IntegrationsProductivity Tools

Productivity Tools

Connect Transient Intelligence to where your documents already live. The pattern is the same in each case: retrieve document content from the source, pass it to /v1/answer, return the cited result.

Pattern

Notion page updated → Make/Zapier fetches page blocks as plain text → POST to /v1/answer → write cited answer back to a Notion database property or comment.

Setup (via Make)

1. Trigger: Notion → Watch Database Items (new or updated row).
2. Add a Notion → Get a Page module to retrieve the full block content.
3. Map the plain-text body from the page blocks into the input field below.
4. Set question from a fixed database property (e.g. an "Analysis question" field on the row).
5. Write citations[0].quote and answer.summary back to a Notion property.

HTTP body

{

"question": "What are the key decisions recorded in this page?",

"input": {{notionPageContent}},

"top_k": 5

}

Pattern

New file lands in a Drive folder → Make/Zapier downloads the file as plain text or exports a Google Doc → POST to /v1/answer → result posted to Slack, Sheets, or email.

Setup (via Make)

1. Trigger: Google Drive → Watch Files in a Folder.
2. Add Google Drive → Download a File (or Export a Google Doc as plain text).
3. Pass the file content as input to /v1/answer.
4. Route the citations response to Slack, a Google Sheet row, or an email via a downstream module.

Supported file types: .pdf, .docx, .xlsx, .csv, .txt, .json. For Google Docs, export as text/plain before passing as input.

cURL — file ingestion and query

curl -X POST https://api.transientintelligence.com/api/models/v1/answer \

-H "x-api-key: $TI_API_KEY" \

-H "Content-Type: application/json" \

-d '{ "question": "Summarise the action items", "input": "'$FILE_CONTENT'", "top_k": 5 }'

Pattern

Two common approaches: a slash command (/ti ask) that takes a question and returns cited answers inline, or a bot workflow triggered by a file share that auto-analyses the document and posts a summary.

Slash command (via n8n or Pipedream)

1. Create a Slack app with a Slash Command: /ti, pointing to your webhook URL.
2. In n8n: Webhook node receives the Slack payload. Parse text as the question and channel_id for the reply target.
3. Pass question + your session context to /v1/answer.
4. Format the citations response and POST back to Slack via chat.postMessage.

Slack response formatter (JavaScript)

const { answer, citations } = await tiResponse.json();

const blocks = [

{ type: "section", text: { type: "mrkdwn", text: answer?.summary ?? "No answer returned." } },

...citations.map(c => ({

type: "context",

elements: [{ type: "mrkdwn", text: `> ${c.quote} — ${c.locator}` }]

}))

];

await fetch(`https://slack.com/api/chat.postMessage`, {

method: "POST",

headers: { Authorization: `Bearer ${SLACK_BOT_TOKEN}`, "Content-Type": "application/json" },

body: JSON.stringify({ channel: channelId, blocks })

});

Example response

This is what TI returns for a typical document question. Use the citations array — not answer.summary alone — as the primary output in downstream steps.

Example — /v1/answer response

{

"answer": {

"summary": "The Q3 report identifies three action items: cost reduction in logistics, a hiring freeze in non-core departments, and a 60-day review of the APAC pipeline.",

"confidence": 0.91

"citations": [

{ "quote": "Logistics cost reduction target: 12% by end of Q4.", "locator": "p.4 §2", "relevance": 0.96 },

{ "quote": "Hiring freeze effective immediately across all non-core functions.", "locator": "p.7 §1", "relevance": 0.93 },

{ "quote": "APAC pipeline under 60-day strategic review.", "locator": "p.9 §3", "relevance": 0.88 }

"run_id": "run_01jt4xk29mfq"

}

Error handling patterns

Empty citations — no evidence found

If citations is an empty array, TI found no supporting passages. Route this to a separate path in your flow — e.g. post a 'No evidence found' message to Slack, or flag the Notion row for manual review. Never forward answer.summary from an empty-citations response as if it were grounded.

401 Unauthorized

The x-api-key header is missing or invalid. Check the key is set in your platform's credential store (Make Connection, Zapier auth, n8n credential) rather than hardcoded in the body.

413 Payload Too Large

The input text exceeds the per-request limit. Split long documents into sections and run multiple calls, or use the /v1/upload → /v1/ask async flow for large files.

Google Docs — garbled input text

Always export Google Docs as text/plain before passing as input. Exporting as HTML or application/vnd.google-apps.document sends markup that inflates token count and degrades retrieval quality.