Chat API
The Doti API allows developers to integrate AI-powered, context-aware search and reasoning into their own systems. From embedding search bars in internal tools to automating ticket triage in Salesforce - Doti enables intelligent, secure access to organizational knowledge through a simple, scalable API.
Authentication
All requests to the Doti API require a Bearer token. Include it in the Authorization
header:
Authorization: Bearer {YOUR_ACCESS_TOKEN}
❗ Tokens must be generated by the Doti team. They are not yet user-creatable via the Portal.
Token Scoping: Agents vs Collections
Every Doti API token is scoped to either an Agent or a Collection, depending on how you want the query to behave.
Agent-Scoped Tokens
When a token is tied to a Doti Agent, it inherits the agent’s entire configuration:
Search scope (data sources & integrations)
Custom instructions (tone, formatting, persona)
Sequential reasoning steps (agentic workflows)
Live updates included: Any change made to the Agent (like updated instructions, new integrations, added Dots) immediately takes effect for all requests using that token — no redeployment needed.
This makes agent-scoped tokens ideal for long-term, evolving use cases like:
Embedded assistants
Autopilot integrations
GPT-based decision makers
Collection-Scoped Tokens
Alternatively, tokens can be scoped to one or more Collections. This is useful when:
You want minimal configuration overhead
You need to expose just a slice of the knowledge base
You don’t need multi-step or formatted responses
You control exactly which resources the API can query - without inheriting an Agent's instructions or behavior.
API Endpoints
🧠 GPT (AI-Powered Reasoning)
Use this endpoint when you want GPT-style generation (summarization, synthesis, drafting), optionally guided by a prompt.
Endpoint:
POST https://api.doti.ai/api/v2/gpt
Request:
{
"prompt": "Summarize in 3 bullets", // Optional
"query": "What’s the current escalation policy?"
}
Response:
{
"message": "1. Escalations go to team leads first...",
"query": "What’s the current escalation policy?"
}
Internal Search
Use this to search across your internal tools (Slack, Confluence, Notion, etc.) and receive answers + citations.
Endpoint:
POST https://api.doti.ai/api/v2/search/internal
Request:
{
"query": "Where can I find the onboarding checklist?"
}
Response:
{
"qid": "abc123",
"question": "Where can I find the onboarding checklist?",
"answer": "You can find it in the Onboarding Handbook [1].",
"sources": [
{
"name": "Onboarding Handbook",
"link": "https://notion.so/doc/onboarding",
"relevancy": "high"
}
]
}
Rate Limiting
Max tokens per request:
40,000
Rate limits apply at the token level - aggressive workloads should batch efficiently.
🚨 Error Handling
200
✅ Success
400
❌ Bad Request – e.g., missing query
401
🔒 Unauthorized – invalid or missing token
403
🚫 Forbidden – valid token, but no access to query scope
429
⏳ Rate Limit Exceeded
Use Cases
Proactive Support Automation (Zendesk, Salesforce)
Integrate Doti via Workato, Zapier, or Make to:
Trigger Doti when new tickets arrive
Auto-generate suggested responses using search or GPT
Attach replies to the ticket for agent review or auto-send
💡 This allows zero-touch handling of known issues using internal knowledge.
Embedded Assistants in Internal Tools
Embed Doti in:
Internal portals
Knowledge systems
Dev tools or dashboards Allow users to query agents or collections via an internal UI.
Augment Any Chat UI
Integrate Doti into:
Intercom
Slack bots
Microsoft Teams
Let your existing interfaces become smarter by piping in Doti’s answers + sources.
Getting Started
Request token(s) via your Doti success team
Define if your use case requires:
Agent-level logic (for sequenced reasoning or permission enforcement)
Collection-level scoping (for lean, specific use cases)
Implement using the endpoint(s) above
Summary
GPT Endpoint
AI-based reasoning with optional prompt
Internal Search
Real-time, permission-aware knowledge retrieval
Scoping
Token can inherit Agent behavior or point to specific collections
Authentication
Bearer token (provisioned by Doti team)
Rate Limits
40,000 tokens/request
Use Cases
Ticketing automation, embedded assistants, live knowledge lookup
🛠️ The Doti API makes your apps, tickets, and bots smarter - with no need to rebuild from scratch. Plug in, power up, and let your workflows think for themselves.
Last updated
Was this helpful?