I spent last Thursday in a conversation that ended with one of the most candid sentences I've ever heard from a CFO. "Look," she said, "we don't have the time or the budget to build real software. We just need to survive until this whole AI thing settles down."
I didn't have the heart to tell her the truth, which is: it's not going to settle down. Not in a way she'd like, anyway.
The distinction that no longer exists
There used to be a clean line between "software companies" and "everyone else." Software companies wrote code; everyone else bought it. Software companies hired engineers; everyone else hired people who used the tools the engineers built.
That line is gone. It was already fraying before 2023. What happened in the last two years quietly ended it.
Today, every one of your competitors is in the process of deciding how deeply software will run their operations. And the ones who decide "very deeply" are going to operate at a cost base, a response latency, and a customer-experience quality that the ones who decide "not much" physically cannot match.
This isn't about automation. Automation gets you efficiency. Operating intelligence gets you a different company.
What "operating intelligence" actually means
Operating intelligence is the capacity of your business to sense, decide, and act through software — without a human in the loop for every step. It shows up in small, boring-sounding places:
- Your sales team's quoting tool notices a prospect asking about a use case you haven't productized yet, and writes a one-pager scoped to their environment before the next call.
- Your ops manager opens their laptop Monday morning and finds that weekend anomalies in supplier data have already been investigated, summarized, and — where appropriate — resolved.
- Your legal team's intake form doesn't just collect information; it drafts the first version of the document, flags the three clauses that matter, and schedules the review with the right partner.
None of these are revolutionary in isolation. Cumulatively, they're a different company. And the gap between the company that has them and the company that doesn't is growing exponentially — because the first company's software is also learning, every day, from every transaction.
Why the old playbook doesn't work
The instinct for most 50-to-500-person companies is to hire a "digital transformation" consultancy, commission a five-year roadmap, and nothing meaningful changes. This path fails for three structural reasons:
1. The software is the strategy now
You can't outsource the thinking about what your intelligence should do and keep a differentiated business. The specific ways your company makes decisions — the stuff your best operators know by heart — are the thing you're trying to encode. That's a product problem, not a contractor problem.
2. You don't need an army, you need leverage
AI-accelerated development teams of three or four people regularly ship what used to take twenty. A small team with deep platform expertise (Google Workspace, Microsoft 365, Claude, Gemini) can ship more real software in one quarter than a traditional SI ships in a year. The shape of the engagement has to match the new economics.
3. Adoption beats features
A system your people actually use every day, built inside the tools they already work in, beats a "perfect" platform rollout that ends in a Slack channel called #launch-issues. This is why we build on top of Workspace and Microsoft 365 — we meet your team where they are.
The "software company" identity shift
Every company will eventually decide, explicitly or implicitly, whether it sees itself as a software-producing organization. That decision shapes hiring, capital allocation, product roadmap, and — most of all — what your best people are expected to do with their time.
The companies making that decision explicitly are investing in:
- A small core team of AI-native software builders who own the internal stack.
- Operating systems on top of Claude, Gemini, and custom MCPs that their people extend week by week.
- A culture that treats "we could build software for this" as a normal first response to a new problem.
The companies not making the decision are, in practice, still making it. They're deciding to let a competitor operate faster, at lower cost, with better customer experiences, and with more room to experiment. That decision compounds.
What "seriously" looks like
If you're persuaded, the question becomes: what does taking this seriously look like in the next 90 days? Our version of the answer, refined across dozens of engagements:
- Pick one workflow that matters. Not the flashiest — the one your best operator would describe as "the thing I'd automate if I could." That tells you where the leverage is.
- Ship a real pilot in six weeks. Not a mockup, not a proof-of-concept slide deck. A system that does the work end-to-end for a defined group of users, embedded in the tools they already use.
- Measure. Time saved, errors avoided, throughput gained. Whatever number would have made the original workflow better if you'd doubled it.
- Then expand. The trust earned from shipping one thing makes the next five easier to scope, fund, and deliver.
There's no five-year roadmap in that list. There's a 90-day one. The five-year version is just the 90-day one, six times, compounding.
The quiet part out loud
The CFO I opened with was right about one thing: she doesn't have unlimited time or budget. Nobody does. But the assumption embedded in her framing — that "this AI thing" is temporary, and that waiting is a neutral move — isn't neutral. Waiting is picking a side. It just feels comfortable because you're not doing anything yet.
Every company is a software company. It's just that most don't know it yet. The ones that figure it out first will own the next decade of their industries.
If you want to be one of those — we should talk.