[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"guide-what-to-negotiate-in-an-ai-vendor-contract::en":3,"guide-related-what-to-negotiate-in-an-ai-vendor-contract::en":17},{"slug":4,"title":5,"excerpt":6,"body":7,"meta_title":5,"meta_description":8,"keywords":9,"category":15,"published_at":16,"updated_at":16},"what-to-negotiate-in-an-ai-vendor-contract","What to Negotiate in an AI Vendor Contract","Most of the risk in an AI purchase sits in clauses nobody reads: training rights, model changes, spend controls and exit. Each is negotiable before signature and effectively fixed afterwards.","\u003Ch2>Negotiate before you are dependent\u003C\u002Fh2>\n\u003Cp>Every term in a software contract is easiest to change at exactly one moment: before you sign the first one. After that, your data is in the system, your team is trained, your integrations exist, and the leverage that produced the original discount has moved to the other side of the table. Assume the first renewal will be a weaker negotiating position than the first purchase, and put the protections in now.\u003C\u002Fp>\n\u003Cp>That means treating the contract as a design exercise rather than a legal formality. The questions below are commercial and operational, not just legal, so the people who will run the tool should read the terms alongside the lawyers.\u003C\u002Fp>\n\u003Ch2>Say plainly what may be done with your data\u003C\u002Fh2>\n\u003Cp>The most important clause in an AI contract is what the vendor may do with your inputs and outputs. Get a clear statement that your content is not used to train models, or if a training use exists, exactly what it covers and how it is switched off. Ask how long prompts and responses are retained, whether human reviewers see them, and under what circumstances.\u003C\u002Fp>\n\u003Cp>Then follow the data outward. Which subprocessors are involved, where they run, and how you are notified when the list changes. A change of subprocessor can move your data into a different jurisdiction without any change to the product you are using, and the only place that is visible is the contract.\u003C\u002Fp>\n\u003Ch2>Pin the model and the notice you get when it moves\u003C\u002Fh2>\n\u003Cp>An AI product is a moving dependency in a way traditional software is not. The model behind it can be updated, retired or altered in its safety behaviour, and any of those can change your output quality overnight. Ask for version pinning where the vendor offers it, a stated notice period for deprecations, and a defined window to run your own evaluation before a change becomes mandatory.\u003C\u002Fp>\n\u003Cp>If the vendor cannot pin versions — many cannot — then negotiate what you get instead: advance notice, a description of what changed, and the right to terminate without penalty if quality on your own measurements falls below an agreed level.\u003C\u002Fp>\n\u003Ch2>Put a ceiling on the price and on the usage\u003C\u002Fh2>\n\u003Cp>Two different ceilings matter. The first is a cap on renewal increases, expressed against the price you actually pay rather than a list price, applying to more than the first renewal, and covering the total rather than just the unit rate. Without it, an introductory discount is a loan repaid at renewal.\u003C\u002Fp>\n\u003Cp>The second is a control on consumption. For usage-priced products, ask for hard spend caps, budget alerts, per-tenant limits and clear reporting you can see more often than monthly. A product that can generate unlimited billing without an operator action is a commercial risk regardless of how good it is.\u003C\u002Fp>\n\u003Ch2>Make service commitments have consequences\u003C\u002Fh2>\n\u003Cp>An availability commitment with no remedy is a sentence, not an obligation. Look for a stated target, a defined measurement method, and a remedy that scales with the impact — and read the exclusions, because a definition of downtime narrow enough can make a strong-sounding number meaningless.\u003C\u002Fp>\n\u003Ch2>Write down how you leave\u003C\u002Fh2>\n\u003Cp>Negotiate the exit while the vendor is still competing. What data you can export, in what format and structure, how often, and whether you can do it yourself. How long data is retained after termination before deletion. What transition assistance is available and at what rate. Whether you may terminate for convenience, and with how much notice.\u003C\u002Fp>\n\u003Cp>For AI systems, name the pieces a general clause misses: evaluation sets and their labels, prompt and configuration history, conversation logs including tool calls, and any fine-tuned model built from your data. If extraction of something is impossible, it is better to have that stated in writing than discovered during a migration.\u003C\u002Fp>\n\u003Ch2>Keep the agreement alive after signature\u003C\u002Fh2>\n\u003Cp>Record the renewal date, the notice deadline, the cap, the owner and the review date somewhere a person will actually look. A large share of unfavourable software spend is not the result of a bad negotiation but of a good one that nobody revisited, renewing quietly on terms that made sense years ago.\u003C\u002Fp>","Price is the least interesting thing in an AI contract. These are the clauses that decide what happens when the model changes, the bill grows or you want to leave.",[10,11,12,13,14],"ai vendor contract","saas negotiation","data processing agreement","price increase cap","exit plan","buying-guides","2026-09-06T03:30:02+00:00",[18,23,27,31,36,41],{"slug":19,"title":20,"excerpt":21,"updated_at":22},"ai-tool-pricing-models-seat-vs-usage-vs-credits","AI Tool Pricing Models: Seat-Based vs Usage-Based vs Credits","The three common ways AI tools charge — per seat, per usage, and by credits — and how to reason about which one will actually be cheaper for the way your team works.","2026-08-05T14:32:26+00:00",{"slug":24,"title":25,"excerpt":26,"updated_at":22},"how-ai-image-generators-differ-diffusion-vs-the-rest","How AI Image Generators Differ: Diffusion vs the Rest, in Plain Terms","A non-technical explanation of how AI image generators work, why the diffusion approach became dominant, and what practical differences to expect between tools.",{"slug":28,"title":29,"excerpt":30,"updated_at":22},"how-to-automate-your-workflow-without-code","How to Automate Your Workflow Without Code","A practical sequence for building automations that survive: picking the right process, mapping it before touching a tool, and handling the failure cases that break most first attempts.",{"slug":32,"title":33,"excerpt":34,"updated_at":35},"how-to-build-a-chatbot-without-coding","How to Build a Chatbot Without Coding","A practical route to a working chatbot using no-code tools: deciding scope, connecting your own content, handling the questions it cannot answer, and knowing what it will cost.","2026-08-05T14:32:27+00:00",{"slug":37,"title":38,"excerpt":39,"updated_at":40},"how-to-build-an-evaluation-set-that-does-not-go-stale","How to Build an Evaluation Set That Does Not Go Stale","Most evaluation sets are assembled once, from whatever was easy to collect, and then quietly stop representing the work. How to build one from failures you already have, keep it fed from production, protect it from the system it measures, and know when it has expired.","2026-09-09T03:30:02+00:00",{"slug":42,"title":43,"excerpt":44,"updated_at":45},"how-to-change-a-prompt-without-breaking-production","How to Change a Prompt Without Breaking Production","Prompts get edited in a text box and shipped in seconds, which is why they break things quietly: no compiler, no stack trace, no obvious moment of failure. Give them the release discipline code gets.","2026-08-24T03:30:02+00:00"]