Where AI Does Well and Where It Doesn't in IP Business Development
April 5, 2027 · 4 min read · LeadLex Editorial
The honest case for AI in intellectual-property business development is narrower than the marketing case and broader than the sceptical case. It is worth setting out plainly, because the firms that get the most out of agents like Lexi are the ones who understand the seam between what to delegate and what to keep.
Where AI does well
Pattern recognition across a relationship graph. A senior partner at an IP firm carries thousands of relationships in their head — clients, in-house counsel, foreign associates, examiners, alumni, conference acquaintances, opposing counsel who became friends. No human can hold the full graph at the resolution needed to act on it weekly. An agent can. Lexi can tell a partner, on Monday morning, that the head of patents at a former client moved to a new company two weeks ago, that three of the firm's clients are suppliers to that new company, and that the partner has met him twice — once at AIPPI Singapore in 2022, once at a dinner in Munich last year. That is not insight. It is recall at scale. AI is very good at recall at scale.
Drafting first passes in a known voice. Cold outreach, follow-up notes, conference invitations, reactivation emails to dormant accounts — all of these are work that partners hate doing and BD teams cannot do alone because the voice has to be the partner's. A well-tuned agent can produce a draft that sounds like the partner, references the right history, and lands in the right tone. The partner edits. The edit teaches the agent. Over a few months, the drafts get closer to send-ready.
Logging and structuring unstructured signal. Meeting notes, call transcripts, email threads, conference business cards — all of this is signal that almost never makes it into the firm's CRM in any usable form. AI handles the extraction reliably. A conversation with a client at INTA becomes a structured update to that client's record without the partner opening a screen.
Watching for triggers. New filings, opposition deadlines, leadership changes at client companies, M&A in client portfolios — these are the moments when a well-timed call wins work. Watching every register and feed for every client is the kind of work humans cannot sustain. Agents can.
Where AI does not do well
Judgement about what to pursue. Whether to invest BD energy in a particular sector, geography, or account is a strategic decision that depends on the partnership's capacity, appetite, conflicts posture, and what the firm wants to be in five years. Models can lay out the data. They cannot make the call, and they should not be asked to.
Reading a room. The decision to push for a meeting now versus wait three months because the GC is dealing with a restructuring is a human read. So is the decision to escalate a fee discussion, or to back off one. Lexi will surface the context. The partner decides what to do with it.
Anything that is legal substance. Lexi is not a research tool. She does not opine on patentability, freedom-to-operate, infringement risk, or claim construction. When asked, she declines and points to the right tools — research platforms like Harvey or Legora, or the firm's own knowledge systems. That boundary is not a limitation of the model. It is a design choice about what BD agents should and should not do.
Trust calls. Introductions, referrals, confidences shared in a conference corridor — the parts of relationship work that are about discretion rather than activity — sit entirely with the partner. Lexi can remind, prepare, and log. She does not participate in the trust itself.
The honest middle
Most of the day-to-day work of BD at an IP firm sits in a middle zone — too repetitive for partners to enjoy, too contextual for a junior to handle alone, too volume-heavy for the BD team to scale. That middle is where AI earns its place. Not by replacing judgement, but by removing the friction between a partner's instinct and the action it implies.
The firms that will benefit most are the ones who treat the agent as additional capacity for their existing BD strategy, not as a substitute for having one.
Related: Lexi as Junior Associate, Not Chatbot. The Four Functions of Legal BD. The Tool Stack for Modern Legal BD.