Legal — Scott Palmer

Republished with permission from the FrUn SMB AI Guide. Canonical chapter: Superhuman Docs → · All chapters

Scott Palmer · Fractional General Counsel · SCO Palmer Firm · LinkedIn

1. Mindset — What an SMB Should Believe Before Spending a Dollar on AI

Ask a legal question of any AI. It will produce fluent, confident, professional-sounding language which may not be correct for your situation, your state, or your deal. SMBs who treat the output as a definitive answer end up signing agreements that look clean but might carry a one-sided indemnity, an auto-renewal, or an uncapped liability clause that nobody flagged. The document looked finished, so it got signed. A single misread or poorly phrased clause can cost your business more than a year of human legal services. AI for legal tasks is a fine start, but a human needs to be in the loop.

Where AI actually helps: summarizing a long contract and translating it into plain English, translating legalese so you understand what you’re agreeing to, generating a first-draft NDA or policy, building a review checklist, and organizing the contracts you already have. These are real hours saved.

Where AI merely feels productive: asking AI to decide whether to sign, to assess your specific legal risk, or to interpret how a law applies to your business. You’ll get a fast, tidy answer that feels like progress. But it could also be the part of the process most likely to be wrong, and the part where being wrong is the most expensive.

Get ready for AI:

• Know the business terms you want, such as price, scope, term, and what you will and won’t accept on liability. AI can’t decide your risk tolerance for you.
• Pull your contracts together in one place, not scattered across inboxes. AI can’t organize what it can’t see.
• You or your legal counsel can sanity-check the output. Garbage into AI produces confident garbage out, and in legal language the garbage issues may be hard to spot.

2. The SMB Tool Stack — Start Cheap, Stay Simple

You do not need legal-specific AI software to get the most value. The tools built for lawyers are priced for lawyers. Start with general-purpose tools and a place to sign.

The two tools you need to start:

• A general-purpose AI assistant: $20-30/user per month. Claude Pro ($20/mo) or ChatGPT Team ($25-30/user/mo) covers nearly everything an SMB needs from “legal AI”: summarizing incoming contracts, drafting first-pass NDAs and policies, and building review checklists. It replaces the blank page and some of the $300-500/hour “what does this clause mean?” emails you may currently send to a lawyer. Use the Business/Team tier, not the free consumer one (see Guardrails).
• An e-signature tool: $15-25/user per month. Dropbox Sign ($15-25/user/mo) or DocuSign ($15-40/user/mo). Not AI, but it pairs with it: once AI helps you produce a clean template, e-sign is what gets it executed without the print-sign-scan-chase cycle. Dropbox Sign is the cheaper, simpler pick for most small teams.

Tools to skip or postpone, and why:

• Dedicated AI contract-review platforms (e.g., Spellbook, $99-199+/user/mo). These are excellent for law firms reviewing dozens of contracts a month inside Word. However, for a 5-50 person SMB, you’re paying 5-10x the price of a general assistant for features you won’t use enough to justify the cost. Postpone until you have a real, repeating contract volume.
• Enterprise contract lifecycle management (CLM), the Ironclad/Conga tier. Built for hundreds of active contracts, an admin to run it, and an annual enterprise contract. If a tool needs a dedicated admin or a sales call to get a price, it’s not the right starting point.
• “AI lawyer” chatbots that market themselves as replacing counsel. The marketing is the warning label. They’re fine for orientation, but they are not a substitute for advice on anything that matters.

A free or near-free starting point with zero budget:

• The free tier of Claude or ChatGPT for summarizing and drafting (mind the data warning in Guardrails), a well-structured Google Drive or Dropbox folder as your single contract repository, and the free tier of an e-sign tool like ZohoSign for low volume. That combination costs nothing and still gets you a review habit, organized contracts, and signable templates.

3. The First 30 Days — Where to Start

Start by building a contract intake and summary habit. Before you sign anything, paste it into your AI assistant and ask for (1) a plain-English summary and (2) a list flagging anything unusual, one-sided, or risky such as term length, auto-renewal, termination, liability, payment, IP ownership, confidentiality. This run-through doesn’t replace a lawyer. It tells you when to call one, which is where SMBs waste the most money on legal spend.

A week-by-week order of operations:

• Week 1: Set up and centralize. Pick one AI assistant and one shared folder. Move every active contract into that folder. You now have a single source of truth and can’t sign blind because you can’t find the file.
• Week 2: Build your review checklist. Create one reusable prompt that asks the AI to summarize any contract and flag the same risk categories every time. Run it on 3-5 contracts you’ve already signed to ensure it gives you what you need. You’ll usually find at least one term you didn’t realize you agreed to.
• Week 3: Templatize your common documents. Have AI draft first-pass templates for your 2 or 3 most-used documents, typically an NDA, a master services agreement or SOW, and a basic offer letter. Don’t deploy them yet.
• Week 4: Get the templates reviewed by counsel and wire up signing. Have your legal counsel review those master templates once (a few hours of fixed cost), then route them through your e-sign tool and document the process so anyone on the team can follow it without you.

