# Sage & Sigil > Plain-English AI policy kits for small law firms, contractors, and defense subcontractors — plus a free AI tools ranking and AI Q&A hub. Sage & Sigil sells plain-English AI acceptable-use policy kits for small U.S. law firms (PII, HIPAA, ABA Formal Op. 512), construction contractors, and defense contractors (NIST SP 800-171, DFARS 7012, CMMC). Educational templates, not legal advice. ## Free resources - [AI Tools Ranking](https://sageandsigil.com/ai-tools-ranking): source-cited models, agents, MCP servers by tier (last reviewed 2026-10-14) - [AI Questions & Answers](https://sageandsigil.com/ai-questions): FAQ, starter prompts, troubleshooting (last verified 2026-10-14) - [Newsletter](https://sageandsigil.com/newsletter) ## Products - [General AI Policy Starter Kit](https://sageandsigil.com/products#starter): $49. An adopt-today AI Acceptable Use Policy and the essentials to roll it out. - [Law Firm AI Kit (PII, HIPAA, ABA ethics)](https://sageandsigil.com/products#law): $149. The Starter Kit tailored to law firms, mapped to ABA Formal Opinion 512. - [Contractor AI Kit (NIST 800-171 / CMMC)](https://sageandsigil.com/products#contractor): $199. The Starter Kit tailored to construction and defense-adjacent contractors. - [CEO “Control AI in 30 Minutes” Kit](https://sageandsigil.com/products#ceo): $29. Three admin steps to take control of AI use in your company this week. - [CMMC Safe-AI Guide with SSP-ready language](https://sageandsigil.com/products#cmmc): $249. AI use mapped to NIST SP 800-171 Rev. 2, with sample SSP and POA&M language. - [Everyday AI Guide (work + family)](https://sageandsigil.com/products#everyday): $19. Simple prompts for your job, and how to teach your family to use AI safely. - [Law Firm Bundle](https://sageandsigil.com/products#bundle-law): $179. Starter + Law Firm Kit + CEO Kit + Everyday Guide. - [Defense Contractor Bundle](https://sageandsigil.com/products#bundle-defense): $399. Starter + Contractor Kit + CEO Kit + CMMC Guide. - [Complete Library](https://sageandsigil.com/products#bundle-complete): $449. Every paid kit (#1–#6). ## About - [About & MSP disclosure](https://sageandsigil.com/about) - [Legal](https://sageandsigil.com/legal) ## Latest AI releases (from https://sageandsigil.com/ai-tools-ranking) - 2026-10-13: Google brings Antigravity into the Gemini Mac app. Google's Antigravity coding agent is now built into the Gemini macOS app; Gemini 4 Argon is available in the Antigravity desktop app and CLI for AI Ultra and eligible Gemini Enterprise customers. Source: https://antigravity.google/blog/gemini-4-argon-in-google-antigravity/ - 2026-10-13: Google releases Gemini 4 Argon. Frontier model, 1M-token unified input/output limit via the Gemini API; available in the Gemini app for AI Pro/Ultra subscribers. Source: https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/ - 2026-10-12: Meta releases Muse Wonder and Muse Wonder Full. Meta's new top models for agents, coding, and computer use, with a 1M-token context. Muse Wonder is on the Meta Model API and Muse Code; Muse Wonder Full is limited to paid Muse Code plans. Meta says its full safety report is coming soon. Source: https://research.meta.ai/blog/muse-wonder - 2026-10-07: Anthropic releases Claude Haiku 5.5. Fast, low-cost model for high-volume work; available on Claude apps, the Claude API, AWS, Google Cloud, and Microsoft Foundry. Source: https://www.anthropic.com/claude-haiku-5-5 - 2026-10-07: OpenAI rolls GPT-6 out to all ChatGPT tiers. GPT-6 Sol (paid tiers) and GPT-6 Luna (Free/Go) in ChatGPT Chat; Enterprise availability depends on admin settings. Source: https://openai.com/index/gpt-6-for-everyone/ ## AI questions & answers (from https://sageandsigil.com/ai-questions, verified 2026-10-14) ### Is it safe to paste client information into ChatGPT or another AI tool? Not into a personal or free account. Only use a tool your firm has approved, on a business plan where training on your data is off and retention is understood. When in doubt, remove names, numbers, and identifying details first. ### Can I use AI on DoD or CUI work? Only with services that meet FedRAMP Moderate or equivalent, as DFARS 252.204-7012 requires for covered defense information. Most consumer AI tools do not qualify. CMMC Phase 2 is currently paused, but NIST SP 800-171 obligations still apply. ### Can I paste an RFQ, part spec, or delivery schedule from a government job into ChatGPT? Not into a personal or public account. Non-public information you get under a federal contract is usually Federal Contract Information (FCI), which CMMC Level 1 expects you to protect. If it is CUI, DFARS 252.204-7012 requires a cloud service that meets FedRAMP Moderate or equivalent. Ask whoever runs your IT which tool is approved for contract data. ### Do lawyers have special rules for using AI? Yes. ABA Formal Opinion 512 (July 2024) applies duties of competence, communication, fees, and confidentiality to generative AI. State bar guidance varies, so check yours. ### Do AI vendors have their own rules for legal, health, or financial work? Yes, and they change. For example, Anthropic's usage policy update (effective Nov 12, 2026) requires a qualified human to review AI output used for decisions about someone's legal rights, health, finances, or livelihood, and requires telling the affected person that AI was involved. Read your vendor's current usage policy alongside your professional rules. ### Do we need an AI policy if employees are already using AI anyway? Especially then. If people are already using AI on their own, a short written policy tells them which tools are approved, what never to paste, and who to ask. Pair it with a quick training on checking AI output, since a policy alone doesn't stop mistakes. ### What should I try first with AI at work? Start with low-risk writing tasks that use no confidential data: rewriting an email, summarizing a public article, or making a checklist. Starter prompt: "Rewrite this email to be friendly, clear, and under 120 words. Keep all facts the same: [paste email with names removed]" ### What's a simple structure for a good prompt? Give it a role, the task, the context, and the format you want back. Starter prompt: "You are an experienced office manager. Create a one-page checklist for onboarding a new employee at a 15-person construction company. Use short bullet points grouped by first day, first week, first month." ### The AI made something up. What do I do? Assume any fact, citation, or number may be wrong until checked. Ask it to show sources, then open them yourself. Never file or send AI citations unverified. Starter prompt: "List each factual claim in your last answer, and for each one tell me how confident you are and what source I could check." ### The answer is too generic. How do I fix it? Add specifics: who it's for, constraints, an example of what good looks like, and ask it to ask you questions first. Starter prompt: "Before you answer, ask me up to 5 questions that would help you make this specific to my situation." ### How do my coworkers and I share AI work safely? Use one firm-managed workspace (a Business/Team/Enterprise plan) instead of personal accounts, so prompts and files stay under company control and leave with the company, not the employee. ### How can I introduce AI to my kids safely? Use it together, check each tool's age terms, never share personal details, and make a game of spotting answers that are wrong. Starter prompt: "Explain why the sky is blue to a 5-year-old in three short sentences, then ask them one fun question about it."