This UNRELEASED Make Feature Will Transform Your AI Workflows | Exclusive Demo

Introduction to Human in the Loop Feature

• Sebastian Mertens from Make showcased a new AI feature called "human in the loop" at the No Code Club conference in Berlin.

• The Make community expressed excitement and had questions about how this feature will work after a LinkedIn post was made.

• A walkthrough of the human in the loop feature will be provided, including a real-life use case at 9x.

Setting Up Human in the Loop

• To review content with human input, users need to add a "human in the loop" step to their scenario.

• A dedicated connection must be created for each human in the loop scenario; an example connection is named "Daily AI LinkedIn Post to Approve."

• Users specify what text should be reviewed and can provide additional context by adding prompts.

Running and Reviewing Outputs

• Additional data can be passed through, such as record IDs, during setup for better tracking.

• After running ChatGPT, it generates outputs including an ID and a URL for further actions.

• The generated URL allows team members to approve or reject content without needing a Make account.

Approval Process

• Team members can decline, adjust (e.g., add emojis), or approve suggested posts via the provided link.

• Actions taken after approval require setting up another Make scenario triggered by completion of the review process.

Webhook Configuration

• A webhook is created to connect back to the original human in the loop step for tracking approvals.

• Best practices suggest naming webhooks consistently with their purpose for clarity.

How to Automate Posting on LinkedIn

Automation Scenarios

• Use routers to post directly to LinkedIn if the status is approved.

• Create loops for different statuses: approved, adjusted, or cancelled for retrying actions.

• Once a decision is made, the URL can no longer be used for approval or adjustment.

Staying Updated with AI and Automation

• Subscribe for updates on AI tools and real-life use cases in automation.

• Discuss a real-life implementation of the Human in the Loop feature at 9x.

Managing Inbound Requests

• Using Tally contact form responses to manage inquiries efficiently.

• Existing Make scenario captures Tally responses and stores them in CRM.

Integrating AI Assistance

• ChatGPT assists by drafting email responses based on user inquiries from the contact form.

• Categorize inquiries into four types: training requests, partnership proposals, job applications, and sales offers.

Response Strategies

• Tailor responses based on inquiry type; express interest or politely decline as needed.

• Provide specific formatting instructions for response prompts sent to ChatGPT.

Human Review Process

• Send suggested responses via Slack for human review before sending out emails.

Testing the System

• Example test with an inquiry from "Elon Musk" about CEO position submission.

Reviewing Suggested Responses

• Check Slack message containing inquiry details and suggested response analysis.

Handling Human in the Loop Process

• The speaker discusses managing the human in the loop process step by step, starting with an email response.

• Approval of the message is confirmed; input from the human in the loop includes a full message and its status as approved.

• Delimiters are used to identify parts of the email, allowing extraction of specific information using regex.

Extracting Information for Email Responses

• The output shows only necessary parts of the message extracted from the human in the loop step.

• A router is set up to send emails based on approval status; declined suggestions lead to creating a draft for manual handling.

• An example email sent regarding employment opportunities demonstrates successful extraction and processing.

Enhancing AI Workflows with Human Input