Gurpreet S. Bal has negotiated across the table from counsel on hundreds of M&A transactions. His view on the current AI moment in deal practice is pointed: "Announcing which AI platform your firm uses is the legal equivalent of publishing your playbook. The people reading it aren't your clients." The information asymmetry that used to separate sophisticated deal teams from less experienced ones is narrowing. How you use the tools — not which tools you have — is where the advantage now lives.
Bal advises buyers, sellers, and investors across complex technology and private equity transactions and has watched AI reshape deal dynamics in real time.
The market consolidation around a small number of legal AI platforms — Harvey, Legora, and a handful of others — means that sophisticated deal teams on both sides of a transaction are increasingly working with the same underlying technology. A buyer's counsel using Harvey to analyze the seller's disclosure schedules is not necessarily doing anything that seller's counsel cannot do in reverse. This symmetry changes the nature of the AI advantage. When only one side had AI-assisted contract review, the advantage was speed and coverage — more issues surfaced faster. When both sides have those capabilities, speed and coverage become table stakes. The advantage shifts to the quality of the prompting, the experience of the lawyers directing the analysis, and the judgment applied to what the AI surfaces. The tool democratizes access to pattern recognition. It does not democratize the judgment required to know which patterns matter in a specific deal.
One of the more tactically interesting applications of shared AI tools in M&A practice is using them to predict opposing counsel's issues list before they send it. If both sides are using similar AI platforms with similar training on market-standard documents, a buyer's lawyer can run the seller's draft agreement through their own AI review — not just to identify buyer-side issues, but to identify the issues that a seller-side AI review would flag as well. The output is a rough map of the negotiating terrain before the redline arrives. This is not hypothetical. It is a natural consequence of AI tools trained on similar corpora of market-standard agreements. The deals where this technique is most valuable are the ones where the other side is less sophisticated — where they are relying on AI output without the human judgment to prioritize it. Against an experienced, well-resourced counterparty who is also running AI, the predictive value narrows, but the technique still informs preparation.
The AI negotiation advantage is real and durable in one specific context: when the counterparty is using AI tools without sufficient experience to evaluate and prioritize the output. AI contract review tools generate issues lists. A junior associate at a small firm, given an AI-generated issues list and insufficient deal experience, may raise every flagged issue — including low-stakes items that experienced deal lawyers would trade away without discussion. Knowing this, an experienced counterparty can use the pattern of issues raised as a signal about the other side's experience level and adjust their negotiating posture accordingly. The advantage breaks down when both sides have experienced deal lawyers directing the AI. At that level, the AI compresses time and improves coverage for both parties equally. The remaining advantage is entirely in the quality of human judgment — who has done more transactions of this type, who knows the market better, who has a clearer sense of what their client actually needs versus what the AI flagged as a deviation from form.
Law firms and legal operations teams have developed a habit — driven partly by marketing, partly by client demand for transparency — of publicly announcing their AI platform partnerships. Press releases naming which AI tools a firm has licensed, case studies about their AI deployment, social media posts from partners about the tools they use. This is a mistake with real consequences in adversarial contexts. If opposing counsel knows your firm has licensed Harvey and is using it on deal negotiations, they know roughly what your AI-assisted issues list looks like, how it is structured, and what it is trained to flag. They can run the same tool on their own draft and see your likely output before you send it. They can craft their initial draft to minimize the surface area the tool will flag. They can train their own prompts to work around the patterns your tool emphasizes. None of this requires industrial espionage. It requires reading your firm's press releases.
The next evolution in legal AI — goal-based negotiation agents — introduces a risk that the market has not adequately confronted. An AI agent tasked with "negotiate the best possible indemnification cap" will optimize toward that goal through a sequence of decisions: what to concede first, what to hold, when to trade. Those intermediate decisions are judgments. They affect the overall deal, not just the provision being negotiated. A concession made by an AI agent in one provision to preserve position in another may not be the trade-off a human lawyer would have made — and may not be what the client actually wants. The lawyer who deployed the agent and approved the final outcome may never have seen the intermediate concessions. This is not a theoretical risk. It is the structural consequence of deploying goal-directed AI in adversarial contexts. Until there are robust frameworks for supervising agentic negotiation AI — which there are not yet — the responsible position is to keep AI in the analysis and preparation role and keep the human in the room for the actual negotiation.
Gurpreet S. Bal is a corporate partner with 16 years advising on private equity, merger transactions, and public offerings for companies and investors at three of the world's top law firms. He has represented clients in hundreds of transactions with aggregate deal value exceeding $60 billion across AI, semiconductors, fintech, and emerging technology. For more information and to get in touch, visit gurpreetbal.com.