AI Privacy Firewall.
Use AI. Keep Your Data.
MyYaad is for anyone who uses ChatGPT, Claude, Gemini — or any AI — with real personal data. It runs on your machine and swaps the sensitive bits for realistic shadows before your prompt is sent. The AI still answers properly. Your real data never leaves your device. Each provider gets different shadows, so even a breach can't be correlated back to you.
No account needed · No cloud · Unlimited & free · Not a trial
Shadow PreviewFollow one prompt.
Out and back.
Your real details stay on your side of the line. Only shadows cross it.
They see shadows.
You hold truth.
Other tools replace your name with [NAME] — the AI gets confused, responses break. MyYaad generates realistic fake data that keeps AI responses useful. And unlike cloud firewalls, nothing leaves your device.
One identity.
A different shadow for every AI.
Each AI gets a different shadow — even if every provider is breached simultaneously, they can't correlate your data across services.
ChatGPT
Ethan Anderson
Vertex-88
salt: f3a2b8f1…
Claude
Sofia Rivera
Halcyon Group
salt: 4e1f8bc2…
Gemini
Marcus Lee
Bluepeak Co
salt: 8d5a3e7f…
DeepSeek
Daniel Brooks
Crestline Ltd
salt: 9c47d2aa…
HMAC-SHA256 · unique salt per provider · shadows cannot be correlated
Four steps.
Zero exposure.
You type normally. MyYaad handles the rest — before anything reaches the AI.
Store Your Data Locally
Add your personal details — name, company, email — to a local vault on your device. The vault is a folder of plain files you can open and inspect. Nothing touches the internet.
Shadows Replace Real Data
Before your prompt reaches any AI, the Shadow Engine swaps real details with realistic fakes. Each provider gets different shadows.
Review Before Send
A pre-send preview shows you every substitution — field by field, real → shadow — before the prompt leaves your device. Confirm or cancel. Nothing reaches the AI you didn’t see first.
You See the Real Response
When the AI replies using shadow names, MyYaad translates them back automatically. Perfect answer, zero exposure.
“Based on your background, Alex_B7K_9...”
→ Restored to “John Mitchell”
Works with the data you
actually share with AI.
MyYaad transforms these data types with realistic, coherent shadows.
Names
Sarah Chen → Person-7f4a3c
Emails
sarah@example.com → usr_k2@fh.io
Companies
Acme Corp → Vertex-88
Addresses
42 Oak Lane → 17 Elm Rd
Financial
£65,000 salary → £58,200
Phone Numbers
+44 7700 900123 → +44 7...
Documents
PDFs, Word, Excel — shadowed as context
Date / date-of-birth detection is in development. For now, add specific dates as vault entries to shadow them by exact match.
Whole documents in.
Only shadows out.
Add PDFs, Word files and spreadsheets, then use several together as combined context. Names, national IDs, SSNs and account numbers inside are swapped for realistic shadows before anything reaches the AI.
PDF · DOCX · XLSX — used together as context · SSNs & national IDs auto-detected
If you use AI to think, write, or work — MyYaad is for you.
Most of our users aren’t lawyers or doctors. They’re people who realised AI is now part of how they get things done, and they’d rather their day-to-day life not end up in someone else’s training data.
Anyone using AI as a thinking partner
You draft emails in ChatGPT. You bounce ideas off Claude. You ask Gemini for advice on a tricky conversation with a friend. Real names come up. Real situations. MyYaad swaps the personal bits for realistic stand-ins before the prompt is sent — so you can keep using AI the way you already do, without leaving a trail.
Founders, freelancers, and self-employed people
Client names. Project codenames. Salary figures. Contract numbers. The exact details that make AI useful are the same details you wouldn’t print on a billboard. MyYaad keeps them on your device while the AI still gets enough context to actually help.
People sharing tricky life stuff with AI
Medical questions. Family situations. Money worries. Difficult emails to write. AI is good at this — but the privacy cost is real. MyYaad means you can ask the question without making your situation part of an AI provider’s permanent record.
