coolmen468
coolmen4683mo ago

Agent: execution failed

I have a link between flow bridge and Voiceflow so people can use Whatsapp to communicate with the ChatBot. Someone asks if they can drink something Thursday, which is possible. But before the Agent can react it goes into an error and doesn't response. I've tried to send the same messages but it worked for me fine. Here is the error message that is visable in transcript:
[Agent]: execution failed
{
"ref": {
"type": "agent",
"nodeID": "67fe8e0f561b31690c3a2a7e",
"agentID": "67e808eeb3cadc6ecc9537d5",
"diagramID": "63067e968cbdb692c97b063a"
},
"metadata": {
"error": {
"status": 404,
"headers": {
"alt-svc": "h3=\":443\"; ma=86400",
"cf-cache-status": "DYNAMIC",
"cf-ray": "954ede162bec82de-IAD",
"connection": "keep-alive",
"date": "Tue, 24 Jun 2025 20:10:15 GMT",
"server": "cloudflare",
"set-cookie": "__cf_bm=szASQLkDnriJHI9BqRaRrjUZnPa1pFy_l0mpjrfxTRM-1750795815-1.0.1.1-vcAhQjEFTG5wn3CvKs.dGIU6YcHQU0iyeU3n__Dp6BFlyDN_s0U4hqh6L2kWhOm3zL94zLkjCZi4e8Wvje7lMiImK8Af0f8g5zxDe52CNKY; path=/; expires=Tue, 24-Jun-25 20:40:15 GMT; domain=.api.openai.com; HttpOnly; Secure; SameSite=None, _cfuvid=wE47v1eF0NOPU4lBv7BBE9k7B9A1srGa3Sm6OpPvGt4-1750795815445-0.0.1.1-604800000; path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None",
"strict-transport-security": "max-age=31536000; includeSubDomains; preload",
"transfer-encoding": "chunked",
"x-content-type-options": "nosniff"
}
}
}
}
[Agent]: execution failed
{
"ref": {
"type": "agent",
"nodeID": "67fe8e0f561b31690c3a2a7e",
"agentID": "67e808eeb3cadc6ecc9537d5",
"diagramID": "63067e968cbdb692c97b063a"
},
"metadata": {
"error": {
"status": 404,
"headers": {
"alt-svc": "h3=\":443\"; ma=86400",
"cf-cache-status": "DYNAMIC",
"cf-ray": "954ede162bec82de-IAD",
"connection": "keep-alive",
"date": "Tue, 24 Jun 2025 20:10:15 GMT",
"server": "cloudflare",
"set-cookie": "__cf_bm=szASQLkDnriJHI9BqRaRrjUZnPa1pFy_l0mpjrfxTRM-1750795815-1.0.1.1-vcAhQjEFTG5wn3CvKs.dGIU6YcHQU0iyeU3n__Dp6BFlyDN_s0U4hqh6L2kWhOm3zL94zLkjCZi4e8Wvje7lMiImK8Af0f8g5zxDe52CNKY; path=/; expires=Tue, 24-Jun-25 20:40:15 GMT; domain=.api.openai.com; HttpOnly; Secure; SameSite=None, _cfuvid=wE47v1eF0NOPU4lBv7BBE9k7B9A1srGa3Sm6OpPvGt4-1750795815445-0.0.1.1-604800000; path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None",
"strict-transport-security": "max-age=31536000; includeSubDomains; preload",
"transfer-encoding": "chunked",
"x-content-type-options": "nosniff"
}
}
}
}
Can someone figure out what this error mean and how I can make sure it won't happen again. Thanks in advance!
3 Replies
Tico | Voiceflow Assistant
The 404 coming back from the Agent step isn’t about WhatsApp or Flow Bridge – it’s the Voiceflow runtime telling you it can’t find the model/version your session is pointing to. Why it happens • Each time you click the ▶️ Run button (or Publish to prod) Voiceflow renders a new version of your assistant.
• If a user is still on an old session that references the previous version-ID and you have since re-rendered, the next Agent call will try to reach a model that no longer exists → OpenAI returns 404 and the Agent step throws execution failed. That explains why it failed for the first user but worked when you re-tested – your new test started a fresh session that pointed to the latest rendered model. How to prevent it 1. After every change, always click Run (or Publish) before sending traffic to the bot.
2. Try not to re-render while live users are in the middle of a conversation – their sessions will still point to the previous (now missing) version.
3. For production traffic, use the Production environment and only publish when you’re ready, so version IDs stay stable during day-to-day edits in Development.
4. If you’re using the Runtime API directly, double-check the versionID you pass matches an existing rendered version. Once the version is rendered and stable the execution failed – status 404 error disappears. Documentation: https://docs.voiceflow.com
Steve B. | IntelleSync.ai
Could be an LLM outage try a different provider and see if that works.
coolmen468
coolmen468OP3mo ago
This is good to keep in mind. We do use the Production environment, we published some things Monday, but this error happend yesterday(Tuesday). Nothing was changed or even touched on Tuesday, so how could the version ID be different if it was Published the day before? Its most likely not a LLM outage, because it works now without changing or editing anything? No Run or Publish happend, it just "fixed" itself? I just want to make sure it won't happen again.

Did you find this page helpful?