What “good” looks like at the end of 30 days:

• Every contract lives in one place.
• You have a repeatable review checklist you run before signing anything.
• You have 2 or 3 counsel-approved templates to reuse without paying per deal.
• You can tell which decisions need a lawyer, and which probably don’t.
• You’ve stopped signing things you haven’t read in plain English.

4. A Real SMB Use Case

• The business: 12-person digital marketing agency, around $2M in revenue, no in-house legal and no operations hire. The owner handled contracts between client calls.
• What was broken: Client master service agreements and vendor agreements were signed ad hoc with no templates. Contracts lived in email threads. One client agreement had auto-renewed at an unfavorable rate because nobody tracked the notice window, and another carried uncapped liability nobody had flagged.
• What AI did: Summarized incoming client redlines in plain English, drafted a standard MSA and SOW template, and produced a one-page “fallback positions” cheat sheet so the owner knew what to push back on.
• What the owner did themselves: Made the actual business calls such as setting the liability cap the agency would accept and deciding payment terms. They asked me to review the master templates and the prompts for the AI reviews before they went into production. The AI never decided anything, it prepared the owner to make decisions.
• Tools used: Claude Pro ($20/mo), Dropbox Sign ($25/mo), and roughly six hours of my time at the start.
• What it cost: About $45/month in software, plus a one-time $2,500 for me to review and clean up the templates.
• Time: Roughly 10 hours of the owner’s time setup spread across a month.
• Result: The team caught and renegotiated two unfavorable renewal clauses on existing deals. The owner stopped sending me one-off “can we accept this?” emails because those got triaged by AI first, cutting roughly $1200/month in reactive legal spend. The fixed setup cost paid for itself in less than two months.

5. Guardrails — What an SMB Should NOT Do

The mistakes I see SMBs make:

• Pasting confidential documents, signed contracts, or customer personal data into a free consumer AI tool that may train on your inputs. Use a business/Team tier with no-training and actual data processing terms, or strip identifiers first.
• Treating AI output as legal advice. It will state the wrong law with total confidence, occasionally cite cases that don’t exist, and miss requirements specific to your state. “Don’t trust it blindly” is good, but the concrete rule is: don’t let AI generate customer-facing legal terms (privacy policy, terms of service) and publish them without a human review. Regulators do not accept “the AI wrote it” as an excuse.
• Auto-generating employment documents like offer letters with restrictive covenants, handbooks, and termination notices. Employment law is state-specific, and a wrong or ineffective non-compete, non-solicit, or a misclassified contractor is one of the most expensive errors a small business can make.

What to keep humans in the loop on, no matter how good an AI tool gets:

• The decision to sign anything, this should always be a human.
• Anything regulated or public-facing: privacy policies, terms of service, financial, health, or consumer claims, and anything employees or customers rely on.
• Employment terminations and anything about personal data of customers/staff.

Privacy, data, and compliance issues SMBs underestimate:

• Confidentiality breaches you create yourself. Pasting a third party’s contract into an AI tool can itself violate the confidentiality clause you signed in that very contract. Check your obligations before uploading anyone else’s document.
• Whether the tool trains on your data. Free consumer tiers frequently do. Know the settings and the terms of use before sensitive content goes in.
• No privilege. Your AI chat history is not protected by attorney-client privilege. Don’t treat the chatbot like your lawyer for anything sensitive or disputed because the entire conversation could become discoverable.

6. Lessons Learned + One Thing You’d Tell Every SMB Owner

From using AI in my legal practice and with clients, here are the hard-won lessons:

• AI is best used as a translator and first-drafter, not a decision-maker. Its single highest-value job is telling you when something needs a human.
• The bottleneck is almost never the tool; it’s disorganized contracts and unclear business terms. Fix the inputs before you buy or subscribe to any software.
• Business-tier AI with actual data processing terms and AI-training prohibitions is worth the few extra dollars a month. Never paste sensitive documents into a free consumer tool to save twenty bucks.
• Templates blessed and regularly updated by your legal counsel easily beat per-deal legal review of a new-to-you, reused, or repurposed contract. Spend the money up front, these are the cheapest legal services you’ll ever buy.

The one thing I’d tell every SMB owner:

Use AI to know when you need a lawyer, not to replace one. It’s always cheaper to catch a legal mistake before you sign the deal.

Part of the FrUn SMB AI Guide. Read on canonical guide →