A custom Playbook captures your organisation's actual positions: what you require in a contract, what you'll accept as a fallback, and what needs to be escalated. There are two ways to build one. You can generate a first draft from a contract you already trust, or you can write the rules yourself from scratch. Either way, the result is a Playbook that reflects your standards rather than a generic starting point.

Option 1: Generate a Playbook from a contract you trust

If you have a contract that reflects your preferred positions, whether a template you've drafted, a heavily negotiated agreement, or a signed deal you're happy with, you can use it as the source for your Playbook.

1. Open the contract in Word and launch the SimpleAI sidebar
2. Navigate to Playbooks and select Create Custom Playbook
3. Choose Generate from open document
4. The AI reads the contract, identifies the legal principle behind each provision, and creates a rule for each one
5. Review the generated rules, adjust any that don't quite capture your position, remove any that aren't relevant, and add anything the AI missed
6. Name and save the Playbook

The AI doesn't copy the contract language into the rules. It extracts the underlying principle — mutual confidentiality, liability capped at 12 months' fees, 30-day termination notice — and turns that into an instruction it can apply to any contract of the same type.

💡 This is the fastest route to a Playbook that reflects your actual positions. The AI does the heavy lifting of identifying your principles. Your job is to review and refine, not to write from nothing.

Option 2: Build rules from scratch

If you'd rather define your rules directly, you can build a Playbook rule by rule without uploading a source document.

1. Navigate to Playbooks and select Create Custom Playbook
2. Choose Build from scratch
3. Add rules one at a time

For each rule, you define what the AI should check and what it should do if the contract doesn't meet that standard.

A well-structured rule follows this pattern:

Ensure that [your required position]. If not, redline to [your preferred language].

For example:

"Ensure that the initial term is two years. If it is more or less than two years, redline to two years."

"Ensure that the liability cap is no more than 12 months' fees. If not, redline accordingly."

"Ensure that confidentiality obligations survive termination for at least two years. If not, redline to add survivorship language."

You don't need to write the instruction in any particular format. Write it in plain English and use the Improve Instruction button. This takes whatever you've written and turns it into a prompt the AI can apply reliably, without you needing to think about prompt engineering.

Adding fallback positions

Most negotiations involve compromise. Fallbacks let you build that flexibility into the Playbook rather than leaving it to individual judgment.

For each rule, you can add one or more fallback positions. The AI applies your preferred position first. If that position isn't met, it surfaces the fallbacks so you can choose which one to apply given the context of the deal.

For example:

• Preferred position: liability capped at 12 months' fees
• Fallback 1: liability capped at 6 months' fees
• Fallback 2: liability capped at the greater of 6 months' fees or £50,000

Fallbacks give whoever is running the review a clear set of options rather than having to make a judgment call with no guidance. The decision still sits with the reviewer, but the Playbook tells them what acceptable looks like.

Adding comments

When a rule triggers a redline or a red flag, you can include a comment that appears in the margin of the document alongside the change. Comments are visible to the counterparty or to whoever is reviewing the document.

Use comments to explain the rationale for a change in a way that's constructive rather than adversarial.

"This clause has been revised to reflect our standard liability position, which we apply consistently across all supplier agreements."

"Confidentiality obligations have been extended to survive termination. This is standard practice for agreements of this type."

Comments reduce back-and-forth by anticipating the counterparty's questions before they ask them. A redline with a clear explanation is easier to accept than one with no context.

Adding guidance notes

Guidance notes are internal instructions that appear alongside the rule during review. They are not visible to the counterparty. They are for your team.

Use guidance notes to explain the rationale behind a rule, give instructions on how to apply it, tell reviewers which fallback to choose in different circumstances, or set out escalation steps if something is flagged.

For example:

"This position is non-negotiable for contracts above £100k. For smaller contracts, Fallback 1 is acceptable without escalation."

"If the counterparty pushes back on the liability cap, escalate to the General Counsel before conceding."

"This clause is sensitive for public sector clients. Check with the commercial team before redlining."

Guidance notes turn the Playbook from a set of rules into something closer to institutional knowledge. A junior lawyer running the Playbook has access to the same reasoning and context that a senior lawyer would bring to the same review.

Choosing between Redline and Red Flag

For each rule, you decide whether the AI should propose a redline or raise a red flag.

Redline is for positions where the fix is clear and the AI can draft the corrective language reliably. Use Redline when you know exactly what you want the clause to say and the AI can get you there without further judgment.

Red Flag is for issues where the right response depends on context you can't pre-define: the size of the deal, the counterparty, the commercial relationship, or factors that vary from contract to contract. A Red Flag surfaces the issue and stops there. The reviewer decides what to do with it.

A Playbook that uses both appropriately automates what can be automated and surfaces what needs human judgment, without collapsing the two together.

Refining the rules over time

Your first version of a custom Playbook is a starting point. The rules will improve as you run it against real contracts and see where it over-flags, under-flags, or proposes language that needs adjusting.

After each run, note where the output didn't match your expectations and update the relevant rules. Most Playbooks reach a reliable level of accuracy after two or three rounds of testing and refinement. The chapter on testing your Playbook covers this in more detail.

Next chapter → Writing Rules That Work

‍