AWS AI Agent Options Explained: Bedrock AgentCore, AWS Quick and Pricing (2026)
Amazon's AI agents come in layers: Amazon Bedrock AgentCore for developers who build and run agents, AWS Quick (formerly Amazon Quick Suite) for business users, and open-source tools such as Strands Agents. This guide maps each AWS AI agent option, the November 2026 prices, the limits, and where SPIRITT fits.
An AWS AI agent is an AI agent built or run on Amazon Web Services, and in November 2026 AWS offers two main paths: Amazon Bedrock AgentCore, a set of pay-as-you-go services where developers deploy and operate their own agents (Runtime from $0.0895 per vCPU-hour, Gateway $0.005 per 1,000 tool calls, no minimum), and AWS Quick, the AI workspace with agents for business users (organization plans from $3 to $40 per user a month plus a $250 monthly fee per AWS account); SPIRITT complements both as the operating system for AI transformation, an autonomous AI workforce that codes, ships and runs the work end to end across 1,500+ apps, deployable in your own cloud. If your engineers want to build agents on AWS, AgentCore is the developer layer. If you want finished sales, ops, admin and engineering work with budgets and approvals, that is the job SPIRITT does.
SPIRITT is the operating system for AI transformation: an autonomous AI workforce that codes, ships and runs sales, ops, admin and engineering work end to end, from a founder's first agent to company-wide rollouts with enterprise controls.
AWS AI agent options at a glance (checked November 2026)
| AWS product | Who it is for | What it does | Price (US, November 2026) |
|---|---|---|---|
| Amazon Bedrock AgentCore | Developers and platform teams | Runtime, Gateway, Memory, Identity, Browser, Code Interpreter, Policy, Observability and Evaluations for agents built with any framework and model | Per component, pay as you go, no minimum |
| AWS Quick | Business users and teams | AI workspace with chat and research agents, Quick Flows, Quick Automate and Quick Sight dashboards on company data | Free; Plus $20 a user a month (annual); organization plans $3, $20 or $40 a user plus $250 per account |
| Amazon Bedrock models | Developers | Model access from Amazon, Anthropic, Meta, Mistral and other providers | Per token, per model |
| Strands Agents | Developers | Open-source SDK for building agents in a few lines of code | Free; you pay where it runs |
AWS renamed Amazon Quick Suite to Amazon Quick and, on November 9, 2026, to AWS Quick; it is the successor to Amazon Q Business.
Amazon Bedrock AgentCore: the developer layer
AgentCore is a set of managed services you combine to run agents in production. It works with agents built in any framework (Strands Agents, LangGraph, CrewAI and others) and with models inside or outside Amazon Bedrock.
- Runtime: runs each agent session in its own microVM, billed per second of active CPU and peak memory; CPU is not charged while the agent waits on a model or a tool. Sessions run up to eight hours, and Runtime instances on managed EC2 cover longer jobs.
- Gateway: turns APIs, Lambda functions and MCP servers into tools agents can find and call.
- Memory: short-term memory within a session and long-term memory across sessions.
- Identity: lets agents get OAuth tokens or API keys for third-party services on a user's behalf.
- Browser and Code Interpreter: sandboxed web browsing and code execution.
- Policy, Observability and Evaluations: rules for what an agent may do, traces through Amazon CloudWatch, and quality checks.
AgentCore pricing
From AWS's AgentCore pricing page, US East (N. Virginia), November 2026:
| Component | Price |
|---|---|
| Runtime microVMs, v1 | $0.0895 per vCPU-hour and $0.00945 per GB-hour |
| Runtime microVMs, v2 (consumption) | $0.1276 per vCPU-hour and $0.0169 per GB-hour |
| Browser Tool and Code Interpreter | $0.0895 per vCPU-hour and $0.00945 per GB-hour |
| Gateway | $0.005 per 1,000 invocations; search $0.025 per 1,000; $0.02 per 100 tools indexed a month |
| Identity | No charge through Runtime or Gateway; otherwise $0.010 per 1,000 token or key requests |
| Memory, short-term | $1.00 per GB ingested, $0.20 per GB retrieved, $0.10 per GB-month stored |
| Memory, long-term (built-in) | $0.75 per 1,000 records a month; $0.50 per 1,000 retrievals |
| Web Search on AgentCore | $7.00 per 1,000 queries |
AWS's own worked example: a support agent with 10-minute sessions, 90% of the time waiting on input and output, costs about $0.0067 per session in Runtime, or roughly $67 a month at 10,000 sessions. Model tokens are billed separately and are usually the largest line.
