There's a step most people skip when they decide to add an AI email agent.
They download the tool. They connect their inbox. They build the agent. And then they wonder why it's not working the way they hoped… or why their credits are burning through faster than expected.
The step they skipped: cleaning the inbox first.
I learned this the hard way with a client. And once I saw it, I started doing it with every single client before we touch any AI.
The CPA with 1,100 Unread Emails
Amanda is a CPA who runs her own practice. Smart, experienced, busy. And completely buried under her inbox.
When I asked to see it during our first session, she hesitated. “I honestly hate even opening it,” she said.
She had 1,100 unread emails.
But here's what was interesting: when we actually went through the categories together, most of the inbox wasn't real work. It was noise.
Bank statement notifications she'd set up years ago. Payment reminder confirmations. CPE training announcements from three different providers. Emails auto-forwarding from a former employee who'd left two years back. And spam filter notifications… about spam. The system was emailing her to say it had caught spam.
About 80% of what was in there had no business being in her inbox. None of it required a response. None of it needed her attention. It was just there, accumulating.
Why Dirty Inboxes Hurt AI Agents
Here's the part most people don't know until they've deployed their first AI agent: agents charge per email processed.
When a Lindy email manager or any similar AI tool runs, it reads every email that hits your inbox. It processes them, categorizes them, decides what to do with each one. That process costs compute credits. Real money.
Spam costs the same as a real client message. A newsletter you forgot to unsubscribe from costs the same as a proposal request. Former employee auto-forwards cost the same as anything else.
So if your inbox is messy before you add AI, you're not solving the problem. You're making it more expensive.
I call the diagnostic step we did with Amanda a RATs audit: looking for anything that's Redundant, Annoying, or a Time Suck before we build anything. For email, that means going through every category of incoming message and asking one question: should this ever reach my inbox in the first place?
For Amanda, the answer was “no” for most of what was in there.
What the Audit Actually Looks Like
We spent about 60 minutes on this. No AI. No agents. Just screen-sharing her inbox and building email rules together.
Every time we found a category of junk, we'd go into the email settings and create a rule to filter it before it hit the main inbox. Bank statements? Route to a subfolder. Spam filter reports? Delete automatically. CPE training emails? Route to a separate folder she checks once a month.
It wasn't glamorous. It felt like cleaning out a garage.
But 60 minutes in, her inbox was already a fraction of what it was. And more importantly: now when we added the AI agent, it would only be processing emails that actually mattered.
One thing I saw that surprised me: she'd had email folder rules set up for years, but they were actually routing important client emails away from the inbox too. Old rules she'd forgotten about. When we turned on her Lindy agent, it wasn't seeing half her client messages because they were getting filed away automatically before the agent could read them.
So we removed the rules for client providers and kept the rules for true junk. Simple fix. But if we'd skipped the audit, the agent would have seemed broken.
What Happened After
A week after we set up her Lindy email manager, Amanda called me.
“It honestly still feels weird. I go in there and it feels like the Wi-Fi is broken.”
Her inbox was empty. Completely processed. She'd saved 13 hours in seven days. She told me she'd finally done several hours of deep work on tax returns, completely uninterrupted, for the first time in months.
Then she said something that stuck with me: “This gives me hope that I can actually do the things I'm passionate about and not feel burnout.”
Amanda had been thinking about selling her CPA practice before we started working together. Now she's thinking about growing it.
That shift came from a 60-minute inbox audit. Not from a sophisticated AI build.
The Order Matters
The instinct is always to jump to the tool. That's where the excitement is. Building an agent feels like progress. Cleaning up email rules feels like admin.
But the boring first step is what makes the exciting second step actually work.
Think of every email as 5 minutes of your attention. Delete 20 unnecessary emails a day and you've just recovered nearly 2 hours… before a single agent processes anything. Clean the inbox and the agent has less to do and less to cost.
The order is: clean first, build second. If you skip step one, you'll spend a lot of time debugging step two.
Try This Before Your Next AI Setup
Before you build or deploy an email agent, spend 60 minutes doing this:
- Open your inbox and look at the last 100-200 emails
- Identify every category of email that doesn't need your attention
- Create filters or rules to route those categories away (or delete them automatically)
- Unsubscribe from newsletters you haven't opened in 90 days
- Remove any old auto-forwards or rules you set up years ago and forgot about
Then add the AI.
The audit takes one hour. The AI agent builds on that foundation. Get the order right and the whole system will run cleaner, cost less, and actually work the way you want.
If you want to learn more about building AI email agents and other automation workflows, check out the Productivity Academy — or reach out about a one-on-one consulting session.
