Not all privacy claims are created equal. Here's how the approaches compare.
Standard AI
Your prompts processed in plaintext on provider servers.
Privacy depends on company policies, not technology.
Susceptible to provider-side access, subpoenas, and breaches.
May be used for model training unless you opt out.
Privacy Proxy
A middleman routes your request to strip identifying info.
Zero-data-retention is a policy promise, not a guarantee.
Your prompts are still processed in plaintext by the AI provider.
You must trust the proxy operator and the AI provider.
Maple's Approach
Hardware-Encrypted AI
Data encrypted on your device before it leaves.
Decrypted only inside a hardware-isolated enclave (TEE).
Your data is not used for model training, advertising, or tracking.
Open-source code and live attestation so you can verify.
For Your Security Team
Technical facts for evaluating Maple's security architecture.
How Maple Protects Your Data
Hardware Isolation. AWS Nitro Enclaves provide CPU and memory isolation, ephemeral storage, and no external network connectivity. External traffic must pass over vsock to a proxy on the parent instance.
Code Integrity. Enclave images are measured at boot. The PCR0 hash uniquely identifies the exact code running inside.
Remote Attestation. A cryptographic attestation document, signed by AWS Nitro hardware, proves the enclave's identity and integrity.
Reproducible Builds. Anyone can build our open-source code from GitHub and compare the resulting hash against the live attestation.
Minimal Trust Model. Trust is concentrated in AWS Nitro hardware and open-source code, with reproducible builds and attestation for independent verification.
Per-User Encryption. Data outside the enclave is stored as encrypted blobs per user. Decryption keys live inside the TEE and on user devices.
Maple Teams brings this end-to-end encrypted, verifiable architecture to company research, writing, coding, and analysis. Use Maple Proxy API to integrate Maple with existing apps and workflows, and MCP to connect thousands of compatible data sources.
A TEE is a hardware-isolated area of a processor that runs code in a secure enclave. Maple uses AWS Nitro Enclaves to isolate AI processing from the parent EC2 instance. Nitro Enclaves have no external network connectivity; external traffic passes over vsock to a proxy on the parent instance. Cryptographic attestation reports measurements of the enclave image and code so they can be verified against the published build.
Is this the same technology Apple uses for iCloud?
Similar concept, different implementation. Apple's Private Cloud Compute uses custom silicon with Secure Enclave. Maple uses AWS Nitro Enclaves with attestation-verified code. Both approaches isolate data processing in hardware and use cryptographic verification of the software environment.
How does cross-device sync work if everything is encrypted?
Your account has its own private key derived from your credentials. Chat history is encrypted with this key before leaving your device and stored in encrypted form on our servers. When you log in on another device, your key is re-derived and used to decrypt your data locally.
Who do I actually have to trust?
Your trust assumptions are minimal and verifiable:
Hardware: AWS Nitro hardware performs as documented (independently audited).
Code: The open-source code running in the enclave does what it says (you can audit it).
Attestation: The cryptographic proof on this page confirms the running code matches the published source.
Can I verify all of this myself?
Yes. Our server code is open source . The attestation document on this page is fetched live from our enclave and verified against AWS's root certificate. You can independently reproduce the build, compare the PCR0 hash, and confirm that the code running in production matches the published source.