How to Start Automating the Right Tasks with AI

Hey there! Welcome back to Founder Mode! This year, I spent more than 10 hours using AI tools to automate my content calendar. In search of the Perfect Day. The tools looked nice and were advanced, but they saved our team no time at all. I still had to schedule meetings, edit documents, and write emails. These tasks could have been automated easily.

I fell into a common founder trap. I automated what was fun, not what mattered. The problem? I aimed to automate fun and interesting tasks, like a really smart calendar. Maybe a throw back to my days building a calendar with Acompli (Peter CMO of Acompli on the podcast this week). However, I should have begun with simple, repetitive tasks that saved me the most time.

Since then, I’ve completely rethought how I automate tasks using AI. The result? I've saved hours each week. Plus, my work output quality has improved.
​
One tool that has been super helpful in this automation journey is Lindy.ai​

The key? I developed a clearer process to figure out which tasks to delegate to AI first. Let me try to explain.

The AI Delegation Matrix

Not all tasks are the same when it comes to AI automation. Here's how I break it down:

1. Time Impact: How much recurring time would you save by automating this task?
2. Quality Impact: Would AI perform this task better, worse, or the same as you?

I made a simple 2x2 matrix by looking at tasks in two ways. It has four quadrants:

Quadrant 1: High Time Savings, Better Quality

AUTOMATE THESE FIRST

These are the gold mines for automation. They save a lot of time while also improving quality.

Examples:

• Health Optimization
• Inbox / research / knowledge management
• Code generation / refactor loops
• Due diligence & M&A
• Customer support triage

Quadrant 2: Low Time Savings, Same Quality

AUTOMATE WHEN CONVENIENT

These tasks may not save much time on their own. A basic automation solution makes it easier to automate tasks.

Examples:

• Document formatting
• Note organization
• Simple data entry
• Social media scheduling
• Meeting notes (I never take notes much)

I use Fireflies.AI to record my meetings. It saves me about 5 minutes each meeting(as I typically don't take notes). However consistency makes it easier for my remote assistants to stay on track.

Quadrant 3: High Time Savings, Lower Quality

AUGMENT, DON’T AUTOMATE

These tasks can save a lot of time. However, fully automating them might lower quality. Use AI as a helper here.

Examples:

• Strategic planning
• Investor communications
• Complex customer issues
• Key hiring decisions
• Product roadmapping

For investor updates, I use AI to create the structure based on our metrics and key events. I spend 30 minutes refining it. This cuts prep time from many hours to 45 minutes. Quality stays high.

Quadrant 4: Low Time Savings, Lower Quality

DON’T AUTOMATE

These tasks don’t save much time, and AI would not perform them well.

Examples:

• Team motivation
• Culture-building activities
• Crisis management
• Delivering difficult feedback
• Core strategic decisions

I tried automating performance reviews. I found that the time saved was small. The results felt generic and impersonal. I’ve moved this into my “don’t automate” category.

Kevin’s 5-Step AI Delegation Process

Once you know which tasks to automate, here’s how I currently use the system:

Step 1: Task Inventory (30 minutes) Make a list of all recurring tasks. Be specific. For instance, "Write weekly team update" is clearer than only saying "communication".

Step 2: Matrix Mapping (20 minutes) Plot each task on the matrix. Focus on time savings and potential quality impact.

Step 3: Rough ROI Calculation (15 minutes)

• Think about how much time you spend each week or month.
• Add up the setup / build time to automate it and test.
• Estimate out the savings you expect.

Step 4: Implementation Plan (45 minutes) Make a plan for each task. List tools, inputs, and timelines.

• Here I try to time box the effort and if I can't do it that is an indication the tooling or my approach isn't right just yet.
• If I fail I keep them on a list to revisit later as tools improve.
• I mentioned Lindy.ai also check out Make.com or a blast from the past Zapier now has lots of AI'ness.

Step 5: Keep Improving (15 minutes weekly) Adjust as needed.

• Track time saved.
• Measure quality impact.
• Some of the automations take maintenance; if, they aren't helping it's fine to kill them.

List of some of the tasks I’ve successfully automated using ChatGPT and AI tools:

Software Engineering & DevOps

• Code generation/refactor loops
• CI/CD and build pipelines
• Automated testing
• Code quality and review
• Infrastructure helpers

Product & Marketing Workflows

• AI-native SaaS builders (MVPs / POCs)
• Advertising & Media Creatives
• Due diligence & M&A Prep

Personal Productivity & EA

• Inbox / research / knowledge management
• Task orchestration
• Writing & communications (more structure vs content)

Aviation-Specific Automation

• Training plans and study buddy for knowledge tests
• Flight ops & certification - keeping up with rules / regulations

Health, Wellness, & Quantified Self

• Data-driven optimization
• Reminders and automated data reporting

Legal / Finance / Admin

• Document analysis
• Hiring & immigration
• Finance reporting and analysis

Real Results: Before and After

Before:

• 55+ hours worked weekly
• 30% of the time spent on low-value tasks
• Constant context switching.
• Frequent bottlenecks when unavailable.

After:

• 35-40 hours worked weekly (saving 10-15 hours)
• 50% reduction in low-value tasks
• The team can move forward without constant input from me.
• Improved consistency in communication.
• Surprising benefit: Team satisfaction has improved!

Common Pitfalls to Avoid

1. Focus on Fun Automation, Not What Matters. Begin with dull, repetitive tasks. Then, tackle the creative and complex ones.
2. Crisp instructions and clear prompts are essential. Give detailed context for better results.
3. Skipping the Human Review Step, AI helps, but we still need to review the output. Never stop testing.
4. Automating Too Much, Too Fast: Start with 2-3 high-ROI tasks. Focus on them first, then expand before you end up with lots of AI helpers generating junk.

What’s one task you could automate this week? Hit reply and let me know—I read every response.

See you next week,

-kevin

Recent Social Posts

Kevin Henrikson @KevinHenrikson I couldn't believe this. A Texas company just spent $100M on de-extinction. Scientists can now bring extinct species back to life. What they discovered about ancient DNA will fundamentally change the future of humankind... 6:30 AM • May 6, 2025 38 Retweets 76 Likes Read 6 replies

​

Kevin Henrikson @kevindegods 90% of NFT projects should shut down today. Most founders won't listen. But those who do will save themselves years of pain and wasted energy. Strategic quitting has become the smartest move in Web3 right now: 7:0 AM • May 6, 2025 34 Retweets 60 Likes Read 10 replies

​

Recent Podcasts

​

Zero-to-One Marketing with P... May 6 · Founder Mode 31:57   ​

Catch up on past emails here.

​​​

​

​

​

​

​

2810 N Church St #87205, Wilmington, DE 19802
​Unsubscribe · Preferences​