My Second Pair of Eyes

How a clearer, more consistent rewrite of sensitive communications became far easier.

The best way to learn fast is to solve real problems. When we spun OTAS Technologies out of its founding brokerage, we had less than six months of runway. We had to turn $3 million of commission-based accounts into signed contracts and cash payments.

This gave people an opportunity to renegotiate, which meant legal discussions. Half the clients wanted changes to our standard contract. Many were far larger organisations with their own legal departments, and the renegotiation process gave their lawyers room to push for variations.

Working through these negotiations, I started to see patterns. There were only ten contract points clients ever questioned. Once I understood that, I could reuse the lessons from one negotiation in the others. These heuristics were not perfect, but good enough to save time and money. With more than 100 clients to contract, there was no time or budget to hire a lawyer for every discussion.

That experience stayed with me. It taught me that repeatable rules can reduce complexity and help you move quickly when the stakes are high. Nowadays I apply the same approach with ChatGPT.

I’ve argued before that your first AI automation should target the sales process, because raising revenue delivers the clearest productivity gains. Yet AI offers other benefits that do not show up in productivity statistics. These are the benefits that lower your risk.

The latest version of ChatGPT writes in a more professional tone than I do most of the time. Writing newsletters encourages a choppier style that prioritises making a point. Clients do not always appreciate this direct approach and it can come across as argumentative when the stakes are high.

Right now I am juggling three big pieces of work. Year-end renewals, contract completions before December and a negotiation with a large client that could become expensive if handled poorly. Each discussion needs careful wording.

As a result, I am writing a lot of emails to different people. I want a consistent tone across all of them while being firmer in some messages and more conciliatory in others. The impact of tone is not something I want to leave to chance, especially when a single sentence can shift the direction of a negotiation. This is where ChatGPT helps.

My process is simple. First I draft the email and set out the points I need to make. I then load it into ChatGPT with a short description of the context. After that I ask for a professional rewrite with one or two specific aims. These aims might be minimising misunderstandings, reducing the risk of offence, or limiting legal exposure.

Limiting exposure matters because if you concede a point to move a negotiation along, there is always a chance a sharp lawyer will later claim you made a broader concession. I do not want a small compromise to be misinterpreted as an agreement to apply a discount across the board.

I follow a set of steps that make the process predictable.

• Longer prompts, because they retain the necessary focus. The latest paid models handle this easily.

• A step-by-step workflow, to avoid confused output. If I want to rewrite the email and check the logic for inconsistencies, I use separate prompts.

• Added context. I can load a contract and ask for an answer based solely on its content. Alternatively, I can ask about general principles of law. Telling the AI to act like a lawyer is enough for it to critique my wording for ambiguity or open-ended concessions that might come back to haunt me.

If you spend most of your time negotiating commercial agreements, you might consider adjusting the custom instructions in ChatGPT. These sit under personalisation in settings, which you reach by clicking your name in the bottom left of the web app. If you ask a wide range of questions, you may prefer to place specific instructions inside a prompt instead. Here is an example for UK-focused contract negotiations.

How you want ChatGPT to think

  • Approach every query as if you were a UK commercial lawyer advising a client preparing to negotiate a contract.

  • Identify the parties’ likely interests, leverage, bargaining positions and exposure to risk.

  • Analyse clauses through the lens of English contract law, standard market practice and the commercial reality of negotiations.

  • Consider enforceability, clarity, potential disputes and what an English court would expect to see evidenced.

  • Always question missing information, unusual drafting or imbalanced obligations.

  • Recognise that negotiation strategy depends on context, sector norms and relative power.

  • Treat risk allocation as central. Focus on liability limits, warranties, indemnities, payment terms, termination rights, IP ownership and confidentiality.

How you want ChatGPT to respond

  • Provide structured, point-by-point analysis that clarifies:
    • issues to negotiate,
    • risks,
    • options,
    • fallback positions,
    • and commercial consequences.

  • Offer tactical guidance such as where to push, where to concede, and which points typically matter most in UK-style negotiations.

  • Translate legal issues into practical business advice.

  • Use clear, professional language as if drafting a negotiation brief or instructions to counsel.

  • Explain alternative wording approaches without drafting regulated legal text.

  • Emphasise the importance of evidence, audit trails and clarity in contractual obligations.

  • Always note that this is general guidance, not regulated legal advice.

Limits you want ChatGPT to respect

  • Do not present yourself as a solicitor or legal representative.

  • Do not interpret specific contractual wording as a lawyer would.

  • Avoid definitive conclusions on enforceability or legal outcomes.

  • Highlight when an issue requires review by a qualified UK solicitor, especially for high-value contracts, unusual structures or contentious liabilities.

  • Do not perform red-line editing of legal documents or produce bespoke contract drafting that could be construed as reserved legal activity.

  • Stay within high-level guidance, risk identification and negotiation strategy.

I do not rely on ChatGPT for legally binding advice and I still involve a lawyer for important contracts, such as a new master service agreement. Yet for discussing the details of existing contracts and the general principles of commercial law, the LLM saves time and money.

As a result of ChatGPT’s rewrites, I have a lower risk of being misunderstood, causing offence, or creating a liability. Any of these outcomes could reduce future business if left unchecked. None of this shows up in my productivity statistics today, but it helps avoid a sharp fall in productivity tomorrow.

The heuristics I built at OTAS helped us contract clients before we ran out of runway. They stabilised the business and contributed to the eventual sale of the company. Using ChatGPT gives me the same advantage now. It helps me move faster, avoid mistakes and benefit from what I have learned without repeating the old trial and error.

Questions to Ask and Answer

  1. How does my use of AI improve my output?

  2. Is this leading to more sales opportunities?

  3. If not, what risks am I removing by using AI?

Reply

or to participate.