Multi-tenant by design
Organizations, projects, and environments isolate each tenant's agents, data, and credentials, whether tenants are your departments or your customers.
The platform
The platform is Deployments.ai. It's a source-available, multi-tenant AI agent platform. Run it in an AWS account in your AWS Organization, or use it hosted and managed by us in our AWS account.
Dev account
v14 immutable · versioned
Agents, tools, and policies
Evaluation suite compares v14 with v13 42 cases
Advisory. Review before you promote.
Prod account
Canary sends a share of runs to v14
v13 still serving the rest
Promote v14 to prod?
Promoted by the release approver
v14 serving · v13 kept for rollback
Every run is traced step by step, with metrics.
Illustrative run with sample data.
Bring your own cloud
It runs where your data already lives
You create a dedicated AWS account in your AWS Organization and give us scoped access. We provision the platform there with OpenTofu. Agent runtime, workflow state, memory, traces, and audit logs stay in that account, under your service control policies, your IAM policies, and your KMS keys. Agents reach your systems only through tools you allow, and reach models only through providers you approve. Read how BYOC works for AI agents, or how the platform compares with Amazon Bedrock AgentCore.
- Built on AWS services your team already knows: Lambda, ECS on Fargate, DynamoDB, S3, KMS, EventBridge, and SQS.
- Separate AWS accounts for dev and prod, with releases promoted between them.
- A model gateway with per-environment policies, using your own provider keys.
- Your OpenTofu state, your logs, and infrastructure on your consolidated AWS bill.
What's inside
Everything agents need to do real work, safely
Agents as code
Agents, tools, and policies are defined in code and shipped as immutable, versioned releases, with canary rollout and promotion from dev to prod — or rollback.
Human approvals
Mark any tool call as approval-required. The run pauses, asks the right person, and resumes on their answer. Why we design approvals in.
Durable workflows and memory
Long-running, resumable runs with schedules, so work survives restarts and can wait on people. Memory gives agents the context they need, scoped to their work.
Tools and integrations
Connect systems through MCP servers, HTTP callbacks, and the integration catalog, plus sandboxed code with no network access.
API, SDK, and CLI
Build on the platform through its public API, SDK, and CLI. Client-executed tools let agents call functions inside your own app.
Traces and evaluations
Every run is traced step by step, with metrics. Evaluation suites compare each new release with the one it would replace, as an advisory check before you promote it.
Governance and metering
Retention policies, tenant export, erasure, signed audit-log exports, and usage metering for each tenant.
Channels
Agents meet people in Slack, in email from your domain, on Telegram and RCS business messaging, or inside your product through the API.
The platform is Deployments.ai. We also run it as a hosted, managed service in our AWS account. The platform's source release on GitHub is coming soon.
White-label
Your brand, wherever your people meet an agent
A white-label agent platform runs under your brand instead of the vendor's: your name and domain, and agent identities your people recognize. For internal agents, we configure the platform under your name and domain as part of the engagement. Agents post in Slack with the names and icons you choose and send email from your domain. Putting the platform in front of your own customers under your brand is offered under a commercial license.
- Your name and domain
- Agent identities in Slack
- Agent email from your domain
- Agents inside your own product
Source available
Public source, coming soon
The platform source is coming soon to the Agentifiers/deployments repository under the Functional Source License (FSL-1.1-ALv2), a Fair Source license, and each release becomes Apache 2.0 two years after it is published. Once it's public, your engineers can read every line that touches your agents, extend it with your own tools, and run it for your own business without us.
- Readable, auditable code
- Extend with your own tools
- No single-vendor dependency
- Hand over to your team any time
Run it your way
Self-hosted, set up for you, or hosted and managed by us
Self-host Deployments.ai for free for your own use once the source is released, have us set it up in an AWS account in your AWS Organization, or use the hosted Deployments.ai service, which we run and manage in our AWS account with the same API. Billing is invoice only: we invoice our fees and hosted Deployments.ai access, while in your own AWS account AWS bills your infrastructure directly and model providers bill usage on your own keys. See pricing for each option.
- Self-hosted: free for your own use, coming soon
- Set up by Agentifiers: fixed scope, in your AWS account
- Hosted Deployments.ai: managed by us, no infrastructure to run
- Invoice billing, no infrastructure markup
See it running on your own work
Bring us a process that eats your team's week, or the agent feature you want to ship to your customers. We'll show you what it looks like on the platform, under your brand.