SavvaiGuard

Provides proof your written AI governance policy is enforced. Audits every interaction by client across ChatGPT, Claude, and Gemini

You're already using AI for client or company work. Can you prove exactly how it was used? SavvaiGuard gives anyone using AI for work — advisors, consultants, analysts, researchers — a complete, auditable record of every AI interaction. Every prompt reviewed, every decision logged, every interaction attributed to the right client — that's how SavvaiGuard proves your AI governance policy is being followed. You decide what gets sent to the AI. SavvaiGuard documents every decision automatically. Without SavvaiGuard, every AI interaction involving client or company data happens with no record — nothing to show, nothing to export, nothing to defend. When a client, manager, insurer, or regulator asks how you handled their data, you need more than a good answer. You need proof. SavvaiGuard installs in under a minute, sets up in a few clicks, and works with the AI tools you already use. No IT support required, no enterprise procurement, no integration project. ADDITIONAL FEATURES FOR MULTI-CLIENT WORK If you manage multiple clients — as a fractional CFO, financial advisor, consultant, or any professional with a multi-client book — SavvaiGuard has additional features specifically for you. Open a new chat for each client — the way you already work. Before you start, select the client in the SavvaiGuard panel. Every interaction in that session is automatically attributed to that client across ChatGPT, Claude, and Gemini. Whenever you need it, generate a per-client compliance report for any date range in one click. - Select a client in the SavvaiGuard panel — every interaction in that session is logged to them - Work across multiple AI tools — one audit trail per client regardless of AI platform - One-click compliance reports per client — PDF or CSV - No manual sorting — SavvaiGuard organizes every interaction by client automatically - Export a full AI governance record for any client, any date range, on demand No spreadsheets. No manual logging. No chasing down which AI interaction belonged to which client. Just a complete, per-client record of how AI was used — organized automatically and ready whenever you need it. HOW IT WORKS SavvaiGuard reviews every prompt before it reaches the AI. When sensitive data is detected, the review panel opens alongside your AI platform and shows you what was found. You decide: - Send Sanitized — submit with sensitive data replaced by tokens. Logged. - Send Original — submit unchanged as a documented, intentional override. Logged. - Cancel — discard entirely. Still logged. Every outcome is recorded equally — what was detected, what decision was made, and when. A governed override is documented with the same level of detail as a sanitized submission. What matters is that every interaction is reviewed and documented. WHAT GETS DETECTED Detection runs across four layers: - Google Cloud DLP — enterprise ML for names, emails, SSNs, EINs, account numbers, and 200+ patterns - Advisory pattern detection — engagement IDs, matter numbers, client codes, invoice references - Client registry matching — registered client company names flagged automatically across every interaction - Engagement-specific identifiers — advisory-specific financial and business identifiers configurable per client TWO MODES — BOTH FULLY LOGGED Analysis Mode (default) Client identifiers are replaced with tokens before submission. Financial data and analytical context are preserved. Full interaction logged. Research Mode The company name passes through to the AI — useful when the AI needs external knowledge about a specific company. All other sensitive data is still detected and handled. Every Research mode interaction is logged with mode noted in the audit trail. Which mode you use is your decision — based on your firm's AI governance policy. SavvaiGuard enforces and documents whichever you choose, consistently, across every interaction. To help you get started, SavvaiGuard includes a free AI Governance Policy Template. It gives you a practical starting framework for governed AI use — so you can begin using AI responsibly from day one. DOCUMENT UPLOAD PROTECTION SavvaiGuard also reviews document uploads before they reach the AI — whether attached through the file picker or drag-and-drop. Common business document and spreadsheet formats are supported, so the documents you already share with AI tools for client work are scanned the same way as your prompts. When PII is detected in a document, you choose how to proceed: - Use Sanitized Text — the document text is tokenized inline before reaching the AI. Logged. - Send Original File — the original file is sent as a documented, intentional override. Logged. - Exclude Document — the document is removed from the submission. The prompt still goes through. Logged. A PII report is downloadable on demand for