Audit and control what AI does with your data and tools

Aptible's AI control layer enforces access controls, audit logging, and credential management for LLM usage and MCP tool calls. Your team can use any model, connect any tool, and deploy any agent without creating security gaps.

Audit and control what AI does with your data and tools

Aptible's AI control layer enforces access controls, audit logging, and credential management for LLM usage and MCP tool calls. Your team can use any model, connect any tool, and deploy any agent without creating security gaps.

Audit and control what AI does with your data and tools

Aptible's AI control layer enforces access controls, audit logging, and credential management for LLM usage and MCP tool calls. Your team can use any model, connect any tool, and deploy any agent without creating security gaps.

Where the ai risk shows up

Teams ship AI faster than they can secure it.

By the time someone asks what your agent sent to a model, or which tools an agent called last week, the answer requires reconstructing logs from multiple places if it even exists at all. Here are the governance gaps we see most often.

Getting BAAs with AI providers is slow, fragile, and doesn't scale

Teams end up locked to one model because it's the only BAA they could get. At some companies, every new AI vendor requires annual approval from health insurance and pharma partners. Teams work around it by avoiding models they need or staying in a compliance gray area.

AI is in production but the audit trail is DIY or missing

Engineers integrate models through provider APIs with whatever logging they had time to build. Customer security questionnaires now include AI sections, requiring reconstructing logs from multiple places (if they exist at all).

Every Claude Desktop user has a different MCP configuration

Developers add servers on their own, manage their own credentials, and run local processes nobody else can see. Designers could have the same access to Snowflake as engineers, with no way to differentiate.

Agents are using their creator's credentials

Agents, service accounts, and workload identities already outnumber human users by roughly 45-to-1. Most have no security controls of their own. An agent running on a human's identity is invisible in the audit log: its calls look like the engineer's calls, and there's no way to trace what it did or what arguments it passed.

Tool access is all-or-nothing, and uncontrolled MCP is a supply chain risk

Claude Desktop and Claude Code give a user everything a server exposes or nothing. There's no tool-level scoping. Any MCP server with shell or credential access is tier-0 supply chain: a changed tool definition affects everyone connecting through it with no visibility.

Why the risks stay hidden

AI tooling is outpacing AI security controls

Why the risks stay hidden

AI tooling is outpacing AI security controls

Why the risks stay hidden

AI tooling is outpacing AI security controls

MCP was designed for individuals, not teams

The protocol connects a single AI client to tools. It wasn't designed to enforce who can call which tools, manage credentials across users, or produce a queryable audit log. Most teams are running at the ceiling of what native tooling supports.

MCP was designed for individuals, not teams

The protocol connects a single AI client to tools. It wasn't designed to enforce who can call which tools, manage credentials across users, or produce a queryable audit log. Most teams are running at the ceiling of what native tooling supports.

Agents inherit whatever their creators can access

There's no concept of a robot identity in most agent setups. An agent running on an engineer's session uses that engineer's credentials and grants. Scaling from one agent to five makes attribution worse, not better.

Agents inherit whatever their creators can access

There's no concept of a robot identity in most agent setups. An agent running on an engineer's session uses that engineer's credentials and grants. Scaling from one agent to five makes attribution worse, not better.

"Don't put PHI in ChatGPT" isn't a security policy

A real security layer requires specifics: which models are approved for which data classifications, which agents have production write access, what logging is required. For most teams, that infrastructure doesn't exist and the organization is relying on trust.

"Don't put PHI in ChatGPT" isn't a security policy

A real security layer requires specifics: which models are approved for which data classifications, which agents have production write access, what logging is required. For most teams, that infrastructure doesn't exist and the organization is relying on trust.

how aptible works

One control layer to manage what AI can see and do

Aptible AI Gateway protects sensitive data before it reaches a model. Aptible MCP Gateway controls which tools your team and agents can call and logs every action. Together they reduce the risk AI introduces to your stack.

