How to create scenarios documentation in make using some AI tools

Questions
,
1

:bullseye: What is your goal?

Detailed documentation for scenarios using tools in Make.

:thinking: What is the problem & what have you tried?

Documentation creation for scenarios

3

Export the scenario blueprint mad feed it to Claude for documenting.

A more advanced way will be to add Claude skills.

An even more advanced way would be to create a Make scenario and use an agent to pull the scenario blueprint, some execution logs and based on the knowledge provided, document the scenario.

4

Hi @Ganesh1

There are a few ways I create documentations using tools in make depending on what you intend to use it for.

For very basic non logging documentation i.e. information that you use on the go, there’s an add notes option that allows you to describe exactly what you’re doing and how it’s going. This is manual and you don’t log anything anywhere you just just open your notes and update it with What you’re currently doing. You can also rename all your modules and give them very descriptive names and even add module to module descriptions that can be very handy and makes the scenarios easy to debug and resuse. Another hack that might need an external tool is to export your scenario blueprint and feed it to an AI and analyze it and document as deep as needed.

If this doesn’t satisfy the question then read further.

On the other hand if you need a general or a deeper level scenario documentations that comes with adequate logging or producing external documentats, one of the best way is to use an AI agent although it could still get extremely manual depending on the depth of the documentation (workflow, dataflow, impact, automatic documentations for every scenario run, etc.)

-Using an AI moduke/Agent:

Let’s say we’re dealing with a scenario that is a basic email support scenario and watches a mailhook for incoming emails to trigger it, then triages the email into 4 categories (business casual, business support, business escalation and business irrelevant), then it responds only to business casual and business support. It then sends “business escalate” to Slack for human response and completely ignores business irrelevant emails. The scenario then logs every successful response into your Google sheets database and irrelevant emails logged on anoher sheet..

NB: All what I have written so far in the point above can be a standalone documentation of the scenario workflow). But let’s take it a few steps further to module to module documentation.

The modules we are dealing with right here are:

  1. A custome mailhook (recieves the email immediately and triggers the scenario.
  2. The email is mapped directly to a Google Gemini get an AI response module. (This module already has business rules, logic and data needed to triage/classify incoming emails into required categories)
  3. Immediately after the email classifier module we have a 4 branch router with filters that routes the data to the right. 3 branches for casual, support, and escalate and the fourth branch used as a fall back for business Invalid.
  4. On the first branch the filter let’s business support emails pass through and meets a Google Gemini generate a response module that takes the output from the triage agent and generates+parse a JSON that has the centent to send to the customer.
  5. Immediately after it is a send an email module where the sender, conversationID, headline and body are mapped directly from the module to the send an email module to reply to the customer.
  6. After the send an email module you add another Google Gemini generate a response module and describe your entire scenario then map the module outputs into it something like: (Mailhook data: “emailID, Body, senderemail, etc” Triage Agent: “Response” Email response Agent “Response” Tokens used: “you map tokens of every module here”) Then you give it a suitable role like: "You’re a make.com workflow documentation expert, your role is to monitor every Business Support run of this scenario and produce a very deep documentation of the entire workflow run plus what this run means for our business. Break your documentation into 3 parts, and your response into a 7 parts json; Tokens used, date etc…).
  7. At the end of that documentation AI module we add a Google sheet that it logs the documentation into.

Or if we’re not using sheets we can append our per run documentation into a Google docs. We repeat the same for all 4 branches the Slack branch has the forst AI module generate a Slack response and a Slack module to forward it to Slack followed by the documentation module and the logging module. The invalid branch might only send an email informing the scenario owner of an invalid email and the documentation ai module plus the logging module.

This looks hectic, manual, and monotonous but the outcome can be magnificent if you’re dealing with extremely high stakes customers like maybe legal advice, finance, or medical related services. Per run documentation will tell you everything that happened when every single customer got a response from your scenario and the documentation agent might even capture errors in operational inefficiencies.

So with Make the possibilities cab be endless but since the question seems a little ambiguous, a deeper case by case answer should cover most use cases using Make tools for documentation. Hopefully if profitable, in the future make might roll out run by run documentation features for a toggle on documentation experience for enterprises and maybe teams.

Regards.