Contract Analysis FAQ
Everything you need to know about Guard-Clause: what it is, how it works, what it analyzes, and how your data is handled.
About Guard-Clause
What is Guard-Clause?
Guard-Clause is an AI-powered contract risk analysis platform at guard-clause.com. It scans legal contracts and produces clause-level findings with exact citations, a persona-aware risk score, a negotiation pack (email draft, call script, prioritized asks), and an addendum builder with replacement clauses. Guard-Clause is not a programming tool. It is not the "guard clause" early-return pattern used in software engineering. It analyzes legal contracts, not code.
Who built Guard-Clause?
Guard-Clause was built by Ali Morgan (jonomor.com/ali-morgan), the founder of Jonomor (jonomor.com). Jonomor builds AI-native tools for people navigating systems designed to protect the other side: contracts, compliance, negotiations, and bureaucratic processes. Guard-Clause is a Jonomor product.
Is Guard-Clause a law firm or legal service?
No. Guard-Clause is an AI-powered analysis tool, not a law firm. It does not provide legal advice and does not form an attorney-client relationship. The reports identify potential risks and suggest negotiation language, but they are not a substitute for professional legal counsel. For high-stakes contracts, use Guard-Clause as a starting point, then consult a qualified attorney for edge cases.
Privacy and Safety
Is it safe to upload my contract to Guard-Clause?
Yes. All connections use TLS 1.3 (256-bit encryption). Your uploaded file is read into server memory, text is extracted, and the file buffer is discarded immediately. The extracted text is held in an ephemeral cache (Upstash Redis) with a 15-minute time-to-live, keyed by a 256-bit cryptographic token known only to your browser session. No human at Guard-Clause reads, reviews, or accesses your contract text. The text is processed by AI (Claude by Anthropic) only.
Does Guard-Clause store my contract?
The raw uploaded file is never stored. It is read into server memory and discarded after text extraction. The extracted text is held in an ephemeral Redis cache for 15 minutes, then permanently deleted. If you choose "Don't store anything" (the default), no database record is created and everything is gone after the cache expires. If you choose "Save my report," only the AI-generated analysis (risk score, findings, negotiation pack, addendum) is stored in the database. The original contract text is never persisted. You can delete saved reports at any time from your dashboard.
Does Guard-Clause use my contract to train AI models?
No. Guard-Clause uses Claude by Anthropic for analysis. Your contract text is sent to the Anthropic API for processing and is not used to train, fine-tune, or improve any AI model. Anthropic's API usage policy prohibits using API inputs for model training. Guard-Clause does not maintain its own training pipeline and has no mechanism to feed your data into any model's training set.
What It Analyzes
What types of contracts can Guard-Clause analyze?
Guard-Clause analyzes employment agreements, freelance and contractor agreements, NDAs and confidentiality agreements, SaaS terms of service, vendor agreements, partnership agreements, lease and rental agreements, severance agreements, agency contracts, retainer agreements, offer letters, and any other contract in a supported file format. The AI adapts its risk analysis to the contract type and your selected role.
What file formats does Guard-Clause support?
Guard-Clause supports PDF, DOCX (Microsoft Word), and plain text (TXT) files. Maximum file size is 10 MB. PDF and DOCX are recommended because they preserve section headers and page numbers, which improves citation accuracy in the findings.
Understanding Your Results
What does the risk score mean?
The risk score is a number from 0 to 100 that summarizes the overall risk level of your contract. 0 means no significant risks were identified. 100 means severe, deal-breaking risks across multiple clauses. The score is computed from the number and severity of findings. A score of 0 to 30 generally indicates a standard, fair contract. 31 to 60 indicates moderate risks worth negotiating. 61 to 100 indicates significant risks that should be addressed before signing.
What is a finding?
A finding is a specific risk or noteworthy clause identified in your contract. Each finding includes a severity level (critical, high, medium, low, or info), a title (e.g., "Unlimited IP Assignment"), a plain-language summary of the risk, a persona impact statement explaining why it matters specifically to your role, and a citation with the exact excerpt from the contract, the section header, and the page number. Findings are sorted by severity so the most important issues appear first.
What is the negotiation pack?
The negotiation pack is a set of tools for pushing back on unfavorable terms. It includes a ready-to-send email draft requesting specific changes, a phone/video call script with talking points and specific clause references, a list of prioritized asks ranked as must-have, strong, or nice-to-have, and fallback positions for each ask in case the other party pushes back. Every ask is linked to a specific finding so you can reference the exact clause.
What is the addendum builder?
The addendum builder generates replacement clauses for each flagged issue in your contract. For every problematic clause, you get the original excerpt, a suggested replacement clause with improved language, an explanation of why the replacement is better, and a paste-ready addendum note formatted for attachment to the original contract. You can copy the addendum directly into an email or document.
What does persona mean in Guard-Clause?
Persona is the role you select before analysis: Freelancer, Employee, Founder, or Agency. The AI adjusts its risk analysis based on your role because the same clause has different implications depending on which side of the table you sit on. For example, a broad IP assignment clause is a routine ask for an employer but a critical risk for a freelancer whose livelihood depends on reusing their work. Persona selection changes the severity ratings, the persona impact statements, the negotiation strategy, and the replacement clause suggestions.
How It Works
How does Guard-Clause analyze contracts?
Guard-Clause uses Claude (by Anthropic) for AI-powered analysis. When you upload a contract, the text is extracted server-side, sent to the AI with your persona and contract type as context, and the AI reads every clause to identify risks, generate findings with citations, produce a negotiation pack, and build replacement clauses. The AI output is validated against a strict schema (Zod) to ensure structural correctness. If validation fails, the system automatically retries with error feedback to the AI for self-correction. The entire process takes 1 to 2 minutes.
Pricing and Plans
Is there a free version of Guard-Clause?
Yes. Every account gets 1 free scan per month. The free scan includes the risk score, a summary, and the top 5 findings (with truncated excerpts). The negotiation pack, addendum builder, full findings, and PDF export are not included in the free tier. To unlock the complete report, you can purchase a Single Scan for $39 (one-time) or subscribe to Pro for $29 per month (unlimited scans, full reports, PDF export, shareable links, and scan history).
What is the refund policy?
Refund policy varies by plan. Single Scan ($39) is non-refundable once the report is delivered; contact support@guard-clause.com if a technical failure prevents delivery. Pro and Team include a 7-day refund on your first month if no scans have been run. Enterprise refund terms are specified in your enterprise agreement. See Terms of Service for full details.
Ready to Try It?
Upload a contract and see what Guard-Clause finds. Or browse an interactive sample report first.