Solo and small law firms comparing ChatGPT to specialized legal AI are asking the wrong first question. Here is the one to answer before either.
A piece ran in The National Law Review on July 3 asking a question I see solo and small firms get stuck on constantly: general-purpose AI or legal AI. ChatGPT and Claude, or Harvey and Legora and CoCounsel.
The numbers in it are worth sitting with. Clio's 2025 Legal Trends data found only 8% of solo practitioners and 4% of small firms have adopted AI widely, against 35% of large firms. But the tools they do reach for are almost all general-purpose: 57% of solo lawyers and 54% of small-firm lawyers are using something like ChatGPT, not a legal-specific platform.
Cost explains most of that gap. ChatGPT Plus runs $20 a month. Thomson Reuters CoCounsel starts around $784 a month for a solo attorney on an annual plan. Harvey and Legora do not even publish pricing, which tells you who they are built for.
Legal technology consultant Ernie Svenson made the point in the piece that I think gets missed: the advantage in this moment actually sits with the solo and very small firm, not against them. Larger organizations move slowly. A one or two person firm can pick up a general-purpose tool and start using it directly, without six months of procurement.
The wrong first question
But here is where I think the article asks the wrong first question. Which tool costs less, or which tool is purpose-built, is a real question. It is not the first one. The first question is whether the firm has actually addressed what happens to client information the moment it goes into any AI tool, specialized or not.
Every firm I talk to in this position has some version of the same admission: I do not have a policy for this. Not "our policy says no AI." Not "our policy allows it with these guardrails." Nothing written down at all. That gap is not a technology gap. It is a governance gap, and it exists whether the tool in question costs $20 a month or $784.
ABA Formal Opinion 512 already sets the bar here. Lawyers are required to understand whether the AI system they are using is self-learning, and to get informed consent before client data touches it. A line buried in a standard engagement letter does not satisfy that. Most solo and small firms have not looked at this yet, which means most of them are already out of step with their own professional obligations, regardless of which chatbot or platform they picked.
Three questions before either tool
So before a firm spends real time comparing ChatGPT to Harvey, I would want them to answer three things first. What client data, if any, is going into this tool. Whether clients have been told and have actually consented, not just signed something they did not read. And who in the firm is responsible for knowing the answer to the first two questions six months from now, when a different associate is using a different tool for a different matter.
This is the same order of operations behind the Conscious and Deliberate AI Adoption Framework: governance and permission come before tool selection, not after. A firm that answers these three questions honestly has done the harder, more valuable work, regardless of which product they end up choosing.
Get those three answered honestly, and the general-purpose-versus-legal-AI decision becomes a much smaller, much more practical question. Skip them, and it does not matter which tool the firm picked.
If your firm is ready to work through these questions properly rather than guess at them, book a call and we will start there.