possible) to [support@vapi.ai](support@vapi.ai) or post it in the [Vapi Discord #support channel](https://discord.com/invite/pUFNcf2WmH) for urgent assistance. This will help the support team investigate and resolve your issue as quickly as possible. Source: [Vapi Support Documentation](https://docs.vapi.ai/support) Sources: - [Vapi Support Documentation](https://docs.vapi.ai/support) - [Vapi Call Analysis Documentation](https://docs.vapi.ai/assistants/call-analysis)
g
@Shubham Bajaj
v
To get help with your issue, please provide your assistant ID and a description or screenshot of the problem to Vapi support via one of these official channels: - [Vapi support page](https://docs.vapi.ai/support) - Email: support@vapi.ai - [Vapi Discord #support channel](https://discord.com/invite/pUFNcf2WmH) This will ensure your request is seen and handled by the support team as quickly as possible. Tagging individual team members (e.g., "@Shubham Bajaj") is not documented as an official support method. Sources: - [Vapi Support Documentation](https://docs.vapi.ai/support) - [Vapi Call Analysis Documentation](https://docs.vapi.ai/assistants/call-analysis)
p
We prefer a call id from the call logs section of the dashboard, but if you only have the assistant id that could also be helpful.
g
@Shubham Bajaj it is still not working please help
I am facing this issue since long time please please resolve this
bd3dd797-9905-4fd6-8208-110263404131 @Shubham Bajaj
g
it is fixed
p
Glad the issue was resolved. Please reach out to us if you have any other questions regarding this issue
3 Views