offline review and manual sanitization. COMPLIANCE DOCUMENTATION Every interaction — prompt or document, sanitized or overridden, Analysis or Research — is logged to a permanent audit trail including: - AI platform (ChatGPT, Claude, or Gemini) - Date and timestamp - Sensitive data types detected - Advisor decision — sanitized, override, document blocked, or cancelled - Mode — Analysis or Research - Client attribution From the SavvaiGuard portal (guard.savvai.ai), generate per-client compliance reports across all platforms, export to PDF or CSV, and produce documentation supporting responsible AI use — on demand, for any client, any date range. KEY FEATURES - Complete, per-client audit trail of every AI interaction - One-click compliance reports per client — PDF or CSV - Reviews prompts before submission across ChatGPT, Claude, and Gemini - Instant client switching — SavvaiGuard tracks context automatically - 4-Layer Detection Engine — enterprise-grade DLP - Analysis and Research modes — both fully logged, both configurable - Document upload review — file picker and drag-and-drop - Session-based client tagging with cross-platform coverage - Privacy-focused — SavvaiGuard records detection results and advisor decisions, not full prompt or document content in the audit trail - Sign in and start — no IT support required, no complex configuration, no procurement process WHO IT'S FOR SavvaiGuard is for anyone using AI for work involving sensitive data: - Fractional CFOs and financial advisors managing multiple client portfolios - Management consultants and business advisors - Accountants, CPAs, and tax professionals - Analysts and researchers working with company or client data - Any professional who needs a documented, defensible record of AI use If you manage multiple clients, SavvaiGuard is especially powerful — every interaction is automatically attributed, every client report is one click away. PRIVACY AND SECURITY SavvaiGuard records detection results and advisor decisions — not full prompt or document content. Detection is performed using a combination of in-browser processing and secure backend services. Only detection metadata and advisor decisions are stored in the audit trail. PLANS Free — Up to 50 scans per month, 2 registered clients, detection panel and audit trail viewing. Sanitization, exports, and compliance reports require Advisor. Advisor — $19/month. Unlimited scans, unlimited clients, full sanitization, per-client compliance reports, and export to PDF or CSV. Start free — up to 50 scans per month. Upgrade to Advisor anytime for unlimited usage. FREQUENTLY ASKED QUESTIONS Does SavvaiGuard store my prompts or documents? No. SavvaiGuard does not store full prompt or document content in its audit trail. Detection may use secure backend processing, but only detection metadata and advisor decisions are retained. Is SavvaiGuard a SOC 2 or HIPAA solution? SavvaiGuard is AI governance infrastructure, not a certification. It helps you implement AI governance controls and evidence that align with frameworks like SOC 2 and NIST AI RMF by providing a real-time audit trail and per-client reports, but it does not itself make your firm compliant. Do I have to sanitize everything for SavvaiGuard to be useful? No. SavvaiGuard documents every AI interaction involving client work, regardless of whether you send a sanitized or original prompt. A governed override is logged with the same detail as a sanitized submission — what matters is that every interaction is reviewed and documented. What is the difference between Analysis and Research mode? In Analysis mode, SavvaiGuard replaces client identifiers (including the company name) with tokens before submission while preserving financial data and analytical context. In Research mode, the company name can pass through when you need external knowledge about a specific company; all other PII is still detected and handled. Both modes are fully logged in the audit trail. Who should use SavvaiGuard? SavvaiGuard is built for fractional CFOs, financial advisors, accountants, and consultants who use AI tools like ChatGPT, Claude, or Gemini in client work and need a provable audit trail and per-client compliance reports. Can I use SavvaiGuard on the free plans of ChatGPT, Claude, or Gemini? SavvaiGuard works technically with free and paid AI accounts, but we recommend using business-grade plans for client work because free consumer accounts may use your content to train models and retain data longer. Does SavvaiGuard work with tools other than ChatGPT, Claude, and Gemini? SavvaiGuard currently supports ChatGPT, Claude, and Gemini in Chrome. All interactions across supported platforms feed into a single per-client audit trail — included at no extra charge. Additional AI surfaces are under evaluation for future releases. SavvaiGuard is developed by Savvai Inc. Support: supportrequest@savvai.ai Portal: guard.savvai.ai Terms: savvai.ai/terms-of-service