Self-hosted AI agent platform that runs entirely inside your AWS account: agents, sandboxes, workflows and integrations, with Amazon Bedrock as the model provider.
Brainbase - Agent Platform deploys the complete Brainbase stack into your own VPC with one CloudFormation template: the agent runtime and scheduler, isolated per-agent sandboxes on AWS Lambda MicroVMs, the web application, an LLM gateway that calls Amazon Bedrock with your account's credentials, Supabase-based authentication, storage and realtime services, and the GitHub, Slack, orchestration and proxy MCP servers. Model calls, your database, secrets, storage, logs and the MCP servers all run inside your account. Three optional features use Brainbase-operated services (workflows, the GitHub connector, and connector agent streaming), and a few optional integrations stay off until you add your own third-party credentials; both are listed under External dependencies in the usage instructions. You pay AWS directly for the compute, storage and Bedrock usage your agents consume, and Brainbase for the platform.
The template creates every resource it needs, generates its own secrets, applies the database schema and builds the sandbox image in-account. Upgrades are a stack update against a new release version; the schema ledger applies only what your database is missing.
Highlights
Runs in your account: VPC-private services, your own Postgres, your Bedrock quota, MCP servers included.
One-template install on your domain; upgrades are a stack update with a schema ledger.
Agent sandboxes on Lambda MicroVMs built in-account; GitHub, Slack and MCP integrations included.
AWS Marketplace now accepts line of credit payments through the PNC Vendor Finance program. This program is available to select AWS customers in the US, excluding NV, NC, ND, TN, & VT.
Pricing is based on actual usage, with charges varying according to how much you consume. Subscriptions have no end date and may be canceled any time. Alternatively, you can pay upfront for a contract, which typically covers your anticipated usage for the contract duration. Any usage beyond contract will incur additional usage-based costs.
Additional AWS infrastructure costs may apply. Use the AWS Pricing Calculator to estimate your infrastructure costs.
This product uses two usage-based pricing dimensions that work together. You pay a platform charge for each appliance-hour the deployment runs. With an annual contract, that platform rate stays fixed for 12 months. The second dimension charges a platform fee for each $1 of estimated model token spend. AWS bills you separately for Bedrock model usage and underlying infrastructure. The two dimensions are independent: one tracks how long the platform runs, the other scales with your model activity. Your total varies with both runtime hours and model consumption.
Top-of-mind questions for buyers
What counts as one appliance-hour for the platform charge?
An appliance-hour measures the time your deployed Brainbase appliance runs in your AWS account. The installation runs as Fargate services, a database, and related resources inside your account. You are billed for each hour that running appliance stays active, independent of model activity.
How do the platform charge and the model token spend charge combine on my bill?
The two dimensions bill independently. The appliance-hour charge tracks runtime. The model token spend charge applies per $1 of estimated model cost. They add together on your invoice. Continuous runtime drives the first; heavy agent and model activity drives the second.
Does the model token spend charge include my actual Bedrock and infrastructure costs?
No. The model token spend dimension is a Brainbase platform charge calculated per $1 of estimated model cost. AWS bills Bedrock model usage and underlying infrastructure separately. Those AWS charges run through your own account, not through this listing.
docs.brainbaselabs.com
Helpful?
Vendor refund policy
No refunds. Pay-as-you-go charges stop when the subscription is cancelled.
How can we make this page better?
Tell us how we can improve this page, or report an issue with this product.
Give us feedbackReport a problem with this product or seller
Legal
Vendor terms and conditions
Upon subscribing to this product, you must acknowledge and agree to the terms and conditions outlined in the vendor's End User License Agreement (EULA).
Content disclaimer
Vendors are responsible for their product descriptions and other product content. AWS does not warrant that vendors' product descriptions or other product content are accurate, complete, reliable, current, or error-free.
Containers are lightweight, portable execution environments that wrap server application software in a filesystem that includes everything it needs to run. Container applications run on supported container runtimes and orchestration services, such as Amazon Elastic Container Service (Amazon ECS) or Amazon Elastic Kubernetes Service (Amazon EKS). Both eliminate the need for you to install and operate your own container orchestration software by managing and scheduling containers on a scalable cluster of virtual machines.
Version release notes
Release 2026-09-16.1 (source commit a988971b7e8c586d2e5ec7b0a7017602f06d73c3). Adds sender-default SES bounce/complaint suppression, a dedicated SNS feedback topic and reputation alarms. Confirm the feedback subscription and assign an operator; production access still requires AWS approval. Image and schema pins are unchanged. Includes 16 container images. Deploy in us-east-1 with https://brainbase-marketplace-templates.s3.us-east-1.amazonaws.com/templates/2026-09-16.1/brainbase-appliance.template.json (sha256 d79552390eee2d5dcb7df08b94b530908ad2b45a477497e2d06baa8ef675acef).
