Simpli Assist

Assist is your agents' AI copilot. It combines knowledge base articles, macros, conversation history, and customer context to surface the most relevant suggestions in real-time. Agents resolve tickets faster with intelligent, contextual recommendations.

Simpli Assist provides contextual suggestions, next-action recommendations, and rich context panels for support agents using LLM-powered analysis.

Configuration

Variable	Default	Description
`APP_PORT`	`8010`	Server port
`LITELLM_MODEL`	`openai/gpt-5-mini`	LLM model for suggestions
`MAX_SUGGESTIONS`	`5`	Maximum number of suggestions to return
`CONTEXT_WINDOW_SIZE`	`10`	Number of recent messages to consider
`CORS_ORIGINS`	`*`	Allowed CORS origins (comma-separated)

Start the server

simpli-assist serve

API endpoints

All endpoints are under the /api/v1 prefix.

POST `/api/v1/suggest`

Get contextual suggestions for the current conversation.

Request:

{
  "ticket_id": "T-321",
  "conversation": [
    {"role": "customer", "content": "How do I export my data as CSV?"}
  ],
  "agent_id": "agent-42"
}

The role field accepts values from the AuthorType enum: customer, agent, or system.

Response:

{
  "suggestion_id": "sg-abc123",
  "suggestions": [
    {
      "type": "kb_article",
      "content": "Exporting Your Data — Navigate to Settings > Export to download your data as CSV...",
      "confidence": 0.92,
      "source": "kb-article-47"
    },
    {
      "type": "reply",
      "content": "You can export your data as CSV from Settings > Export. Select the date range and click Download.",
      "confidence": 0.88,
      "source": null
    },
    {
      "type": "macro",
      "content": "data_export_instructions",
      "confidence": 0.75,
      "source": "macro-12"
    }
  ]
}

Suggestion types: reply (draft response), macro (template to apply), kb_article (relevant article), escalation (escalation recommendation).

POST `/api/v1/next-action`

Recommend the next best action for a conversation.

Request:

{
  "ticket_id": "T-321",
  "conversation": [
    {"role": "customer", "content": "This is the third time I've reported this!"},
    {"role": "agent", "content": "I sincerely apologise for the repeated issue."}
  ],
  "current_status": "open"
}

Response:

{
  "action_id": "a-def456",
  "recommended_action": "escalate",
  "reasoning": "Customer has reported this issue multiple times, indicating a recurring problem that needs senior attention.",
  "confidence": 0.85,
  "alternatives": ["apply_macro", "reply"]
}

POST `/api/v1/context`

Build a rich context panel for the current conversation.

{
  "ticket_id": "T-321",
  "conversation": [
    {"role": "customer", "content": "I can't export my data"}
  ],
  "customer_id": "cust-99"
}

GET `/health`

Health check.

Integration example

import httpx

client = httpx.Client(base_url="http://localhost:8010")

# Get suggestions for current conversation
suggestions = client.post("/api/v1/suggest", json={
    "ticket_id": "T-456",
    "conversation": [
        {"role": "customer", "content": "How do I cancel my subscription?"},
    ],
}).json()

for s in suggestions["suggestions"]:
    print(f"[{s['type']}] {s['content']} ({s['confidence']:.0%})")

# Get next action recommendation
action = client.post("/api/v1/next-action", json={
    "ticket_id": "T-456",
    "conversation": [
        {"role": "customer", "content": "I want to cancel immediately"},
    ],
}).json()

print(f"Recommended: {action['recommended_action']}")
print(f"Reason: {action['reasoning']}")

Next steps

The Ticket Lifecycle — See how Assist fits into the agent workflow
Integration Overview — Embed Assist in your helpdesk's agent panel
Model Selection — Choose the right model for real-time suggestions

On this page