3 Ways I Use AI to Save 20+ Hours Every Week
I have three jobs (just like most New Yorkers). I'm a content creator, a public speaker, and a tech engineer all at the same time. I'm also technical and creative enough to build my own tools, write scripts, post content, and connect APIs. What I do not have is time. There were two times where I almost missed my mom's birthday because I was "grinding" at two startups that no longer exist. That was the moment I decided I needed to start building systems, workflows, and incomes that work regardless of external factors.
That is the whole point of this article. The constraint is not ability. The constraint is hours in a day. So I started building AI workflows because I needed systems that could run without me, doing the repetitive and time-consuming work in the background while I stay focused on the things that actually require my presence.
These are the three workflows that have given me back 20+ hours every week, and my exact blueprint how you can build them too ⬇️
1️⃣ I Built a Custom Instagram Analytics MCP Using Claude and Cloudflare
<div class="frich-tip">Time saved: 5+ hours per week on analytics, strategy, and KPI reviews.</div>
The Problem
There is no native integration between Instagram and AI tools. Meta does not hand your data over to Claude or any other AI assistant out of the box. So every week I was logging into Meta Business Suite, cross-referencing it with third-party analytics tools, and manually pulling numbers to review my strategy. It was repetitive, slow, and not the best use of my time.
So I built my own integration from scratch.
The Solution: A Custom MCP Server Connected to the Instagram Graph API
MCP stands for Model Context Protocol. It is the way Claude connects to external data sources and tools. The concept is straightforward: if you give Claude a server it can communicate with, it can pull real data and take real actions on your behalf. I built that server myself and named it MCP IG.
Step-by-Step: How to Build Your Own IG MCP
Step 1: Get API Access Through Meta Business Developer Tools
You need official credentials before you can access any Instagram data. Here is the process:
- Go to developers.facebook.com and create a developer account.
- Create a new app and select Business as the app type.
- Connect your Instagram Business or Creator account to the app.
- Add the Instagram Graph API product.
- Request the permissions you need. At minimum you want instagram_basic and instagram_manage_insights to access your analytics data.
- Generate a long-lived access token. This is the credential your script will use to authenticate against Instagram servers.
Take your time getting the right permissions approved. That step is what unlocks the specific data you want to pull.
Step 2: Identify the API Endpoints You Need
The Instagram Graph API gives you GET request access to a wide range of data. The endpoints I rely on most are:
- Follower growth over time
- Reach and impressions by post
- Saves and shares by post
- Content breakdown by type, including Reels, carousels, and static posts
- Best posting times based on audience activity patterns
- Hashtag performance
Each of these is a separate GET request. Once you know which endpoints map to the data you care about, you can write the script around them.
Step 3: Write a Script That Calls Those Endpoints
I wrote a script that takes my access token, hits the relevant Instagram API endpoints, and formats the responses into something readable and actionable. The script is stateless, meaning it runs on demand and returns fresh data every time it is called. Every GET request visible in the MCP IG screenshot is a tool Claude can call against my account in real time.
Step 4: Host the Script on Cloudflare Workers
I deployed the script to Cloudflare Workers, a serverless platform that runs code without requiring you to manage any infrastructure. Cloudflare Workers are fast, reliable, and free for low-volume personal use cases like this one. Once deployed, your Worker URL becomes the endpoint Claude calls every time you ask it for Instagram data.
Step 5: Connect MCP IG to Claude
Inside Claude, navigate to Settings, then Customize, then Connectors, and select Add Custom Connector. Paste in your Cloudflare Worker URL and Claude immediately has access to every tool your MCP exposes.
Once that connection is live, I can open Claude and ask for a breakdown of how my content performed over the past 30 days. Claude calls my API in real time and comes back with a full strategic summary. There are no spreadsheets to pull, no dashboards to switch between, and no manual work involved.
Five hours a week I used to spend gathering and interpreting data is now a single conversation with Claude.
2️⃣ I Use Descript to Edit Videos
<div class="frich-tip">Time saved: 10+ hours per week on editing and content repurposing.</div>
The Problem
Editing is the most time-intensive part of content creation for me. I would record strong material and then spend hours cutting it down, removing filler words, adding captions, and exporting separate versions for different platforms. That process was consuming an unreasonable portion of my week.
The Solution: Descript
Descript operates on a completely different premise than traditional video editors. Instead of working on a timeline, you edit a transcript. The video and the words are treated as the same thing, so editing one automatically edits the other.
Step-by-Step: My Descript Workflow
Step 1: Upload Your Recording
Drop your file directly into Descript. It handles video, audio, screen recordings, and Zoom imports. Whatever format you record in, Descript can process it.
