The Software That Doesn't Feel Like Software

Aaron Levie, the CEO of Box, posted something earlier this month that's been stuck in our heads:

I think everyone is substantially underestimating the total demand for software and automation in areas that don't feel like 'software'. Not talking about software that's another app on your phone. Software that just automates things for companies all day long.

That's the whole post. It's also the most accurate description of the small business AI opportunity we've seen written down anywhere.

If you run a small business in Western North Carolina — a property management company, a dental practice, an HVAC outfit, a regional accounting firm — the kind of "software" Levie is describing is not what you currently have. What you have is a mix of generic SaaS tools, a few spreadsheets, some shared inboxes, and a lot of work that gets done because someone in the office knows it has to get done.

The next wave of value isn't another app you have to sign into. It's automation that handles the work nobody currently has time for, and that you never have to look at unless something goes wrong.

What "Software That Doesn't Feel Like Software" Looks Like

The dental office in Hendersonville has a routine: every Friday, the office manager pulls a report of patients overdue for their six-month cleaning, drafts personalized emails to each one, and sends them out. It takes her two hours.

Most people, including most software vendors, would describe this as "needing a CRM." So they sell her a CRM. The CRM has 47 features she doesn't use, requires her to learn a new interface, and the report still takes her an hour because she has to format the output for each patient.

What "software that doesn't feel like software" looks like in this scenario: every Friday at 4 PM, the patient list gets generated automatically, the emails get drafted automatically using the practice's actual voice, and a single notification appears in the manager's existing inbox: "34 reminder emails are ready for review. Approve to send, or click any name to edit." She spends 15 minutes scanning, makes one or two edits, hits approve. Done.

She didn't sign into anything new. There's no app icon. Nobody trained her on a system. The work just got done, and she had final say on what went out.

That's the difference Levie is pointing at. The same task, accomplished differently. One version is software. The other is just the work happening.

Why This Has Been the Underestimated Market

For two decades, the dominant model in software has been another app for another problem. Need to schedule meetings? Calendly. Need to track customer interactions? CRM. Need to manage projects? Asana. Need to handle email? Front. Need to share files? Dropbox. Need to send invoices? Bill.com.

Every one of those is a real product solving a real problem. But for the typical small business, the cumulative effect is that running the business now requires fluency in a dozen different SaaS tools, each with its own interface, login, billing, and learning curve. The promise was that software would make work easier. The reality, for many small businesses, is that they spend more time managing software than they did managing the work itself.

This is the gap Levie is naming. The opportunity isn't to build another tool that requires another login. It's to build automation that operates underneath the existing tools, doing the work that nobody currently has time to do, in formats that already exist, without requiring anyone to change how they work.

We've covered the common failure mode of AI tools that don't fit how a business actually operates before — and most of those failures are the same failure: someone tried to bolt on a new tool instead of automating the work that was already happening.

The Plumber Test

There's a useful test for any AI deployment, especially in a small business: would this be better if my plumber didn't have to know about it?

If the answer is yes, you're in the territory Levie is describing. Software that doesn't feel like software is software that does its job without making the user understand what's happening. The plumber doesn't need to know what an LLM is. The plumber needs the right invoice to go to the right customer, on time, in the format the customer's accounting system can read.

Most AI marketing right now violates this test. It's full of dashboards, configurations, prompt engineering tutorials, and demos of agents talking to each other. That's all interesting if you're a developer. It's noise if you're trying to keep a 4-truck HVAC company running.

The interesting AI deployments we're seeing in 2026 are the ones where the AI is essentially invisible. The customer-facing employee gets the right information at the right moment. The owner gets a notification only when something needs a human decision. The reports show up in the inbox they always show up in. The work gets done. Nobody opens an "AI tool."

This is also why the comparison between Zapier, n8n, and Make matters less than it used to. Those tools made automation accessible, but they still produced "software" — workflows you had to maintain, dashboards you had to monitor, error handling you had to figure out. The next layer is automation that runs without needing to be tended.

What Levie Is Telling Investors, Translated for Operators

The reason Levie's post matters more than it might seem is who he's talking to. As the CEO of Box, his audience is enterprise software investors and executives. When he says the demand is "substantially underestimated," he's signaling where the next decade of enterprise software value is going.

The translation for a small business owner: the tools that will matter most for your business in 2027 are not on the market yet, and the ones that are on the market and being marketed loudly are mostly competing for a smaller slice of attention than they think. The real value is in the unglamorous work: the back-office automation, the workflow under the workflow, the work that nobody talks about because nobody finds it interesting.

This has practical implications for how you should evaluate AI vendors right now.

Be suspicious of demos. A demo that requires you to learn a new interface to see the value is selling the interface, not the value. The real test of a small business AI tool is what your day looks like in three months when you've stopped thinking about it.

Ask about output, not features. Vendors pitch features because features are easy to put on a slide. The right question is "what shows up in my inbox / on my desk / in my CRM after this is running for a month?" If they can't answer in plain language, that's information.

Watch for tools that don't require their own dashboard. The best AI deployments for small businesses are the ones that integrate with the systems you already use — your existing email, your existing scheduling, your existing accounting. A new dashboard is a new system you have to learn. A modification to your existing email flow is just better email.

The Plumber in Asheville Test, Applied

If you run a small business and you're trying to decide what to do about AI, here's the practical version of Levie's thesis:

1. List the recurring work that currently gets done because someone in your office remembers to do it. Not the work that's documented in a system. The work that lives in someone's head — Friday reports, monthly reminders, the seasonal communications, the follow-ups.

2. Pick the one that takes the most time and is the most boring. That's your starting point. Not the most complex thing. Not the most strategic thing. The most tedious thing.

3. Ask: what would have to be true for this to happen automatically, in the formats my team already uses, without anyone having to learn a new tool? That's the brief. That's the AI deployment worth doing.

The companies that get this right are not going to be the ones that announce loudly that they're "AI-powered." They're going to be the ones that quietly start making more money with the same headcount, because the work that fell through the cracks now just gets done. From the outside, nothing about the business will look different. From the inside, everything will be.

---

Blue Octopus Technology builds back-office automation for small businesses — the kind that doesn't require a new login, a new dashboard, or a new vocabulary. If you've got recurring work that's quietly costing you money, let's talk.