<aside> <img src="https://prod-files-secure.s3.us-west-2.amazonaws.com/b2fb5bbf-8a6a-4467-b59e-b3f603bd451b/65531e63-ecd3-468a-ad53-bdd500107dc3/swiftlaw_logo_(1).png" alt="https://prod-files-secure.s3.us-west-2.amazonaws.com/b2fb5bbf-8a6a-4467-b59e-b3f603bd451b/65531e63-ecd3-468a-ad53-bdd500107dc3/swiftlaw_logo_(1).png" width="40px" /> Lawyers are insurance. So what tasks are worth insuring $$$?

</aside>

Key questions

1. Who doesn’t want insurance?
2. What are (potential) failure rates across practices or tasks?
3. What situations actually require insurance [lawyers]?
4. What are situations where insurance is mispriced [task is not worth the lawyer fees]?

Here’s a framework to answer the above questions. answers must be low/medium/high/[variable]

• Complexity of the legal task
• Frequency of need per user
• Risk if failure
• Customization requirements

• of target users

• Cost of task

examples for understanding:

1. Non-disclosure agreements (NDAs) for freelancers

• Complexity of the legal task [low]
• Frequency of need per user [high]
• Risk level [low]
• Customization requirements [low]

• of target users [10M+ users/year]

• Cost of task [$200 - $1000+]

Takeaway: This is already automated away by free customizable templates online + ChatGPT/AI models. Low willingness to pay.

2. Filing a civil harassment restraining order

• Complexity of the legal task [medium]
• Frequency of need per user [low]
• Risk level [medium]
• Customization requirements [low]

• of target users [6-7M users/year]

• Cost of task [$2000 - $5000]

Higher risk + obscure domain knowledge → makes usage of generalized AI models unfit. Since output is a court form - there is a binary benchmark/outcome to measure success. Trust can be created by showing proof of prior success.

Takeaway: A software could charge $100 - $200. There is a mispricing of risk. This is an opportunity.