In this work, I argue that almost all categories of legal tasks can be, in principle, performed more efficiently by legal AI (i.e. specialized AI for legal services). However, in legal services, the monopoly of the legal profession is embedded in law and is enforced through the Unauthorized Practice of Law (UPL) regulations. UPL acts as a barrier to large-scale introduction of AI tools in legal services, which ought to be removed. Legal AI can enhance access to justice and consumer welfare and promote disintermediarization of law, thereby empowering the consumer of legal services. I address some key concerns associated with this, namely the problems of accountability for legal AI and problems arising out of AI bias, hallucination and the blackbox effect. These problems are all subject to technological solutions which in turn depend upon a proper regulatory scaffolding involving appropriate assurance, standardization and product and service liability norms. A contour of such regulation might already be emerging. Concerns arising from possible social and human consequences of diluting a lawyer’s role are considered as well. The regulations in the United States and the United Kingdom are considered to share learnings. I conclude that in the near future, the legal services market may be pivoted on legal AI with a far diminished role for lawyers.