Nehal Azhar
03/15/2025, 6:23 AMVapi Support Bot
03/15/2025, 6:24 AMfileIds
, searchType
, and scoreThreshold
2. Confirm that the file IDs of your knowledge base files are correctly added to the assistant's model configuration under knowledgeBase.fileIds
3. Ensure your file format is supported and doesn't contain problematic elements like complex formatting or non-text elements
4. Try clearing your browser cache and cookies or using a different browser
5. There might be syncing or indexing delays, so the assistant might recognize the files after some time
For creating a knowledge base with API (import option only):
This is actually expected behavior currently. There are syncing issues with creating knowledge bases directly through the VAPI API, which is why only the import option is available.
The recommended workaround is to:
1. Create your dataset directly in Trieve
2. Then import this dataset into VAPI using the import method
Here's a code example for importing from Trieve:
bash
curl -X POST https://api.vapi.ai/knowledge-base \
-H "Authorization: Bearer YourVAPIApiKey" \
-H "Content-Type: application/json" \
-d '{
"name": "Your Knowledge Base Name",
"provider": "trieve",
"searchPlan": {
"searchType": "semantic",
"topK": 3,
"removeStopWords": true,
"scoreThreshold": 0.7
},
"createPlan": {
"type": "import",
"providerId": "YourTrieveDatasetID"
}
}'
Let me know if you need any clarification on these steps!
Sources:
- [Discord Thread about file reading issues](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1336306333661663243)
- [Discord Thread about file linking](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1336727607533899786)Vapi Support Bot
03/15/2025, 6:24 AMNehal Azhar
03/15/2025, 6:48 AMNehal Azhar
03/15/2025, 6:51 AMNick K - CEO @ Trieve
03/15/2025, 9:53 PM