Five days before recording this episode, Heather added an AI agent called Viktor to her Slack workspace. Within the hour it had introduced itself to her entire team, worked out on its own that the podcast was central to the business, counted the back catalogue, audited which episodes had transcripts, and offered to fix the gap. It had also spent two thirds of her credit balance, and when she asked where the money had gone, the answer was not what she expected.

That story is the way into a full account of the AI tools Heather is currently paying for: what each one does, what it costs, and whether it earns its place. The total comes to $329 a month, which is just under $4,000 a year, and it is a number she had never added up before preparing this episode.

The second half is a practical method for auditing your own stack. Where duplication hides, what your existing property management software already does that you may be buying twice, why monthly subscriptions beat annual ones until you are certain, and how to hand the whole exercise to AI rather than building another spreadsheet. Heather also covers what is genuinely worth learning right now and what most operators can safely leave alone for another six months.

Key Takeaways

• Most AI spending is not a decision. It arrives as a run of small, individually defensible choices, and it lands on top of a software bill that was already growing.
• When an agent runs on credits, the work you ask for is usually the cheap part. The expense sits in recurring background jobs and proactive activity nobody requested.
• Ask any metered tool, within the first week, where your credits went and what it does when you are not watching.
• A capable agent needs a scope brief before it gets access, not after. Decide what it can see, who can instruct it, and what it may do without asking.
• Per-seat pricing is where software bills quietly grow. A tool that costs $84 a year can cost $420 a year across a team.
• Audit by job, not by tool. Write down the job, then list everything you pay for that does it.
• Check what your existing platforms already do. Most PMS and operations tools have shipped AI features that operators are separately paying for elsewhere.
• Only take an annual plan once you are confident you will still be using the tool in a year. The discount is always offered when you are most enthusiastic and least informed.

‍Tools Mentioned

• Claude, Max plan, $100 a month
• ChatGPT, $20 a month
• Descript, $288 a year
• ClickUp, $84 a year per seat
• Lovable, $50 a month rising to $100
• Wispr Flow, $15 a month
• Littlebird, $20 a month
• Fathom, $15 a month, being cancelled
• NotebookLM, free
• Viktor, from $50 a month

Reading and Listening

• Everyday AI, with Jordan Wilson
• Lenny's Newsletter and Podcast, with Lenny Rachitsky
• Nick Saraev on YouTube
• How To AI, Ruben Hassid
• Sabrina Ramonov

Also Mentioned

• Breezeway, StayFi, Enso Connect, referenced as examples of platforms whose existing AI features are worth checking
• ToolWYZE, in development, launching September

This Episode Is Sponsored By

StayFi

Stay In Touch

This episode is a snapshot rather than a recommendation list, and Heather plans to revisit it every few months. If you run the audit on your own stack, she would like to hear what you found.

‍