Additional details
Usage instructions
Before you start. Create-resource permissions, a public Route 53 zone, Bedrock and SES production access for unrestricted email. us-east-1 only.
Deploy. Subscribe; create a stack from the linked template: DomainName, HostedZoneId, AlarmEmail, optional AppAllowedCidrs (allowed IPv4 ranges) and FirstUserEmail. ~15 min. Creates app., api., llm., supabase., hooks. records plus one per integration, and SES sender/DKIM. FirstUserEmail adds an account; public signup stays closed. Open https://app.<DomainName> and request a magic link once SES is ready. Confirm SNS alarm/feedback emails; monitor bounces/complaints. SES sandbox recipient restrictions still apply.
What the template creates and why. A VPC with public/private subnets; ALB/ACM; 15 Fargate services; one m7g.large with PostgreSQL on encrypted EBS; nine install Lambdas; ECR/CodeBuild image clone; two S3 buckets; secrets, logs, alarms.
IAM roles and their purpose. Scoped task roles: app/billing use aws-marketplace:MeterUsage, billing also aws-marketplace:SearchAgreements; LLM gateway calls Bedrock; scheduler/app manage MicroVMs; storage uses its bucket. Execution roles pull images/read secrets; Lambda/CodeBuild access created resources and the registry. No role has administrative access.
Keys and secrets. Install generates credentials in Secrets Manager (bb-appliance/supabase-shared, bb-appliance/app-shared); none is hard-coded or sent to Brainbase. Rotate with affected services, then redeploy.
Encryption. EBS, S3 and Secrets Manager use AWS-managed KMS at rest; public endpoints use ACM TLS; internal traffic stays in the VPC.
Backup and recovery. PostgreSQL has dedicated EBS with nightly snapshots (7 kept). Restore: create a volume from a snapshot, replace the database instance's data-volume mapping, restart; ask support for details. Stack deletion retains the volume, snapshots and both secrets; delete when done.
Health monitoring. Alarms email AlarmEmail. Check ECS for running = desired and https://api.<DomainName>/healthz.
Quotas. Check Fargate vCPU, Lambda concurrency and Bedrock throughput. Check applied Lambda Max allocated MicroVM memory (L-CD1C0CC4) in us-east-1: 8 GB per RUNNING or SUSPENDED VM plus headroom. New accounts may allow only one VM. Request needed increases in Service Quotas before agent use; contact AWS Support if rejected.
Costs. Pay AWS for one m7g.large, 15 Fargate tasks, a NAT gateway, an ALB, EBS/S3 storage, Bedrock tokens and sandbox compute as used; pay Brainbase for listed dimensions. Included services cost no extra.
External dependencies and data flows. At install, the stack downloads crane from github.com and pulls Marketplace-registry images. Brainbase runs three external services, is responsible for their availability/security, and the product is withdrawn from AWS Marketplace if they are discontinued: workflows (workflows.workforce.brainbaselabs.com), the GitHub connector (brainbase-github-connectors.onrender.com) and connector agent streaming (kafka-llm-service.onrender.com). That data transits Brainbase only while in use. Optional integrations (Pipedream, Portkey, Twilio, LiveKit, Deepgram, Resend) stay off until credentialed; that data then flows to the provider under your terms. Slack differs: SLACK_ENABLED plus placeholder credentials only boot the connector; no data flows until you install a real Slack app. No key is issued for analytics host us.i.posthog.com, so no analytics are sent. OAuth issuer app.brainbaselabs.com is a token claim only; no traffic goes to it. With the shipped configuration, everything not named in this section stays in your account.
Upgrade. Update the stack with the new template URL: the schema ledger applies pending migrations, images update, the data volume survives. On a schema-step failure, --disable-rollback keeps the database.
Email support at help@brainbaselabs.com, 9am-7pm Pacific Time, Monday-Friday; first response within one business day. No availability SLA: the appliance runs on infrastructure you operate.
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
MaiAgent is an enterprise-grade AI agent platform that enables businesses to build, deploy, and manage intelligent chatbots across multiple channels. Powered by leading LLMs (including GPT-4 and Claude), MaiAgent helps teams automate customer support, internal Q&A, and business workflows 24/7, with no coding required.
Noca.ai is an AI Agent platform that lets users build business apps, workflows or "AI Agents" using plain language prompts, no programming required. It offers natural-language UI & logic, integrations with CRM and ERP systems, workflow automation, and enterprise-grade security.
Cyara Agentic CX Trust Platform (formerly Automated CX Assurance Platform) is the independent trust layer for customer experience. Test every AI agent, journey and load profile before launch. Monitor every path to your agents 24/7. Observe every live conversation for quality and compliance. Proof before launch. Vigilance in production.
Be the first to review this product. We've partnered with PeerSpot to gather customer feedback. You can share your experience by writing or recording a review, or scheduling a call with a PeerSpot analyst.