Developers using AI tools daily
You paste real code, real config, real names into Copilot, Cursor, ChatGPT. MyYaad runs locally, talks to a daemon on 127.0.0.1:3000, uses per-provider HMAC-SHA256 salting so each AI sees a mathematically different fake for the same input. Everything is inspectable: the vault is plain files, the architecture is documented in /how-it-works.
Regulated professionals
Lawyers, clinicians, teachers, and compliance-heavy roles have specific obligations on top of personal privacy. There are dedicated pages for legal, healthcare, and education — but the underlying tool is the same one everyone else uses.
Deep-dives for compliance-heavy roles.
If your work falls under HIPAA, FERPA, SRA, ABA, or similar — we've written longer pages on exactly how MyYaad fits. The product is the same one everyone else uses; these pages just answer the regulator-shaped questions.
We get it. There are alternatives.
Here's why they fall short.
"Just use a free PII blocker"
Generic [NAME] placeholders confuse the AI — your responses break. No vault means nothing is remembered. No provider isolation means one breach exposes everything.
MyYaad’s realistic shadows keep AI responses useful.
"Route it through a cloud firewall"
Cloud firewalls like ZeroTrusted.ai still send your data through someone else’s server. Their server sees your real data before stripping it.
MyYaad transforms locally — your real data never touches a network.
"Use a private AI chatbot instead"
Private chatbots like Venice.ai or CamoCopy lock you into their models. You lose access to ChatGPT, Claude, Gemini.
MyYaad works with every AI — you pick the best model, we protect the data.
We do not collect, transmit, or store any of your data. There is no MyYaad cloud.
Privacy is not a setting.
It's the architecture.
Other tools say ‘trust us, we don't log.’ Or route your data through EU servers and call it private. MyYaad doesn't ask you to trust anyone — it keeps your real data off the network.
- Local vault — plain files you control (your real data)
- Shadow Engine (all processing)
- Provider salts (OS keychain)
- Pre-send preview — every substitution shown before send
- Document parser + search index
- Shadow values only (fake but realistic)
- No real names, emails, dates, or numbers
- Different shadows per provider
- Cannot be reversed without your device
- Breach-resistant: shadows are useless without your device
What MyYaad protects —
and what it can’t.
Privacy is architecture, not policy — and that cuts both ways. Here’s exactly where the architecture protects you and where it doesn’t, so you use AI with eyes open.
- Your typed prompts. MyYaad swaps sensitive parts for realistic fakes before the prompt leaves your device. This is the channel MyYaad is built to protect.
- Document content you upload as context. Word, PDF and Excel files are parsed locally; shadowed text is what reaches the AI. The original file never leaves your device.
- What the model already remembers about you. If you used ChatGPT, Claude or Gemini before MyYaad (or with MyYaad off), those providers may already hold your real details from past chats — and may use them in replies. MyYaad stops new data from leaking; it can’t retract what a model already stored. Tip: clear the chatbot’s saved memory / custom instructions before relying on MyYaad.
- What a connected tool hands the model directly. If an integration (Gmail, Drive, Calendar, an MCP connector) reads your real data and feeds it to the model, the model sees the real thing through a channel MyYaad never touches. MyYaad protects what you type and upload, not what a tool feeds the model.
Read “MyYaad protects my data” as “my real data isn’t in the prompts I send” — not “this AI can never know anything real about me.” Saying it plainly is the brand.
Frequently asked questions
Is MyYaad just for compliance roles or for everyone?
For everyone. Most MyYaad users aren’t lawyers or clinicians — they’re people who use AI day to day for work, study, life admin, drafting, thinking. The compliance use-case pages exist because some professions have specific obligations on top of personal privacy. The underlying tool is the same.
Does MyYaad work with ChatGPT?
Yes. MyYaad’s Chrome extension supports ChatGPT, Claude, Gemini, DeepSeek, Grok, and Perplexity. Each provider gets a different shadow value for the same real input — so even if every provider were breached simultaneously, your data couldn’t be correlated across services.
