A three-part series exploring what a firm built from the ground up with AI would look like
Part II: Three Practical Steps for Creating a Unified System of AI Legal Service Delivery
In a previous article, I discussed how law firms can rethink pricing and why redesigning service delivery is the key to building a thriving, sustainable practice. Artificial intelligence is reshaping legal markets into a winner-take-most environment—drawing clients toward large, highly scaled firms and pushing smaller players to the margins unless they adapt.
This article focuses on how firms can adapt their service delivery models with AI technologies now.
To ground this discussion, I’ll draw on insights from noted legal futurist and author Richard Susskind. At this year’s Clio Cloud Conference in October, Susskind offered several recommendations for law firm innovation. I’ll focus on three of the most practical.
1. View Legal Work as Content
As LegalTech.ca reported from the conference: “The lawyers of the future, he [Susskind] suggested, will be content providers whose expertise is encoded into products, platforms, and intelligent workflows.”
The internet already hosts abundant legal content—platforms such as Cooley GO, LegalZoom, and OneNDA have long made precedents available at low cost. But these are static documents. Without tailoring to specific facts, they leave clients exposed unless a lawyer intervenes.
Now, AI drafting tools make it possible to generate dynamic documents—customized in real time—with light lawyer oversight.
Meanwhile, many lawyers are discovering that clients are turning first to ChatGPT and similar tools, then bringing AI-drafted contracts and demand letters to firms for correction. The clients who check in with their lawyers are the prudent ones who realize that large language models (LLMs) rarely produce a reliable legal work product alone. The less prudent ones likely rely on these outputs unreviewed.
The opportunity for firms is to place themselves at the center of this emerging ecosystem—connecting precedents, LLMs and human judgment into coherent workflows without AI hallucinations.
Consider a simple example: A small business owner needs a reseller agreement. The firm sends a link to an AI-powered intake agent, which gathers the relevant details and passes them to a drafting agent that produces the agreement with tracked changes for a lawyer’s review. The lawyer finalizes the document—perhaps with a quick email or DM to the client to clarify missing details. Start to finish, the process can take under an hour at minimal cost.
2. Hire New Types of Talent
Susskind observed that for firms to evolve from traditional practice to AI-driven practice, “new hybrid roles—legal data scientists, knowledge engineers, and low-code builders—will be central to this transformation.”
Collectively, we can think of these professionals as legal engineers: specialists fluent in both law and technology.
Why do firms need legal engineers? Because AI doesn’t integrate itself.
Lawyers who simply bolt AI tools onto existing workflows often end up frustrated—spending more time double-checking hallucinations than they otherwise save. A lawyer might recognize a flawed output that a client would miss, but without structured inputs and safeguards, even expert oversight can’t prevent the system from generating mediocre work. Weak prompts produce weak results.
A solicitor friend pointed out that ChatGPT tends to produce 20-clause documents for any use case – no more, no less – which he finds rather suspect.
Legal engineers understand the difference between generative-AI systems (like Harvey) and extractive or classification-based ones (like Relativity), and how to design the right type of workflow for each.
Smaller firms may not have the budget for full-time legal engineers. In that case, they can pool resources—sharing part-time expertise or licensing platforms purpose-built by legal engineers for small-firm deployment. (See the pilot project mentioned at the end of this post for an example.)
3. Don’t Lose Sight of the Lawyer as a Human Being
Susskind closed his remarks with a reminder: lawyers still provide enduring value through empathy, ethical reasoning, and human judgment.
AI cannot credibly guide a client through emotionally fraught situations, build relationships with opposing counsel, navigate ethical gray zones, or read the subtle cues of a witness on the stand.
Therefore, every AI-enabled legal workflow should include a human in the loop—a professional who exercises judgment in the spaces where machines fall short.
The primary benefit of AI is to free up time from busy work and devote it to relationship building and legal strategy. In doing so, firms can advise more clients with higher quality insights—precisely the work most lawyers want to do.
For anyone interested in exploring what an AI-native law firm with these principles could look like, visit legal-now.com—a fictional legal services concept I’ve published to spark ideas and conversation. I welcome input on it and would love to speak to legal engineers – lawyers and developers – who want to get involved in a project to build such a firm!
Steve Lowry is a lawyer turned technologist, cofounder of the Artificial Intelligence network of British Columbia (AInBC) and the currently CEO of Broadsight, an AI-based workflow platform for communications professionals.
To continue exploring these topics and take part in the conversation, reach out to Steve on LinkedIn.
