Stop having a CRM

The CRM's real problem was never the interface. It was the bundle.

To use the interface, you had to move your data into it. Your customer list, your job history, your photos, your estimates, your correspondence — all of it had to migrate inside the vendor's walls before any of it was usable. Interface and storage came welded together. And once your data was in there, you were not a customer any more. You were a hostage with a monthly payment.

The Bundle is why switching costs are brutal. It is why every migration is a six-week project the whole company dreads. It is why you are still paying for a system half your crew refuses to open.

Understand what The Bundle actually taught two generations of business owners: that the price of using your own information is surrendering custody of it. That you should expect to rent access to the record of your own work. Nobody ever argued for that arrangement out loud. It arrived as a technical constraint and hardened into a business model, and then everyone stopped noticing it — which is the most profitable thing a bad idea can do.

There is no law of physics that says an interface has to own the data it displays. That was an artifact of how software had to be built. It is not any more.

We exist because The Bundle is broken, and we refuse to ship it.

Split the two apart and the picture changes completely.

Storage stays where it already lives. Your documents are in your Drive. Your correspondence is in your mail. Your schedule is in your calendar. Your photos are already on your phone. You are not migrating anything, because it is already where it belongs — in accounts you own, under your control, backed up by companies whose entire business is not losing your files.

The interface becomes language. You do not open an app and navigate to a record. You say what you want.

Here is the objection I would raise if someone pitched this to me, and it is the one the agents-will-replace-everything crowd waves past.

A CRM is not only an interface. Those forms are also a data contract. Every required field and locked dropdown is a schema being enforced on a distracted salesperson at 4:45 on a Friday. The UX is miserable on purpose. It is what makes “show me every job awaiting supplement approval” answerable eleven months later.

Plain English is a better interface and a worse enforcer. Nobody describes the same thing the same way twice. Left alone, conversational input becomes a pile of prose you cannot query — which is worse than the CRM you just cancelled.

So the schema does not disappear. It moves.

The agent holds it. You speak freely; structure gets applied on the way in rather than demanded on the way in. The agent writes back a real record — a row, a document, a calendar entry, a file in a named folder — into your own storage, in a shape it can query later. The discipline the form used to extract from you by force, the agent supplies quietly on your behalf.

That is the difference between a product and a demo. Anyone can bolt a chat box onto a database. The work is in the domain model underneath: knowing what a supplement is, what has to be true before a material order goes out, which four things every job record needs no matter how the conversation went.

Which is why this does not generalize to everything at once. It arrives one industry at a time, wherever someone knows the domain well enough to encode it.

Take a roofing contractor — an industry where the incumbent software is expensive, universally complained about, and impossible to leave.

An aerial measurement report hits the inbox. The agent reads it, extracts the squares, pitch, eaves, ridges and valleys, prices a material order at supplier rates and a labor order at crew rates, drafts the estimate, and files all of it in that job's folder in the contractor's own Drive. The carrier's statement of loss arrives; the agent pulls every line item, compares it against the estimate, and flags what is missing.

Then, once a day, the owner asks one question out loud from the truck: what's outstanding across every open claim?

That question is the entire operating ritual. Not a login. Not a dashboard review. One sentence, said the same way every day, that returns the state of the business.

He never opened an app. He never entered data. And if he decides in a year that he hates the whole arrangement, he keeps every document, every email, every calendar entry — because none of it ever moved.

Per-seat pricing was always pricing for interface access. That is what a seat is: permission for one more human to look at the data through the vendor's window.

When the interface is language, there is no window to sell.

The money moves to the two things that are actually scarce — secure, governed access to a business's own systems, and the domain expertise encoded in what the agent knows how to do. Access and competence. Not seats.

The SaaSpocalypse headlines are treating this as a bloodbath, and for anyone selling windows, it is. But the businesses on the other side of it are not getting a cheaper CRM.

They are getting to stop having one.