Contents 🚀
Introduction 👋
With the right guidance, Juro's AI Assistant can review, draft and summarise your documents in lockstep with your organisation's policies and procedures. Providing the AI Assistant with the context and guardrails to work within enhances its outputs and overall contribution to your contract workflows. This can be given in the form of a playbook, which can be set up on your templates and generated contracts.
|
💡 NOTE: Access to the AI Assistant is assigned as follows:
Templates: Team Admins
Generated contracts: Team Admins and Editors
Juro playbooks and prompt outputs are supported in all languages.
|
Guardrails 🎳
What is an effective guardrail? 🛤️
Guardrails are the parameters that give the AI Assistant a better understanding of your organisation's contracting procedures and principles.
Providing it with clear instructions enables it to produce the most accurate results.
In the context of your Juro playbook, well-defined criteria is the guidance the AI Assistant will rely on to draft, summarise and review your contracts.
Adding guardrails to your playbook 🎨
Within your template/document, navigate to the AI PLAYBOOK section of the AI Assistant tab in your template:
|
|
An effective guardrail will include context, positions and principles, limits and output expectations:
1. Context
|
As with any AI tool, it's helpful for the Assistant to know what their role is in the process. This may include the job title, who the Assistant is acting for and any specialist expertise. For example:"You are an English-qualified lawyer and an expert in commercial SaaS contracts operating under the laws of England and Wales. You acting for the Supplier. The services typically provided are cloud-based, and specific line items are specified within each document."
|
2. Positions and principles
|
Next, add guidance on your organisation's positions, principles and must-haves for your organization when drafting or negotiating documents. For example:
"Below are [#] policies in our Master Services Agreement Playbook. For an agreement to be compliant, all of the following policies must be adhered to. Any deviation from these positions means the document does not comply with our business standards and cannot be accepted."
Suggested areas to add must-haves with examples:
We would recommend adding all areas, principles and policies required by your business per template or document type. Concise numbered lists will be really helpful for the Assistant. |
3. Limits
|
The AI Assistant should also be given clear definitions of what it can't do and can't accept during the contracting process for this document type.
These limits should be added across each area with clear rejection criteria. As an example"Under our playbook, we never accept: (a) liability caps for the customer in excess of $100,000; (b) any governing law other than English law; a default interest rate of higher than 4% per annum above the Bank of England base rate; or (d) any warranty period shorter than 12 months."
|
4. Output expectations
|
An additional helpful section to add is one defining any expectations from the AI Assistant regarding outputs. Admins/Editors will prompt the Assistant to review the document against this playbook; adding a section on expectations can help enhance the outputs these members receive. An example:
""When asked whether a document or contract complies with the playbook, I want you to give your answers in the format delimited by triple quotes.
The text marked as "EXAMPLE" shouldn't be quoted directly; it should be used as an example of how to proceed when you explain whether each playbook policy has been complied with: "[YES or NO], this document [DOES or DOES NOT] comply with our playbook. [SCORE OUT OF 11].
Explanation for each policy: EXAMPLE: 1. The purpose is specified in Section X. EXAMPLE: 2. THE definition of Confidential Information is general and non-exhaustive in Section X.""
|
Sharing an example for an NDA playbook containing context, must-haves, limits and output guidance:
"Below are 11 policies in our Non-Disclosure Agreement Playbook. For an agreement to be compliant, the following policies must be adhered to.
1. The purpose must be specified.
2. The definition of Confidential Information must be general and non-exhaustive.
3. We only accept the following governing laws: i) English Law ii) the law of any country within the EEA (including: Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic (Czechia), Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden.) iii) the law of any state within the United States (including: Alabama, Alaska, Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Florida, Georgia, Hawaii, Idaho, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maine, Maryland, Massachusetts, Michigan, Minnesota, Mississippi, Missouri, Montana, Nebraska, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, South, Dakota, Tennessee, Texas, Utah, Vermont, Virginia, Washington, West, Virginia, Wisconsin, Wyoming).
4. The courts must have exclusive jurisdiction to resolve any disputes.
5. The recipient must be permitted to disclose Confidential Information to people who need to know it for the purpose.
6. The recipient may only use the Confidential Information for the stated purpose - no other reason.
7. The recipient may disclose Confidential Information if required to do so by law or by a regulator.
8. The obligations of confidentiality should last a minimum of 2 years and a maximum of 5 years after either (i) disclosure of the Confidential Information or (ii) termination of the NDA.
9. We don't accept any indemnities in the NDA.
10. The NDA must have an end date, or alternatively, the parties must have a right to terminate the NDA by giving notice.
11. The NDA must be mutual, not a one-way NDA.
When asked whether a document or contract complies with the playbook, I want you to give your answers in the format delimited by triple quotes. The text marked as "EXAMPLE" shouldn't be quoted directly; they should be used as an example of how to proceed when you explain whether each playbook policy has been complied with.: """ [YES or NO], this document [DOES or DOES NOT] comply with our playbook. [SCORE OUT OF 11] Explanation for each policy: EXAMPLE: 1. The purpose is specified in Section X. EXAMPLE: 2. THE definition of Confidential Information is general and non-exhaustive in Section X. "
Accessing the AI Assistant 🐎
After you've finished writing your playbook, we suggest learning more about writing impactful legal prompts in our dedicated article.
Once you've done this, there are two ways to access the AI Assistant:
1. Using the Actions bar in your contract: Click the Sparkle icon in your contract's Actions bar. In the sidebar that opens, write your prompts in the REQUEST panel.
|
|
By default, the AI Assistant scans the whole document for its responses. We suggest focusing on specific clauses or parts by selecting them first, or explaining to the Assistant what clause you'd like it to focus on. This will produce a quicker, more. specific result.
|
2. By highlighting a portion of text within your document (e.g. a clause) to see the AI window appear.
|
|
💁♀️ As always, our Support Team is happy to help you with anything further if needed. Start a chat with us right here by clicking the Intercom button in the bottom-right-hand corner of this page.
Alternatively, you can email your query to support@juro.com 🚀
|