Aptible AI Gateway protects sensitive data before it reaches a model. Aptible MCP Gateway controls which tools your team and agents can call and logs every action. Together they reduce the risk AI introduces to your stack.

request beta access

BAA Coverage
Audit Logging
Access Control
Agent Identity
Team Deployment

One BAA covers every LLM provider

AI Gateway's single BAA covers all usage across supported providers. Switching models or adding providers doesn't require a new compliance review or vendor approval cycle. You get one contract, one audit artifact, one place to point when customers ask.

seamlessly switch models under a single baa

  • One BAA covers every LLM provider

    AI Gateway's single BAA covers all usage across supported providers. Switching models or adding providers doesn't require a new compliance review or vendor approval cycle. You get one contract, one audit artifact, one place to point when customers ask.

    seamlessly switch models under a single baa

  • Every LLM request and tool call, logged with full attribution

    AI Gateway logs every prompt and response with user identity, model, scope, and timestamp. MCP Gateway logs every tool call with user identity, tool name, server, arguments, and IP address. Arguments are encrypted at rest. When a security review asks what your AI did, the record is already there.

    date range

    last 7 days

    server

    All

    tool

    All

    user

    All

    Occurred at

    Jan 17, 2026

    13:11:43 UTC

    Jan 17, 2026

    13:09:13 UTC

    Jan 17, 2026

    13:07:49 UTC

    Jan 17, 2026

    13:05:22 UTC

    user

    qualification-agent

    Robot

    Sally G.

    sally.green@acme.com

    Jane D.

    jane.doe@acme.com

    qualification-agent

    Robot

    server

    notion

    github

    github

    notion

    Tool

    notion_notion-update-view

    create_pull_request

    create_branch

    notion_notion-update-view

    args

    2 args

    1 arg

    1 arg

    2 arg

    user agent

    claude-code/2.1.141

    claude-code/2.1.141

    claude-code/2.1.141

    claude-code/2.1.141

  • Scoped keys for LLMs, tool grants for teams and agents

    AI Gateway scopes LLM access by team and environment. MCP Gateway assigns tool access by role, so the same Notion server can give engineers read access and leads write access. PHI de-identification coming soon.

    Access Grants

    Search...

    role

    Deploy Owners

    Robots (No PHI)

    Robots (PHI Access)

    Account Owners

    servers

    github

    notion

    sentry

    + 3 more

    github

    notion

    github

    notion

    sentry

    + 2 more

    github

    notion

    sentry

    + 5 more

    allowed tools

    23/47

    12/20

    18/31

    55/58

  • Agents get their own identity, not a borrowed one

    An agent running on a human's credentials is invisible in the audit log: its calls look like the engineer's calls. Robot users give agents their own API keys and role-based grants, separate from any human.

    ← Robot Users

    qualification-agent

    API keys

    id

    0301g87e0-8c67-123b-1er5-d6cc1b22c33b

    created

    Jan 17, 2026

    13:15:52 UTC

    Role Memberships

    role

    CRM Team

    Send Email

    added

    Jan 17, 2026

    13:11:43 UTC

    Feb 11, 2026

    12:31:09 UTC

    allowed tools

    23/47

    4/20

    servers

    notion

    sentry

    pylon

  • One connection for your whole team, auto-configured

    Instead of each engineer maintaining their own MCP configuration, the team connects once to the gateway. Claude Desktop and Claude Code can be auto-configured via MDM profiles so every team member gets access to the right tools without individual setup.

BAA Coverage
Audit Logging
Access Control
Agent Identity
Team Deployment

One BAA covers every LLM provider

AI Gateway's single BAA covers all usage across supported providers. Switching models or adding providers doesn't require a new compliance review or vendor approval cycle. You get one contract, one audit artifact, one place to point when customers ask.

seamlessly switch models under a single baa

For teams building with PHI

AI Gateway is designed for regulated workloads. BAA coverage, PHI de-identification, and seven-year audit log retention are enforced by default.


