How to get better results using ChatGPT and Make.com

[Ryan_Baillie]

What are you trying to achieve?

Automate tailoring my resume using an automated workflow and chat gpt prompts.

Right now it takes a job description from a website URL pasted into a column on google sheets. It uses a Get Request to pull the job description and there are prompts in chat gpt to tailor my resume however there are issues. There’s much more than just the important part of job description like: info at the top and bottom of the webpage, links for buttons etc. - this is causing major issues for chatgpt.

• it needs a lot of tokens to complete a request
• the quality of the output is not as thorough as if I used the same prompts in chat gpt with only the job descriptions.

I believe this could be improved using filters or advanced parsing methods but I’m stumped on making custom filters. It requires a code language im not familiar at all, I think it’s called an aray? Chat gpt seems to have issues generating code for it as well. Has anyone else been in a similar workflow or know where the issues are coming from here?

Steps taken so far

I built the workflow, so far it’s working great. With enough tokens it does create a new google doc using the facts from my resume and it somewhat tailors the resume. Although the quality is noticeably worse than doing it manually.

I’ve read these help centre articles:
https://www.make.com/en/help/scenarios/filtering#operators-935231
https://www.make.com/en/help/ai-in-make/openai--dall-e---chatgpt-

Stuck on Parsing Logic

Hi All, I’m sure I seem like a tool for posting this, but I just want to see if you can help me find a solution. I’m trying to mirror this scenario off of YouTube which seeks to achieve the following, but I keep getting this error in the Project - Set Multiple Variables module – message: Failed to map ‘1. value’: Function 'parseNumber finished with error! Valid separators are “,” and “.”. Does anyone have any thoughts? Thanks. RSS Module (RSS feed items): This step watches an RSS feed fro…

https://www.make.com/en/help/tools/text-parser
https://www.make.com/en/help/functions/array-functions#map-complex-array--key--key-for-filtering---possible-values-for-filtering-separated-by-a-comma--

Am I looking in the right direction with these?

Screenshots: scenario setup, module configuration, errors

blueprint.json (68.8 KB)

[Amir_Ashkenazi]

Hey Ryan,
You can accomplish your task with this simple one-module scenario. It uses Airtop to fetch the job description and tailor the resume, completing the process in just 2–3 minutes. To run it, you’ll need a free Airtop API key.

Full disclosure: I’m the CEO of Airtop, an intelligent web automation platform deeply integrated with Make, and I’d love to hear your feedback.

Tailor My Resume.json (9.2 KB)

You can also extend this workflow by fetching job descriptions from a Google Sheet and sending the results via email or saving them to a Google Doc. Let me know if you have any questions!

Best,
Amir

[Ryan_Baillie]

Hey super interesting. You mentioned using ‘Airtop to fetch the job description and tailor the resume’. I reviewed some documentation about it, can you link more specific resources that could solve for this?

I reviewed:

Airtop – Browser Automation for AI Agents

Airtop and LinkedIn Integration | Workflow Automation

and lightly skimmed

What is Airtop? — Airtop | Documentation

I’m a little lost here on the different modules available and how they can be used.

[Ryan_Baillie]

Hey Amir! sorry I overlooked the blueprint you shared at first, I’m pretty new on make. I imported it and this seems awesome!

I’m going to continue playing with it and I’ll let you know how it goes.

[Amir_Ashkenazi]

Yes. Please start with this blueprint
Tailor My Resume.json (9.2 KB)

PageQuery is the only module you need for extraction of data. Other modules allow you to interact with web content, we can leave it for the next use case

[Ryan_Baillie]

Thanks! When do you typically see this response from a URL 'Unable to satisfy the objective".

I have google sheets as a trigger watching for new rows. I add my URL to the spreadsheet in column D.

It outputs my URL into airtop, it seems to load fully, it gets stuck at this position for a little while

then it creates this output

I have the module connected to the google doc module, which is supposed to put the content from airtop into a google doc in my drive. Am I overcomplicating this still?

blueprint.json (10.9 KB)

[Amir_Ashkenazi]

Seems slightly over complicated…
You can check the URL and prompt in the Airtop Studio
If you are looking to get information from sites that require authentication, you will need to create a profile on Airtop and connect it by logging into the service

[DavidGurr_Make]

@Ryan_Baillie I don’t think you need the HTTP Make a request module at all.

Airtop fetches the page itself. In the Page Query module you need to make sure that Mode is set to Automatically create session.

Then map the URL from your sheet into the Url field (same as you did in the HTTP Make a request module).

Then specify your prompt in the prompt field.