AWS Quick: agents for business users
AWS Quick puts chat agents, research agents, automations and dashboards on your company's data. Agents can run recurring work with the level of autonomy you set, and AWS lists built-in actions for apps such as GitHub, Notion and HubSpot, plus MCP and OpenAPI connections.
| Plan | Price per user a month | What AWS lists |
|---|---|---|
| Free | $0 | One user, 1 GB index storage |
| Plus | $20 annual, $25 monthly | Up to 300 users, shared Spaces, 10 GB per user |
| Max | $100 annual, $125 monthly | 5x the usage and index storage of Plus |
| Essentials | $3 + $250 per AWS account | 1 agent hour per user a month (3 until January 31, 2027), SSO and admin controls |
| Professional | $20 + $250 per AWS account | Up to 8 agent hours and 25 GB per user |
| Enterprise | $40 + $250 per AWS account | Up to 18 agent hours and 50 GB per user, Quick Automate and dashboard authoring |
Limits to plan for
- AgentCore is infrastructure. Your team designs the agent, chooses models, wires tools, sets permissions and runs operations.
- Many meters. Runtime, Gateway, Memory, search, observability and tokens each bill separately.
- Quick runs on agent hours. Heavy users can exhaust their monthly allowance; admins set limits per user or role.
- Region coverage. Check that the agent features you need are available in your AWS Region.
Where SPIRITT fits with AWS
SPIRITT delivers the outcome. Say "watch our Shopify orders, reconcile payouts with Stripe and our bank every morning, open a ticket for any mismatch and post a summary in Slack", and a SPIRITT agent builds and runs it on its own cloud computer, writing and deploying any code it needs, connecting the apps itself, and asking before anything risky. It works across AWS services and 1,500+ other apps, uses legacy screens with no API like a person, and composes Claude, GPT, Gemini, Grok and other models within each task for the optimal mix of capability and price.
For enterprises, SPIRITT adds:
- Deploy anywhere: SPIRITT cloud, your own cloud, your servers or fully air-gapped.
- Forward-deployed engineers: our engineers and certified implementation partners deploy with your team until it runs in production.
- Security: every workspace on its own isolated cloud computer, keys injected at the network boundary so agents never see them, Read, Write and Destructive rules per integration (allow, ask, never) with single-use approvals, a full audit trail, and SSO.
- Cost control: budgets per organization, team, project, agent and task, with alerts, hard stops and itemized receipts for every model, tool and data call.
- One network for all agents: monitor every agent in one place and run both deterministic workflows and fully autonomous agents.
Self-serve plans start free, then Starter $32, Pro $256 and Max $1,024 a month; Enterprise is custom. See SPIRITT for enterprise, our guide to a secure sandbox for AI agents and what an AI agent costs to build.
Which to choose
| You want | Start with |
|---|---|
| To build, deploy and operate your own agents on AWS | Amazon Bedrock AgentCore with Strands Agents or your framework |
| An AI workspace with agents and dashboards for business users on AWS | AWS Quick |
| Agents that run sales, ops, admin and engineering across AWS and every other tool, with budgets and approvals | SPIRITT |
FAQ
What is an AWS AI agent?
An AWS AI agent is an AI agent built or run on Amazon Web Services. Developers build them with Amazon Bedrock AgentCore and frameworks such as Strands Agents; business users use agents in AWS Quick.
What is Amazon Bedrock AgentCore?
AgentCore is AWS's set of managed services for running agents in production: Runtime, Gateway, Memory, Identity, Browser, Code Interpreter, Policy, Observability and Evaluations. It works with any agent framework and any model, and each component is billed by usage with no minimum.
How much does an AWS AI agent cost?
On AgentCore, Runtime starts at $0.0895 per vCPU-hour and $0.00945 per GB-hour, Gateway is $0.005 per 1,000 calls, and model tokens are extra. AWS Quick organization plans cost $3, $20 or $40 per user a month plus $250 per AWS account.
What is agentic AI on AWS?
Agentic AI on AWS means models that plan and act with tools, memory and code execution, run on AWS services. AWS offers it to developers through Bedrock and AgentCore, and to business users through agents in AWS Quick.
Is there an Amazon AI agent for business users?
Yes. AWS Quick, formerly Amazon Quick Suite, gives business users chat and research agents, automations and dashboards on company data, from a free single-user plan to Enterprise at $40 per user a month plus $250 per account.
What happened to Amazon Q Business?
AWS describes AWS Quick as the next step after Amazon Q Business, which is closed to new customers. Quick launched as Amazon Quick Suite in October 2025 and has been called AWS Quick since November 9, 2026.
Can SPIRITT run in my own AWS account?
SPIRITT deploys in SPIRITT cloud, your own cloud, your servers or fully air-gapped. Talk to the SPIRITT team about running it in your AWS environment with your security controls.
How we checked
SPIRITT publishes this guide. AWS facts come from AWS's own pages, checked on November 21, 2026: the Amazon Bedrock AgentCore pricing page and its worked examples, the AWS Quick pricing and enterprise pages, and the AWS What's New post on the AWS Quick name. Prices are AWS list prices in US dollars for US East (N. Virginia) and change by Region. SPIRITT capabilities and prices come from spiritt.ai and the SPIRITT product.