Before you press play, here’s what you’ll see in this session:
👇 This training only takes about 10 minutes. And while you’re watching, listen for the moment where I explain why “aggregated data” can stillinfluence an AI model’s behavior.
That point changes how many attorneys read AI data.
In traditional SaaS agreements, vendors often include clauses allowing the use of aggregated or anonymized data to improve the service.
That approach generally works because SaaS systems use data for analytics and operational improvements.
AI systems operate differently.
When data is used for training, the model’s internal parameters are updated based on patterns in that data.
Even if the original text is no longer stored in readable form, those patterns can influence how the model behaves.
This is why clauses written for SaaS environments do not always translate safely into AI systems.
Not because they’re complex, but because most attorneys were never taught when an indemnity actually makes sense.
So negotiations drift into instinct, precedent, or “market practice” arguments.
In this video, I introduce the one question that unlocks indemnities:
.avif)
.avif)
In many AI-enabled products, the SaaS vendor is not operating the underlying model.
Instead, the vendor integrates a third-party AI provider.
This creates multiple layers:
🗂️ Customer data layer
💻 SaaS vendor platform layer
🧠 AI model layer
Different types of data usage and training permissions can exist at both the SaaS vendor layer and the AI model layer, which means risk allocation can involve multiple parties.
Most attorneys and contract managers negotiating AI provisions understand that the technology matters.
But they often lack visibility into the underlying architecture🏗️.
Without that context, contract drafting becomes abstract and negotiations become harder than they need to be.
I explain all of this in the video, so check it out.
Over the past year I kept seeing the same pattern. Very capable SaaS lawyers were being asked to negotiate AI provisions, but the contracts referenced issues that simply don’t exist in their traditional understanding of technology agreements.
So I created the Practical AI Contract MasterClass to close that gap.
Over 3 weeks we walk through the issues that actually appear in AI contract negotiations.
You’ll learn:
•🏗️ the architecture behind AI-enabled products
•📑 how to interpret AI addenda and vendor AI policies
• ⚖️ where indemnities collapse in layered AI systems
• 🧠 how training permissions actually work
• ✍️ how to draft precise data-usage restrictionsThe focus is practical.
✔ Real clauses
✔ Real negotiations
✔ Real drafting strategies
.avif)