Most of this series has been about the fear that stalls custom AI deals: what happens when the builder is gone. We have walked the three engagement models, the documentation that makes them work, and the support SLA that holds them up. This last post turns the whole thing around.
The point of all that structure is not defense. It is to give you something most vendors will never offer, because they are too afraid of it to put it in writing. The ability to fire them cleanly. And the willingness to hand you that ability is the single most reliable signal that you are dealing with a vendor worth keeping.
Lock in is what insecure vendors sell
There is an old playbook in enterprise software. Make the system hard to leave. Bury the data in a proprietary format, keep the integrations undocumented, price the exit so high that staying is always cheaper than going. The dependency is not a side effect of the product. It is the product. Retention through inability to leave.
It works, in the narrow sense that customers stay. It also tells you everything about the vendor's confidence in their own work. A company that retains customers by trapping them is a company that does not believe it could retain them on merit. The lock in is a confession.
A vendor confident in the work hands you the keys without flinching. The relationship continues because the work is good, not because leaving is impossible. When the exit is easy and the client stays anyway, both sides know the partnership is real. That is worth more than any lock in clause.
Ownership transfer is a feature, not a failure
The instinct is to treat a client wanting to own the system outright as a relationship breaking down. Flip it. The option to take full ownership is a feature of the product, the same way a warranty or a money back guarantee is a feature. It does not signal that things went wrong. It signals that the client was never at risk in the first place.
A defined ownership path delivers the source code, complete documentation, and a structured knowledge transfer window so your team can run the system independently. You can take it non exclusive, keeping your cost lower while the generalized solution stays in the vendor library, or exclusive, removing that specific build from reuse entirely. Either way, the exit was built before you needed it, which is the only time building an exit is worth anything. An exit you negotiate during a crisis is not an exit. It is a hostage release.
Optionality is what actually closes the deal
Here is the part that surprises vendors who have never tried it. Offering the clean exit does not lose deals. It closes them.
Put yourself on the buyer's side of the table. Two vendors offer similar systems at a similar price. One answers the continuity question with reassurance and a contract you cannot leave. The other hands you a documented system, a severity based SLA owed by a team, source escrow with defined triggers, and a written ownership path you control. Which one is the safer signature. The optionality is not a concession that weakens the deal. It is the thing that makes the deal safe enough to sign.
When the client controls the exit, the dependency that felt like a risk becomes a choice. They are not stuck with you. They are choosing you, every month, because the alternative is available and they keep deciding not to take it. That is a far stronger position for both sides than a contract that holds someone in place against their preference.
The reframe, in one sentence
You should be able to fire your AI vendor cleanly, and a vendor worth hiring is one who built that exit before you ever asked for it. The clean exit is not the weakest point of the engagement. It is the proof that the rest of it is honest.
This is the whole reason Heed published a Support and Continuity policy with the sample contract clauses in the open. You are never locked in. You are never stranded. Three doors, all open: we run it, escrow protects it, or you own it outright. Bring the question to the table and we will show you the support model, the escrow option, and the ownership path before you commit to anything.