Everyone Has the Seat Now

Overall adoption rose to 55.86% of businesses on Ramp, up 0.62 points from a revised August (Ramp AI Index, September 2026).

But two figures in the release sat uneasily next to each other.

In Ramp's mid-market segment, the median company spent $4.56 per employee per month on AI in September (Ramp AI Index dashboard, September 2026). Among the top 1% of all firms on Ramp, the median was $9,568 per employee per month (Ramp AI Index writeup, Oct 14, 2026, September data).

Both groups can buy the same tools from the same vendors. Access is not what separates them.

Ramp's lead economist, Ara Kharazian, summed up the moment in one line: "It feels like we're at the top of the S-curve, where the growth in the market will have to come from more intense adoption rather than more customers" (Ramp writeup, Oct 14, 2026).

For owners and operators of traditional American businesses with 50 to 500 people, that sentence changes the question. Our read: adoption is no longer the bottleneck. Intensity is.

Ramp sorts the businesses on its platform into three size bands. In September, mid-sized businesses on Ramp reached 62.50% adoption, up 1.13 points, the fastest gain of any band (Ramp AI Index, September 2026). Small businesses rose 0.53 points to 50.03%. Large businesses rose 0.50 points to 65.72%.

A year earlier, in September 2025, the middle band stood at 47.67% (Ramp AI Index, September 2025, as published in October 2026).

So in Ramp's data, the companies most like the ones we work with are not laggards on the buying side. More than three in five now pay for at least one AI tool.

Buying is not using. Ramp's spend panel shows how thin the typical commitment still is. In September, the median monthly AI spend per employee was $29 in Ramp's micro-SMB segment, $11.91 in its SMB segment and $4.56 in its mid-market segment (Ramp AI Index dashboard, September 2026).

The bigger the company, the thinner the spend per head. Our read: many larger firms buy seats for a few people and spread the cost across a large headcount. The daily work of most employees has not changed.

Then there is the top. Ramp's writeup repeats an earlier finding: 80% of spend on OpenAI and Anthropic comes from 1% of firms (Ramp writeup, Oct 14, 2026).

Our read: a seat at a few dollars per person does not move a number on the income statement. Removing a named bottleneck does. A quote that takes three days. A handoff that waits on one person. A renewal nobody touches until it is late.

Among finance and insurance firms on Ramp, adoption reached 73.69% in September, up 0.57 points (Ramp AI Index, September 2026). Only the information sector, which Ramp labels technology and media, was higher, at 79.92%.

The sector also spends more per head than most. Its median was $50.98 per employee per month, against $5.80 in construction and $3.45 in health care (Ramp AI Index dashboard, September 2026).

For an agency, an MGA or a regional carrier, the competitive question has moved. In Ramp's data, most of your peers already pay for the tools. Our read: the edge now shows up in who still rekeys the same submission three times, and who does not.

Among North Carolina businesses on Ramp, adoption stood at 52.75% in September, 16th of 51 (the 50 states plus DC), up one spot from 17th (Ramp AI Index, September 2026; ranks computed by Setanta, since Ramp publishes none). It rose 0.45 points from a revised August, a bit slower than the 0.62 gain across all businesses on Ramp.

The year tells more. In September 2025, North Carolina sat at 36.10%. That is a gain of 16.65 points in twelve months (Ramp AI Index, September 2025 and September 2026). Ramp's September sample counts 1,909 North Carolina firms, and 1,007 of them paid for AI.

Florida ranked 9th at 55.10%. Tennessee was 21st at 51.45%, Alabama 22nd at 51.12%, Georgia 24th at 50.39% and South Carolina 40th at 45.92% (Ramp AI Index, September 2026). All six states sit below the 55.86% rate across all businesses on Ramp.

Our read: North Carolina is drifting up with the rest of Ramp's board, not pulling away from it. The story here is no longer whether firms adopt. It is what those 1,007 paying firms do with it.

Anthropic was paid by 43.77% of businesses on Ramp in September, up 0.24 points. OpenAI reached 40.63%, up 1.12 points, the biggest monthly gain among the large labs (Ramp AI Index, September 2026). The gap narrowed to about three points.

A year earlier the picture was reversed. In September 2025, OpenAI stood at 41.02% and Anthropic at 16.02%. In the series as Ramp now publishes it, Anthropic first passed OpenAI in May 2026 (Ramp AI Index, Sep 2025 to Sep 2026).

The two shares together add up to far more than overall adoption, so many firms pay for both.

Our read: picking the model is the easier decision. Picking the bottleneck it removes is the one that pays.

Our mission is to modernize traditional American businesses. These are the firms that insure homes, build roads, move freight and keep the lights on. Most of them are owner-led, and in Ramp's data most firms like them have now bought something.

For most of these companies on Ramp, the buying decision is behind them. The harder work is ahead: fitting AI to how a real business runs, with its own people, rules and systems. The firms that pull ahead will pick one or two real workflows, put someone close to the work, and count bottlenecks removed rather than licenses bought.

Our aim is simple. Setanta makes your whole company as good as your best person.

If you want a plain look at where work waits inside your own company, our Operating Diagnostic is free.