Document analysis used to depend on attorney experience and team size, but tools like Harvey and Legora now let any subscribing lawyer generate a structured issues list from a redlined agreement in minutes. As adoption grows, buyer's and seller's counsel increasingly run the same document through materially similar AI workflows. The structural consequence is that if both sides use the same tools with default or lightly customized prompts, both receive substantially similar analyses — and the advantage goes to whoever realizes this first.
For most of legal history, the quality and speed of document analysis was a function of attorney experience and the size of the team reviewing the paper. AI legal platforms have changed that calculus. Harvey, Legora, and similar tools allow a lawyer at any firm with a subscription to generate a structured issues list from a redlined purchase agreement in minutes — flagging missing representations, non-standard indemnification caps, unusual closing conditions, and deviation from market practice. As these platforms become more widely adopted, both the buyer's counsel and the seller's counsel increasingly run the same document through materially similar AI workflows. Gurpreet Bal observes that this convergence has a structural consequence that most practitioners have not yet thought through: if both sides are using the same tools with default or lightly customized prompts, both sides are receiving substantially similar analyses. The question is who realizes this first.
When you send a redline, run that same document through your own AI platform before the other side responds. If opposing counsel uses a standard Harvey or Legora workflow without material customization or independent legal layering, the issues list they hand their client will closely resemble what your tool generates. That lets you predict which provisions they will push back on, accept, or miss entirely — so you can brief your client in advance on likely objections, which provisions are defensible, and where early concessions build credibility and help close faster.
The practical implication is straightforward. When you send a redline of a purchase agreement to the other side, you can run that same document through your own AI platform before they respond. If opposing counsel is using a standard Harvey or Legora workflow without material prompt customization or independent legal layering on top of the output, the issues list they hand to their client will closely resemble what your own tool generates. That means you can walk into a negotiation — or prepare your client — with a reasonable prediction of which provisions the other side will push back on, which they will accept, and which they may have missed entirely. Gurpreet S. Bal advises clients to use this window deliberately: when delivering a redline, counsel can brief the client on the issues the other side is likely to raise, which provisions are defensible, and where concessions are worth offering early to build credibility and close faster.
The advantage holds when opposing counsel treats the AI issues list as the work product itself, with minimal refinement. It breaks down when the other side uses AI as an accelerant but applies independent judgment, domain-specific prompts, and transaction experience to filter, prioritize, and supplement the output. A lawyer who sends raw AI output to the client is using AI as a substitute for analysis; one who rejects non-market flags, adds issues the tool missed, and briefs on the delta is using it as a multiplier — and the advantage accrues to that second kind of work.
The predictive advantage is real but conditional. It holds when opposing counsel is using AI output with minimal refinement — treating the issues list as the work product rather than as a first draft to be interrogated by an experienced lawyer. It breaks down when the other side has sophisticated practitioners who use AI as an accelerant but apply independent judgment, domain-specific prompts, and transaction experience to filter, prioritize, and supplement what the tool produces. The distinction matters: a lawyer who runs a purchase agreement through Harvey and sends the output to the client is using AI as a substitute for analysis. A lawyer who runs the same document, rejects three of the eight flagged issues as non-market, adds two the tool missed based on the specific deal context, and then briefs the client on the delta is using AI as a multiplier. Gurpreet Bal advises that the negotiating intelligence advantage accrues to the side whose counsel is doing the second kind of work — because they are both faster than the lawyer who does no AI work and harder to predict than the lawyer who does only AI work.
Before any negotiation response arrives, counsel should brief the client on the likely issues the other side will raise, organized by priority — real concerns that warrant substantive engagement, standard market positions their AI will flag but experienced counsel would not push hard, and positions that reveal something about the other side's risk tolerance or deal motivation. This is not speculation; it is the output of running the same analysis the other side is likely running, filtered through counsel's experience, and treated as a standard part of the pre-negotiation briefing.
For M&A counsel advising sellers or buyers in competitive processes, the AI mirror dynamic suggests a specific approach to client communication. Before any negotiation response arrives, counsel should brief the client on the likely issues the other side will raise, organized by priority — which are real concerns that warrant substantive engagement, which are standard market positions the other side's AI will flag but experienced counsel would not push hard, and which are positions that reveal something about the other side's risk tolerance or deal motivation. This briefing is not speculation; it is the output of running the same analysis the other side is likely running, filtered through counsel's actual experience with what moves in a negotiation and what does not. Gurpreet S. Bal integrates AI-assisted issues anticipation into the deal preparation process for M&A transactions at Foley and Lardner, treating the predictive analysis as a standard part of the pre-negotiation briefing — particularly in transactions where the other side's counsel is not known to the firm and their level of AI sophistication is an open question.
Firms are publicly announcing which AI tools they use through press releases, LinkedIn posts, and website announcements, intending to signal innovation. From a negotiating standpoint, the effect is to tell every future counterparty exactly which tool to run the documents through to predict your analysis, removing one of the few remaining variables in the opponent's uncertainty. Signaling that your firm uses AI is a credibility message; specifying exactly which platform you use, with default workflows, is closer to publishing your playbook.
There is a specific mistake that firms and individual practitioners are making right now that compounds this dynamic: publicly announcing which AI tools they use. Press releases, LinkedIn posts, and firm website announcements celebrating the adoption of Harvey, Legora, or any other named platform are common. The intent is to signal innovation to clients and recruits. The effect, from a negotiating standpoint, is to tell every counterparty in every future transaction exactly which tool to run the documents through to predict your analysis. It removes one of the few remaining variables in the opponent's uncertainty. If opposing counsel has announced their platform and you have access to the same tool, you no longer need to guess what their AI will generate — you can run it yourself with a reasonable expectation that the output will match. Gurpreet S. Bal advises practitioners to think carefully about what tool-specific announcements communicate to the market beyond their intended audience. Signaling that your firm uses AI is a credibility message. Specifying exactly which platform you use, with default workflows, is something closer to publishing your playbook.
Gurpreet S. Bal is a Partner at Foley and Lardner LLP in Silicon Valley, where he advises technology companies, founders, and investors on mergers and acquisitions, venture financings, IPOs, and corporate governance. He has represented clients in hundreds of transactions with aggregate deal value exceeding $60 billion across AI, semiconductors, fintech, and emerging technology.