Workshops Use Cases Architecture Projects Blog
About Contact Offers Talks
Home›AI technologies›AI agentsDeutsch

AI agents: models that call tools, running in isolation

An agent is a model plus the tools it may call. An agent runtime runs it, keeps sessions apart, gives each agent an identity and memory, and records what it did.

Amazon Bedrock AgentCoreFoundry Agent ServiceVertex AI Agent EngineStrands Agents, LangGraph or CrewAI
Book a free 30-min call →Strands Agents Workshop →

What are AI agents?

Agents are built with frameworks such as Strands Agents, LangGraph or CrewAI. The framework decides how the agent thinks and which tools it calls. The runtime decides where it runs, what it may reach, and how one session is kept apart from the next.

Several agents are an answer to several jobs, not to one hard one. When one prompt would do, one prompt is the better design. And two agents should never share one identity, for the same reason two colleagues should not share one login.

Goes inA task, and the agent code you want run
Comes outThe result, an isolated session, its memory and a record
Typical usesMulti-step work with tools: research, planning, editing, operations

Four ways to run it

Each of the big clouds offers it as a managed service, and you can also run it yourself. Same task, a different bill and a different console.

Amazon Bedrock AgentCore
Managed on AWS

Amazon Bedrock AgentCore[1]

The managed agent runtime on AWS: runtime, memory, identity and a tool gateway for agents built with any framework and any model.

Official documentation →
Foundry Agent Service
Managed on Microsoft Azure

Foundry Agent Service[2]

The agent service inside Microsoft Foundry.

Official documentation →
Vertex AI Agent Engine
Managed on Google Cloud

Vertex AI Agent Engine[3]

The managed agent runtime on Google Cloud.

Official documentation →
Strands Agents, LangGraph or CrewAI
Run it yourself

Strands Agents, LangGraph or CrewAI[4]

Open-source agent frameworks you can run on any container platform you operate. Strands Agents (Apache 2.0) comes from AWS; LangGraph and CrewAI are MIT.

Project page →
Amazon Bedrock AgentCore: The managed service on one page, with the same service on the other platforms.
The managed service on one page, with the same service on the other platforms.[5]

How it looks in an architecture

Slides from my course on AI architectures at FH Burgenland, drawn from real projects.

A CrewAI crew on Bedrock AgentCore, each agent with one job, and six validation layers around them.
A CrewAI crew on Bedrock AgentCore, each agent with one job, and six validation layers around them.[5]
Six agents, six jobs, and a different model behind three of them.
Six agents, six jobs, and a different model behind three of them.[5]
One question to the event assistant, traced from the message to the answer.
One question to the event assistant, traced from the message to the answer.[5]

Where I used it

The video pipeline

In my video pipeline, a CrewAI crew of eight agents on Bedrock AgentCore decides the edit, each agent with one job. Six validation layers sit around them, because agents that pick clips will happily pick a clip that does not exist. The eight agents cost about one cent per video.

Read the write-up →

The event assistant

In my event assistant, six agents each have one job, one set of tools and their own identity. In another multi-agent system of mine the agents shared one identity, and credentials ended up in plain text in the run logs. Since then: one identity per agent, isolation per session.

See the reference projects →

The geospatial platform

In my geospatial platform, agents run on both sides of the line and exchange a question and an answer, never records. That is what keeps the private data in the building.

Architecture on ai-solutions.wiki →

Managed and self-hosted together: hybrid

Agents make hybrid possible without moving data: an agent in the cloud and an agent inside your building each work with their own data and exchange questions and answers. The records themselves never travel. The question to answer first is which data class may be seen where.[6]

The geospatial platform: two arrows cross the line, and both carry a question or an answer.
The geospatial platform: two arrows cross the line, and both carry a question or an answer.[5]

What would help you next?

Learn to decide

The course “From Managed to Hybrid” teaches where AI should run: what each layer costs, what you operate and where your data has to stay. It is in preparation, with a free live session first.

Join the course waitlist →
Linda MohamedLinda Mohamed
AI and cloud consultant, AWS Community Hero, Vienna

Build it with me

Tell me what you want to build. In 30 minutes we check which of the four ways fits your data, your rules and your budget, and whether AWS funding for a proof of concept can cover part of it. How AWS funding works →

Book a free 30-min call →15-min question slot

You already know what you need

Book the workshop directly:

ai-solutions.wiki

Read more on ai-solutions.wiki

My open knowledge base, with longer comparisons, pricing notes and sources.

Sources

  1. Amazon Bedrock AgentCore documentation. https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/what-is-bedrock-agentcore.html
  2. Foundry Agent Service documentation. https://learn.microsoft.com/en-us/azure/foundry/agents/overview
  3. Vertex AI Agent Engine documentation. https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/overview
  4. Strands Agents, LangGraph or CrewAI. https://strandsagents.com/
  5. Linda Mohamed, course slides “The projects”, MASE-3, FH Burgenland, September 2026
  6. GIS and AI architecture on AWS, ai-solutions.wiki. https://ai-solutions.wiki/solutions/geospatial/gis-ai-architecture/
  7. AgentCore vs Bedrock Agents, ai-solutions.wiki. https://ai-solutions.wiki/comparisons/agentcore-vs-bedrock-agents/
  8. Strands Agents, ai-solutions.wiki. https://ai-solutions.wiki/tools/strands-agents/
  9. CrewAI, ai-solutions.wiki. https://ai-solutions.wiki/tools/crewai/
  10. CrewAI vs LangGraph, ai-solutions.wiki. https://ai-solutions.wiki/comparisons/crewai-vs-langgraph/
  11. How I built an AI film crew on AWS, lindamohamed.com. https://www.lindamohamed.com/blog/how-I-built-an-AI-film-crew-on-aws

Links checked September 2026. Product names and features change: check the linked pages before you decide.