A fair question

"Why not just use ChatGPT, or Copilot?"

It is the right question, and it deserves a straight answer. Cloud AI works today and costs little per seat. This page is the comparison we would want if we were the buyer: appliance, cloud, DIY, enterprise suite — including the rows where the cloud wins.

The honest table

Four ways to put AI on your documents

The difference is not the AI model. It is where your documents live, who can reach them, and what you are still paying in year three.

PerimeterCloud AIDIY open-sourceEnterprise suite
Where your data livesYour server roomProvider's regionsYours — once builtVendor cloud / heavy on-prem
Who can reach your documentsYour staff, under your access rulesThe provider, under its terms and its jurisdictionWhoever runs the stackVendor admins, per contract
Time to working30 daysInstant — the cloud wins hereMonths6–12 month rollout
Upfront costHardware + setupNone — the cloud wins here tooEngineering time from day oneA procurement project
Who you need to hireNo oneNo one~3 ML engineersA project team
Cost shapeOne box + annual licence$30–90 per user per month, foreverSalaries, forever$100k+/year floors
Verifiable in personYes — pull the cableNoIf you built it rightRarely
Works without the internetYesNoYes, if fully self-hostedUsually no
Update pathModel updates included in supportThe vendor updates on its scheduleYour engineers, ongoingVendor roadmap, change requests
What happens when you leaveThe box is yours. It keeps workingAccess ends when payment stopsYou keep the stack — and the upkeepAn exit and migration project

Read the cloud column fairly: it wins on instant start and zero upfront cost. If your documents could be public anyway, it is the rational choice. The comparison changes the moment the documents are the business — bid pricing, client files, supplier terms.

The per-seat meter

Cheap per seat. Expensive per company

Cloud AI is priced to look small. Multiply it by your headcount and by the years, and the shape changes.

$30–90

Per user, per month — the going rate for cloud AI assistants and copilots.

List pricing, mid-2026
× 100 seats

A typical mid-market floor. That is $3,000–9,000 every month — $36,000–108,000 every year, every year, and none of it ever becomes yours.

Arithmetic, not a forecast
One box

Perimeter is one box plus an annual licence. The cost does not scale with headcount, and the hardware is yours from day one.

No per-seat meter
Build vs buy

The build-vs-buy line sits around three engineers

DIY stacks on open-source are real, and some are good. They are also an engineering project: a server, ingestion pipelines, retrieval tuning, evaluation, monitoring, model updates — and someone on call when it breaks.

Our honest rule of thumb: if you can dedicate about three ML engineers to it, building can make sense — you get exactly what you design, and it is yours. Below that line, the stack becomes a second job for whoever owns it, and it competes with the work the company actually sells.

Perimeter runs the same class of open-weight models a DIY stack would. The difference is who carries the project: we ship the box working, and support keeps it current.

Put simply

Four options, one sentence each

Cloud AI is a very clever stranger. A DIY stack is a second job. An enterprise suite is a mortgage. Perimeter is an appliance: it arrives, it works, it stays inside your walls.

Cloud AI: a very clever stranger
DIY stack: a second job
Enterprise suite: a mortgage
Perimeter: an appliance
Next step

Take the comparison with you

Get the one-page matrix for your IT and your MD — or skip the paper and see Perimeter answer on your own documents, live.

Live in 30 days · box, setup, training
No per-seat meter · one box, one annual licence
Zero external calls · pull the cable — it keeps working
Honest trade-off · cloud wins on instant start; we win on everything after