Two aspects of founder life make me care a lot about my AI setup. First, there is so much context switching. In one day, I’ll go from working on the product and our technology, to growth, to hiring, to finances.
Second, I work hard to keep administrative tasks from taking over my time so I can focus on the meaningful parts of building the company. No one warns you how much admin comes with being a founder.
I use tools that have AI in them like Fathom, Clay, and Claude Code, but I also spin up purpose-built toolsets that work across these systems.
Here are agents I created in the past week:
Planning My Event Schedule with Google Calendar and Airtable
I wanted to identify opportunities to give technical talks in the coming month. I signed up for a number of events on Lu.ma, which placed them on my calendar, but a calendar view is not a good way to make prioritization decisions or manage outreach.
I created an agent that combines the Google Calendar MCP with the Airtable MCP. It pulls upcoming events into a structured table and adds fields for priority, status, links, and notes. This gives me a ranked view of opportunities and a lightweight system for tracking outreach.
Prompt excerpt:
Find all the AI events on my calendar in the next month and then add them to the AI events table in Airtable. These events might be labeled as meetups or have the word hackathon in the title and definitely no guests.
Call Prep with Mixpanel and Web Search
There are plenty of call-prep tools out there, but I needed something that merged internal context with external research. My toolset pulls user data from our Mixpanel, combines it with public info, and prepares a focused brief before calls.
Deep Dives in Gmail
Email is an unreliable long-term memory. Important details are often buried across threads and attachments, and searching for them manually takes time.
I use an agent connected to Gmail for semantic search and deeper analysis. It's simple, and incredibly valuable.
Prompt excerpt:
How much did we agree to pay our tax person <name>? Did we already send the payment? Which thread had the final version of our agreement?
End-of-Month Expense Reporting with Ramp
Ramp requires adding memos to expenses at the end of every month. Doing it manually is slow. Instead, I use an agent with the Ramp MCP to identify expenses that are missing memos, then look up the vendors, suggest categories, and draft the required memos.
Figma and Slack for Website Updates
When we were iterating on a new website, I wanted an agent to understand the design and then identify Slack messages from our users that would make great testimonials on the site. I used the Figma MCP to provide context on the site's layout and tone, and then used the Slack MCP to search for appropriate feedback from customer channels that could become the quotes we'd publish on the website.
Contacting Candidates with HeyReach + Web Search
When hiring, I use HeyReach to run outbound LinkedIn campaigns across different candidate segments. Once someone accepts a connection request, I need to quickly retrieve the reason I reached out.
I use the HeyReach MCP to identify which campaign they belong to, which tells me why I reached out.
In HeyReach, see which campaign this person is part of www.linkedin.com/in/profile_identifier
I often pair this with lightweight web search to pull in quick context such as their current company's size and recent changes.
HeyReach’s UI also doesn’t have a flexible dashboard so I use the MCP to craft my own queries:
In HeyReach, give me the aggregated statistics for all campaigns that include [Hiring] in the campaign title
My Reflections
Building Tadata informs how I use agents as a founder, and reflecting on how I use agents as a founder informs how we build Tadata. It's fascinating to observe which agents become part of my ongoing arsenal, and which just save me time for one week or one big task.
If you’re a founder, or simply interested in applying agents to real work, I would encourage experimenting with small toolsets. The value compounds quickly, especially when attention is the scarce resource.
