There's a line I keep coming back to.

I had a conversation the other day that perfectly captured the exact problem we're seeing across the board: "Traditionally, 100% of your time goes into entering and reconciling the data. Zero per cent goes into actually using it."

That's most businesses right now. The tools are good. The data is there. Nobody's reading it.

That's the thing that's changed.

In 2024, Reddit signed a deal with Google. Not for advertising. Not for traffic. For their data.

$60M a year — just to access what Reddit's users had been posting for 18 years. Their stock shot up. Wall Street suddenly saw what the AI companies had already figured out: data is the asset.

Your business has been building the same thing.

Every invoice you've processed. Every customer interaction logged in your CRM. Every inventory movement, support ticket, and sales order. Years of it, sitting in your tools, doing nothing.

The difference between Reddit and most SMBs isn't the quality of the data. It's knowing what to do with it.

I was recently at the Sunrise Tech Conference, and Cliff, the founder of Canva, summed it up perfectly.

Here's where most businesses are today.

Xero knows your finances. Your CRM knows your customers. Your inventory system knows your stock. Excel knows whatever your ops manager put in there last Tuesday.

Each tool does its job. None of them talk to each other. And nobody has time to pull it all together.

So what happens? You log into Xero to check cash flow. You log into HubSpot to check pipeline. You open a spreadsheet to check costs. You make a decision based on three separate windows and a gut feeling.

That's not analysis. That's just vibes with receipts attached.

The command centre

What we're building for clients at Phlo is connecting all of those sources into one place, a command centre where you can ask questions across your entire business in plain English.

Not just a dashboard that shows you what happened. Something that tells you why it happened, where it's happening again, and what to do about it.

The tools you're already using, your ERP, CRM, accounting software, ops data, they don't change. You just stop looking at them in isolation.

Big companies figured this out a long time ago.

They'd bring in McKinsey or BCG, hand over their data, and wait six weeks for a deck. The deck would tell them where they were bleeding money, which customers were actually profitable, and where the hidden growth was. That engagement cost $500K.

Most SMBs never got one.

Here's the thing — that work wasn't magic. It was pattern recognition across large amounts of data. And that's exactly what AI is very, very good at.

Claude Opus — the model we use for this kind of analysis — was specifically built for financial analysis, due diligence, and research synthesis. It can hold your full year of transactions, your CRM history, your cost data, and your inventory movements in a single session and reason across all of it at once. Anthropic designed it for exactly this: complex, multi-source analysis in plain language.

What does it actually find? The kinds of things nobody has time to look for:

A product line that looks profitable on paper — until you factor in the returns eating 40% of the margin.

A customer segment buying less often but at 3x the average order value, completely ignored by the sales team.

A cash flow gap that shows up every September and has for three years, but nobody's ever planned for it.

Suppliers being overpaid because nobody's compared invoices against contract rates in the last 18 months.

None of this is exotic. It's in the data. It's just never been looked at.

And 91% of SMBs that are using AI this way say it directly boosts their revenue. That stat doesn't surprise me.

Note: the data in this demo has been masked and changed from the live source — It's illustrative, not a live readout of their business.

Plantman builds and rents water trucks across WA. One ERP, one CRM, two sides of the business — never looked at together. We connected them and let the AI read the whole thing.

Here's the kind of thing it surfaces:

What didn't work first: live API connections pulled through occasional data anomalies. Clean exports worked better — We would self-correct those errors on export anyway.