Make

Tracira has a native Make app. Add monitoring to any scenario with a single module, no HTTP setup required.

Verdict response

With Wait for the verdict, the module returns:

{
  "ok": true,
  "id": "b1c2d3e4-...",
  "status": "pass",
  "verdict": "pass",
  "confidenceScore": 0.95,
  "explanation": "The reply stays within the 20% discount cap."
}

Branch on Status: pass means nothing needs a person, flagged means the output waits on the Review page, and error means the check could not finish. Verdict (pass, fail or unsure) and Explanation say why.

Holding an action for approval

When your AI decides to do something with side effects (issue a refund, delete a record), set After check to Wait for a human to approve. That reveals Action Name and Action Summary, both required, plus optional Action Parameters, Callback URL and Fire callback on.

The two halves do different jobs. The action describes what is being reviewed: the summary is what reviewers read verbatim to approve or reject, so write it as a clear sentence like Refund EUR 49.00 to Alice Martin (order #8841). The choice itself decides whether a person is asked at all, and it holds the output on the Review page whatever your rules conclude. Tracira never runs the action: your scenario does, once a person approves.

Rules still run either way, and they remain the right tool when only some actions need a person: a data-field rule on action.params.amount flags refunds over a threshold and lets smaller ones through. When a rule flags the output, its explanation is what the reviewer reads.

To act on the decision, start a second scenario with Watch decisions (below), or set a Callback URL pointing at a Make webhook. You can also leave AI Output empty when there is no message to review and only a step to approve.

Reacting to decisions: Watch decisions

Watch decisions is an instant trigger that starts a scenario the moment a person decides in Tracira. By default it fires on Approved by a human, Rejected by a human and Edited by a human. You can also select Taken over by a human (a reviewer handled it themselves, so nothing should wait for an approval), Taught after the fact (a reviewer coached the AI after an output already went out), Flagged for review, Passed all rules and Evaluation error. Saving the trigger registers it with Tracira automatically; there is nothing to set up in the dashboard.

The Decision it carries uses the API values: approved, rejected, changed (edited) and handled (taken over). When a reviewer edits an output, Output already holds their corrected version, so a scenario that maps Output delivers what the human approved. Corrected Output and Original AI Output are there when you need to tell the two apart.

Scenario A: [ AI drafts ] → [ Tracira: Send an output ]            (submit for review)
Scenario B: [ Tracira: Watch decisions ] → [ Filter: Decision = approved ] → [ Send the reply ]

Logging a conversation

For a chatbot or support thread, send one output per exchange - do not paste the full message history into every output:

• Input Text: only the newest user message (the one your AI just replied to)
• AI Output: the reply to it
• Session ID (under Advanced settings): your conversation or thread ID - the same value on every turn

Tracira stitches all outputs sharing a Session ID into a single conversation thread, so reviewers read the whole exchange in order with nothing repeated. Your AI request can still carry the full history - Tracira only needs the new turn.

Leave your system prompt out entirely: it's configuration, not conversation. The best place for it is Tracira itself - see Hosted instructions below.

Hosted instructions (self-improving prompts)

Tracira can host the instructions (system prompt) your AI runs with, so the prompt lives next to the people reviewing its output instead of buried inside a scenario module or a spreadsheet. Reviewers see, edit, and restore every version on the AI instructions page in the dashboard; your scenario always drafts with the latest.

[ Gmail ] → [ Tracira: Get instructions ] → [ AI drafts ] → [ Tracira: Send an output ]

Attaching files

For small files, map them into Input Attachments (files the AI worked from) or Output Attachments (files the AI produced) on the Send an output module. Pick Upload file and map the binary, or From URL and paste an HTTPS link. Tracira auto-detects whether each attachment is an image, audio, or document.

Whenever an output carries more than one file, fill in What is it? on each one: Before photo, After photo, Signed contract. The reviewer sees that above the file instead of IMG_0042.jpg, and it comes back on the attachment in Watch decisions, so a later step can pick out the file it needs without matching file names.

The label describes the file's role in that output, so it goes on the attachment, not on the upload:

Large files: the Upload a file module

Inline attachments travel inside the Send an output request, which is capped at 4.5 MB. For larger files (a multi-page scanned PDF, a high-res image) add the Upload a file module before Send an output. It uploads the binary straight to Tracira storage (no size cap from the request body) and returns a key, supporting files up to 32 MB.

[ AI module ] → [ Tracira: Upload a file ] → [ Tracira: Send an output ]

Getting a file back: the Download a file module

Every event on the Watch decisions trigger carries an Attachments array listing the files stored on that output, each with a File name, a File Key, a File URL, and the What is it? label the sender gave it.

That is what makes a redo possible for document work. A scenario that parsed a PDF has finished long before the reviewer reads it, so when they send it back with a comment the scenario no longer holds the file. Add Tracira → Download a file, map the attachment's File URL (or File Key) into File, and it comes back as binary data you can feed straight to your AI module along with the Reviewer Comment.

[ Tracira: Watch decisions ] → [ Tracira: Download a file ] → [ AI module ] → [ Tracira: Send an output ]

On that final Send an output, set Revision Of to the original Output ID so the two versions form a chain, and under Input Attachments or Output Attachments add an item with source Already in Tracira, mapping the same File Key. The document stays on the new version without being uploaded again.

Updating an output afterwards

Some details only exist after the output has been sent: the ticket number your CRM assigns a second later, the category a downstream step decides, the outcome of the call. Add Tracira → Update an output wherever that detail appears and write it onto the output.

• Metadata rows merge into what is already stored: only the keys you list change, and a row with an empty value removes that key.
• File Labels rename a file already on the output. Map its File Key and say What is it?.
• Session ID, Subject ID and Actor ID (under advanced settings) can be filled in later too.

[ Tracira: Send an output ] → [ CRM: Create ticket ] → [ Tracira: Update an output ]