MCP Gateway adds the agent layer: tool-level access control for agents handling PHI, with audit logging that satisfies the same auditability requirements as any other PHI access event.

AI Gateway is designed for regulated workloads. BAA coverage, PHI de-identification, and seven-year audit log retention are enforced by default.

MCP Gateway adds the agent layer: tool-level access control for agents handling PHI, with audit logging that satisfies the same auditability requirements as any other PHI access event.

AI Gateway is designed for regulated workloads. BAA coverage, PHI de-identification, and seven-year audit log retention are enforced by default.

MCP Gateway adds the agent layer: tool-level access control for agents handling PHI, with audit logging that satisfies the same auditability requirements as any other PHI access event.

Learn more about HIPAA-Compliant AI

Access Grants

Access Grants

role

CRM Team

Robots (No PHI)

Robots (PHI Access)

Account Owners

servers

notion

sentry

+ 3 more

github

notion

github

pylon

sentry

+ 2 more

github

pylon

sentry

+ 5 more

allowed tools

23/47

12/20

18/31

55/58

Access Grant Details

Role

*

CRM Team

Tools Available (20)

Search...

github

0/24

notion

8/17

All tools (wildcard *)

notion_notion-create-comment

notion_notion-create-database

notion_notion-create-pages

notion_notion-create-view

notion_notion-duplicate-page

notion_notion-fetch

notion_notion-get-comments

notion_notion-get-teams

notion_notion-get-users

notion_notion-move-pages

notion_notion-query-database-view

notion_notion-query-meeting-notes

notion_notion-search

notion_notion-update-data-source

notion_notion-update-page

notion_notion-update-view

sentry

12/12

shortcut

0/20

Use Cases

How teams use Aptible's AI control layer

Use Cases

How teams use Aptible's AI control layer

Ship AI features that touch sensitive data without building the security layer from scratch

Route LLM traffic through AI Gateway and get audit logging, de-identification, and BAA coverage enforced automatically.

Expand LLM usage beyond one provider without new BAAs

Teams locked to Bedrock or a single provider because it's the only BAA they have can access all supported models through AI Gateway's single BAA. Switching models or testing new providers stays a product decision, not a compliance event.

Answer customer security questionnaires with actual evidence

Customer security reviews now include AI sections. Pull audit records for LLM usage and agent tool calls directly from the platform, without a reconstruction effort.

Secure developer tooling, not just production features

Engineers using Claude Code with internal tools connected through MCP are part of the same risk surface as production features. MCP Gateway applies the same access controls and audit logging to developer workflows automatically.

Separate PHI and non-PHI contexts as your team and agent count grow

As teams scale, the same agent or engineer may need different tool access depending on the task. Assign distinct permission profiles for workflows that touch regulated data vs. those that don't, enforced at the gateway rather than relying on the agent to self-limit.

aptible vs no AI security

What changes when AI security is enforced at the infrastructure layer

DIy on aws

LLM audit logging

Every request and response logged automatically

Build and maintain your own pipeline, or nothing

MCP tool audit logging

Every tool call logged with user identity and arguments

No record outside local client logs

BAA coverage

One BAA covers all LLM usage

Separate BAA per provider, no MCP coverage

PHI protection

De-identification before model requests (coming soon)

Manual implementation per application

Tool-level access controls

Per-role tool grants, enforced at the gateway

Server-level only

Credential management

Shared and personal credentials centrally managed

Individual tokens in individual config files

Agent identity

Robot users with scoped grants, separate audit trail

Agents use human credentials, invisible in logs

Team deployment

Single connection, MDM-deployable

Each engineer configures their own setup

Cost controls

Hard limits per scope that stop requests

Billing alerts after the fact

Evidence on demand

Available immediately from the platform

Requires reconstruction across multiple systems

Keep shipping. Safety happens automatically.

Deploy in minutes.

Keep shipping. Safety happens automatically.

Deploy in minutes.