GenAI vendor lock in assessment.

Twelve questions across architecture, data gravity, contract, and operations produce a single lock in risk score with a buyer side action list.

Vendor lock in is rarely the result of one decision. It is the cumulative effect of architecture choices, data gravity, contract language, and operational habits. Most enterprises only notice it at renewal, when there is no time and no alternative ready.

This assessment scores your real exposure across the three layers that decide AI vendor lock in. The output is a single score, a tier, and a buyer side action list. It takes about five minutes.

Why a lock in score matters

Lock in is the inverse of leverage

Renewal leverage is whatever your alternative looks like on the day. Lock in is everything that makes the alternative more expensive than the renewal. The score makes the gap visible.

• Architecture lock in. Bindings to vendor specific APIs, SDKs, and prompt formats.
• Data lock in. Embeddings, fine tunes, vector stores, and pipelines that only run with one vendor.
• Contract lock in. Auto renew, single model commits, and missing swap rights.

When to run it

First run at the start of an AI roadmap. Then quarterly during high growth. Always before a renewal or before a major architecture decision.

How the score works

Twelve questions across the three layers. Each question scores zero to three. The total maps to four risk tiers.

The three layers

Four questions cover architecture, four cover data, four cover contract and operations. Each layer is weighted equally in the final score.

Risk tiers

Total score of zero to twelve puts you in green. Thirteen to twenty four is amber. Twenty five to thirty six is red. Above thirty six is critical with structural lock in already in place.

What you get

A summary score, a tier, a layer breakdown, and a personalized buyer side action list with the changes that move the score most for the effort.

Start the assessment

Twelve questions. About five minutes. No email required to see the score, but we recommend sending it to yourself for the action list.

The questionnaire

Pick the answer that best describes your current state. Where it is mixed, pick the one that covers most of your AI footprint.

The score is not a verdict. It is a map. The actions next to your number are what change the conversation at renewal.

What the tiers mean

Each tier has a typical profile and a typical first move.

Green. Low lock in

Architecture is portable. Data is replicable. Contracts are clean. You have real renewal leverage. Keep the discipline as adoption grows.

Amber. Medium lock in

Some vendor specific bindings. Renewal leverage exists but takes work. Fix the cheap contract clauses first. Then plan the architecture changes.

Red. High lock in

Heavy vendor specific architecture. Significant data gravity. Limited contract flexibility. Renewal leverage is low. Start the workbook now, not at notice deadline.

Critical. Structural lock in

Single vendor dependency end to end. No second vendor stood up. Auto renew with no swap rights. The renewal will hurt unless you start the unwind today.

How to move the score

Each layer has a high return action that moves the score quickly.

Architecture actions

Put a thin abstraction layer between your applications and the foundation model API. Even a basic wrapper means a model swap is hours, not weeks.

Data actions

Keep embeddings and fine tune data portable. Store training data on infrastructure you control. Vendor proprietary stores are the hardest lock in to unwind.

Contract actions

Strip auto renew. Add model swap rights. Cap renewal price increases. Build the rate card you compare every future deal against.

Suggested reading

• AI platform contract negotiation framework. Cross vendor reading from the wider library.
• Enterprise AI contract negotiation playbook 2026. Cross vendor reading from the wider library.
• GenAI contract renewal strategy 2026. Cross vendor reading from the wider library.

What to do next

1. Run the twelve question assessment in the next five minutes.
2. Email the result to yourself and your AI council members.
3. Walk through the layer breakdown with architecture and procurement.
4. Pick one action per layer and assign an owner.
5. Re run the assessment in ninety days to track movement.
6. Use the result as the baseline for your next renewal conversation.
7. Contact us for a structured advisor review when you score in the red or critical tier.

Frequently asked questions

Is the assessment really free?

Yes. There is no cost. We use the aggregate data to keep our benchmarks current. Your individual response is never shared.

How long does it take?

About five minutes. Twelve questions, mostly multiple choice.

How accurate is the score?

The score is calibrated against five hundred plus enterprises we have advised. It is a strong directional signal, not a forensic audit.

Can multiple people in our team run it?

Yes. The pattern of scores across roles often surfaces useful disagreement before the renewal lands.

What if our architecture is mixed across vendors?

Pick the answer that covers most of your spend. The score then reflects your dominant vendor exposure.

Does scoring red mean we should switch vendors?

Not necessarily. It usually means you should fix the contract and stand up an alternative, not rip and replace.