Legal AI: 10 Things We Look For

The practice of law has remained largely unchanged for decades. The industry’s technology and innovation resistance is attributable to a multitude of factors, ranging from law firm business models (e.g. billable hours that disincentivize automations) to more nuanced, cultural operating practices (e.g. associates “paying their dues” by working long nights on mundane tasks for training purposes).

Whether the industry is ready or not, potential AI-enabled efficiency gains have become too robust to ignore. Legal work leverages and produces copious amounts of critical and unstructured data that can now be leveraged to both accelerate and improve work outputs, in addition to drastically improving workflows. (A reality clients are becoming increasingly aware of, creating an added layer of external pressure for firms to adopt AI tooling).

After speaking with dozens of legal AI companies over the last year+, and investing in others like Noetica, below are ten learnings and business attributes we consider when evaluating startups and the intersection of law and AI.

We look for products that:

1. Expand beyond SaaS/seat based pricing. The legal market is not massive and hyper fragmented with a long tail of small and solo practitioners (it’s further fragmented by the wide range of legal specialities). The largest ACVs and revenue opportunities will require a departure from simple SaaS pricing (e.g. capturing percentage of transactions, savings generated, etc.).

2. Unlock/generate new data sets. Many lawyers want to create information asymmetries. They are incentivized to buy products that will give them an information edge against peers. (Or, ensure their peers don’t have a data advantage over them).

3. Involve complex transactions and non-legal stakeholders (mostly relevant for big law and top-tier firms). Involving other parties increases product stickiness and overall market size.

4. 10x speed/quality of tasks that are not billed in-full to clients. Certain aspects of legal work are not fully passed through to clients because lawyers are expected to be perpetually “up to date” or “experts”. Products that accelerate non-billable legal functions will likely be easier to sell.

5. Target law domains that deploy flat-fee, or per-transaction pricing. Firms/lawyers with flat fee structures are often SMBs, or solo practitioners, and customer growth is the key to unlocking incremental revenue. These transactions are also more “off the shelf” and relatively consistent. These practices may be most willing to adopt AI automation tooling, as it frees up time for customer, and thus revenue, growth.

6. Don’t have “black box” automation solutions. Lawyers want visibility into AI product outputs for easy fact and logic checking. The stakes are too high for many legal professions to not have a “human in the loop”.

7. Target law domains with strong network effects, and word of mouth growth potential. (e.g. legal domains where there is a uniting commonality, but also further specialization, so it’s likely less competitive.)

8. Target legal domains with newfound relevance because of AI and/or rapidly changing regulations (e.g. copyright infringement)

9. Are primarily built for firms (vs. in-house counsels). In-house legal teams are generally smaller and less-strategic. The “land and expand” opportunities for in-house counsel tools will likely be more limited.

10. Unlock customer growth opportunities and net new cases, especially for SMBS and solo practitioners. (e.g. Dorrow)

If you’re building in the space, we’d love to connect — julia@flybridge.com.