Step 2: Let It Transcribe
Descript transcribes your recording automatically in a matter of minutes. The accuracy is strong and significantly better than most auto-caption tools I have used. Any errors are fixed by simply retyping the correct word directly in the transcript.
Step 3: Edit the Transcript Like a Document
Read through your transcript and make edits the way you would edit any written document:
- Remove filler words using the built-in AI feature that finds and removes every instance of 'um, uh, like, and you know' across the entire recording in one click.
- Cut sections by highlighting the text and deleting it. The corresponding video footage is removed automatically.
- Rearrange sections by cutting and pasting transcript text. The video follows every move you make.
There is no scrubbing through footage and no frame-by-frame cutting required. Everything happens in the document view.
Step 4: Add Captions
Captions are generated directly from your transcript. You can style them, position them, and correct any errors, all from the same interface. There is no need to export to a separate caption tool or wait on a transcription service.
Step 5: Export for Every Platform
From a single edit I export a landscape version for YouTube, a vertical cut for Reels and TikTok, and a square version for LinkedIn. Descript handles the reformatting so I never have to re-edit the same content multiple times.
Descript replaced three separate tools for me: a video editor, a caption generator, and a transcription service.
The Repurposing Layer: Descript and Claude Together
Once Descript gives me the full transcript of a recording, I paste it into Claude and ask it to turn that content into an Instagram caption, a tweet thread, a newsletter section, a YouTube description, and three short-form video hooks. One recording becomes a full week of content across every platform. The original thinking happened once. The distribution happens everywhere.
3️⃣ I Use Claude and Motion to Audit and Protect My Time
<div class="frich-tip">Time saved: 8+ hours per week by cutting wasted time and reducing decision fatigue.</div>
The Problem
Working three jobs means every hour has to be accounted for. For a long time I was filling my calendar with social commitments, reactive tasks, and meetings that felt productive in the moment, but were not meaningfully moving anything forward. I was busy all the time without building towards what mattered.
I needed an honest look at where my time was actually going, and a system to protect the hours that counted most.
The Solution: Time Blocking, Claude Audits, and Motion
Step 1: Time Block Every Activity, Not Just Meetings
I put everything in my calendar as a time block. Deep work, content creation, admin tasks, workouts, social commitments, and creative time all get their own blocks. If something is going to take time, it gets scheduled.
Most people only calendar their meetings and leave the rest of the day unstructured. That approach makes it easy to end up reactive rather than intentional. When your whole day is blocked, you are forced to be honest about what actually makes it onto your schedule.
Step 2: Run a Weekly Audit With Claude
At the end of each week I describe my time blocks to Claude, including what I spent time on, roughly how many hours each category took, and how productive each area actually felt.
Then I ask Claude to audit it. I want to know where I am over-investing, what is taking longer than it should, what patterns are emerging across weeks, and what could be automated, batched, or removed entirely. Claude gives me a prioritized breakdown I can act on right away, and because it is based on my actual calendar data rather than generic advice, the feedback is specific and useful.
Claude surfaced that I was spending more than half my time on social commitments. That one insight was enough to make me start saying no more intentionally.
Step 3: Use Motion for Automatic Rescheduling
Motion is an AI calendar tool that automatically reorganizes your schedule when things shift. When a meeting moves, when something runs long, or when a new priority comes in, Motion adjusts everything else so your day still holds together without you having to manually rebuild it.
Before I started using Motion, any disruption to my calendar meant spending time replanning the rest of the day. Now Motion handles all of that in the background and I focus on executing whatever is in front of me.
Step 4: Protect the Blocks That Matter Most
Based on what Claude flags each week, I identify where I need to invest more time and what needs to be reduced or removed. The highest-priority blocks get treated the same way I treat external commitments. They do not get moved, cancelled, or shrunk to make room for things that are less important.
The goal is not to fill every hour with productive output. The goal is to make sure that how I spend my time actually reflects what I am trying to build.
I built these workflows because I am technical and I was running out of time. The ability to build things was there. The hours to manually do everything were not. These three systems now run in the background of my week every single day, and they have returned more usable time to me than anything else I have tried.
The setup took real work upfront. That is also why it works, because most people will not put in the time to build something like this. If you do, you end up with a significant operational advantage that compounds over time.
Want to stay updated every time I drop a new AI tool or workflow?
I send a newsletter whenever I find something worth sharing, whether that is a new integration, a workflow I just built, or a tool that actually made a dent in my week. Sign up so you are the first to know when the next one drops.