Does any of my data leave my device?
No real data leaves your device. The Chrome extension talks only to a local desktop app on your machine via 127.0.0.1:3000 — a loopback address that cannot reach the public internet. What the AI provider receives is a shadow: a realistic but fake value. Your real values stay in a vault folder on your computer.
Is MyYaad really free?
Yes. The core privacy features are free forever — unlimited vault entries, all current and future AI providers, provider isolation, pre-send preview. No account required. We make money from optional sync and enterprise tiers, not from your data.
What does MyYaad NOT protect?
MyYaad protects what you type and the documents you upload. It can’t retract what an AI provider already remembers about you from past chats, and it doesn’t cover data a connected integration (Gmail, Drive, Calendar, an MCP connector) hands the model directly. See the full threat-model boundary.
Is MyYaad HIPAA / FERPA / SRA compliant?
MyYaad is a tool that reduces AI exposure of regulated data — not a compliance certification. Pre-send preview lets you confirm what’s being sent before it leaves your device, and the local-first architecture means real PHI / student records / privileged client data isn’t transmitted to AI providers. Your organisation’s own HIPAA, FERPA, or SRA review still applies. Use-case detail: Healthcare · Education · Legal.
Built in public.
Here’s what’s shipped, what’s in flight, and what’s next — honestly.
Bookmark the Changelog for the running list.
The bigger problem isn’t just privacy. It’s lock-in.
MyYaad’s job today is to keep your real data out of AI prompts. That’s the foundation. But the bigger problem — the one that’s coming for everyone who uses AI seriously — isn’t just privacy. It’s lock-in.
Right now, every AI provider remembers a different version of you. ChatGPT has its memory, Claude has its projects, Gemini has its context. You can’t take any of it with them. If you want to switch — for price, for capability, for principle — you start over. We’re building the user-owned layer that breaks that.
Portable context
Your vault is already on your device. The next step is making it portable: your shadows, your context, your AI-relevant identity bundles, all in a wallet you own and can take across providers. Move from ChatGPT to Claude without losing how the AI knows you. The provider-specific salting we already ship (each AI gets a mathematically different fake) is the technical foundation.
Private AI agents
When AI agents start acting on your behalf — booking, buying, writing, calling APIs — the privacy problem multiplies. Every agent step is another chance to leak. We’re building MyYaad’s shadow engine into an agent middleware layer, so the same locality and substitution guarantees apply when AI is doing things, not just talking.
Developer platform
SDKs (Python, TypeScript), LangChain and LlamaIndex integrations, and an Agent API — so other tools and other agents can plug into MyYaad’s privacy layer instead of building their own. Privacy as infrastructure.
These are on the roadmap, not shipped today. The privacy firewall — what’s protecting your prompts now — is the foundation we’re building this on. Bookmark Changelog as each piece lands.
Unlimited. Free. Forever.
MyYaad is in early access. Core privacy features are free forever. Pro and Enterprise tiers coming soon.
Free
For individuals and teams who value privacy.
- Unlimited vault entries & documents
- All AI providers (current + future)
- All shadow transforms + provider isolation
- Pre-send preview — see every substitution before send
- No account required · No cloud · No limits
Pro
For professionals who sync across devices.
- Everything in Free, plus:
- E2E encrypted cloud sync + multi-device
- Deeper context (cloud RAG + enhanced model)
- Safe rewrite suggestions + email support
- App-level encryption (XChaCha20)
Enterprise
For organisations with compliance needs.
- Everything in Pro, plus:
- Policy engine + admin console + SSO
- Compliance evidence export (PDF/CSV)
- Agent API + SDKs (Python, TypeScript)
- Hardware key encryption + role-based access
Your data. Your rules.
If you use AI, MyYaad is for you. Download it, set up a small vault with the personal details you don't want leaking, and carry on using the AI tools you already use — only now your real data stays on your machine. Free. Local. Ready in under a minute.
macOS 12+ · Windows 10+ (64-bit) · Chrome extension included