Use Bedrock’s managed agent orchestration when the model must interpret an open-ended goal, gather missing details, choose among approved tools, query a knowledge base and replan. Use Lambda-centered code or AWS Step Functions when the workflow needs fixed sequencing, explicit retries, approvals, waits, idempotency or tightly controlled business rules. They are complementary rather than interchangeable: a Bedrock agent can decide which tool to use, while Lambda executes that tool. For a new implementation, also account for product lifecycle: Amazon Bedrock Agents Classic is no longer open to new customers, so evaluate AgentCore or a custom architecture before committing to the Classic runtime.
What each option actually does
Amazon Bedrock Agents Classic
Bedrock Agents Classic is a managed runtime for model-driven orchestration. It can interpret a request, split the goal into steps, ask the user for missing information, invoke APIs through action groups, query a knowledge base and use the result to plan the next step. AWS manages the prompt stages and provides features for memory, monitoring, encryption, permissions and API invocation.
As an Amazon Associate I earn from qualifying purchases.
The important distinction is that the agent is not the business API itself. The model predicts an action; Bedrock packages the selected operation and parameters; your fulfillment code performs the operation and returns a result that can feed another orchestration step.
Recommended Free Tools
Lambda-first orchestration
A Lambda-first design puts your application code in charge. You decide when to call a model, how to represent state, which tools are available, how arguments are validated, which identity is authorized, how failures are retried and whether the workflow ends or continues. This gives you a contract-controlled state engine rather than a managed planner.
#1 Best Overall
- Valued Carpenter Pencil Set: You will get 2 pcs solid carpenter pencils with 26 piece 2.8 mm refills, 1 replaceable sharpener, 1 plastic storage box.The complete carpenter pencils combination allows you to finish your work faster and more easily
- Deep Hole Marker Pencil: The deep-hole construction pencils adopts 45mm elongated tip design, which is more convenient to mark in the small hole or in other tight areas that other carpenter markers cannot reach
- Carpenter Pencils with Sharpener: The sharpener is screwed into the top of the work pencil, which won't get lost either. Built-in pencil sharpener that keep the lead with pointed and smooth to Improves line of sight in fine work
- Stronger Solid Lead: This work pencil is matched with a 2.8 mm thick lead , which is much thicker and stronger during the drawing process of construction work, it will not break or damage easily
- Marks on Various Surfaces: 3 colors solid construction pencil can marks on various surfaces,such as metal, plastic, wood, paper etc. Ideals for woodworkers, contractors, craftsmen, builders, merchants and masons
AWS’s custom-orchestrator pattern describes state transitions using events such as START, MODEL_INVOKED and TOOL_INVOKED. The orchestrator can return outcomes including INVOKE_MODEL, INVOKE_TOOL and FINISHED. You can implement a different state contract, but the principle is the same: every transition is explicit and testable.
Where Step Functions fits
Step Functions is a workflow control plane, not a conversational agent. It is usually the better fit when a transaction has a known order, must pause for a callback or approval, needs durable retries and error branches, or can run longer than a single function invocation. Lambda tasks can remain the workers inside that state machine.
Bedrock Agents vs. Lambda: the decision at a glance
| Decision axis | Bedrock Agents Classic | Lambda-centered orchestration | Step Functions with Lambda tasks |
|---|---|---|---|
| Autonomy | Model selects actions, gathers parameters and can replan. | Your code selects tools and transitions. | Transitions follow a defined workflow graph. |
| Control | Managed prompt and orchestration stages with configurable tools and permissions. | Full control over state, prompts, validation and stop conditions. | Explicit branching, retries, catches, waits and callbacks. |
| Tool integration | Action groups use an OpenAPI schema or function details and a Lambda fulfillment function, or can return control to the caller. | You define the model-tool contract and invoke any service your code can reach. | Each task invokes a service or function in a predetermined sequence. |
| Long-running behavior | Suitable for conversational, multi-step planning; transaction timing still depends on the invoked services. | A Lambda function has a 15-minute maximum timeout. | Designed for waits, callbacks, retries and error paths that should outlive a function call. |
| Governance and observability | Managed traces, permissions, encryption, monitoring and optional guardrails. | You assemble logging, traces, policy checks and guardrails. | Workflow history and service integrations complement your application-level controls. |
| Lifecycle | Amazon Bedrock Agents Classic is no longer open to new customers. | Runs on generally available Lambda and your chosen model integration. | Uses Step Functions and Lambda without depending on the Classic agent runtime. |
Runtime quotas and pricing are not universal properties of an architecture. Bedrock limits are model- and Region-dependent, with token-per-minute and sometimes request-per-minute quotas; account-specific values can differ and adjustable quotas require a request. A real cost model must include model tokens, Lambda memory and duration, the number of orchestration turns and any workflow services.
When a managed agent is the right shape
Open-ended goals
Choose managed orchestration when users describe outcomes rather than procedures: “Find the delayed orders, explain the likely cause and draft a customer response.” The agent can decide which approved operation to call, ask for an order identifier if it is missing, consult a knowledge base and use intermediate results to select a next step.
Many optional tools
An agent is useful when several safe read or low-risk write operations are available but no single fixed sequence applies to every request. The model can select among those operations while your action-group contract limits what it is allowed to call.
Conversation is part of the task
If collecting parameters conversationally is a core requirement, managed stages and memory reduce the amount of session plumbing you must build. You still need to decide what information can be retained and which operations require a fresh authorization check.
Rank #2
- Ergonomically Designed: Work in tight areas with a compact design that gets into tough spots
- Both tools are designed to fit into difficult to reach spaces
- Complete a wide range of applications with a high-performance motor that delivers up to 300 UWO, and an impact driver delivering up to 1,400 in-lbs of torque
- Both the DEWALT impact driver and drill driver feature integrated LED work lights with a convenient 20-second delay, ensuring enhanced visibility in dimly lit or challenging work areas
- One-Handed Loading - Keep one hand free with a 1/4 in. hex chuck that accepts 1 in. bit tips
Knowledge retrieval plus action
Bedrock can combine knowledge-base queries with action-group calls. For example, it can retrieve a policy, ask for a missing account number, then call a permitted API. This is different from a retrieval-only chatbot because the answer can drive an operation.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchHow Bedrock action groups connect tools
An action group is the contract between the agent and an executable tool. You provide either an OpenAPI schema or function details, plus a Lambda fulfillment function. One action group can contain up to 11 API operations and uses one Lambda function. That function must inspect the operation-specific event and produce the response shape expected by the agent runtime.
- Define narrow operations. Describe each API operation with explicit parameters, types, required fields and useful descriptions. Prefer separate read, preview and commit operations over a single function that accepts arbitrary commands.
- Attach the fulfillment path. Configure the action group to invoke one Lambda function, or choose the return-control pattern when the caller—not Bedrock—must execute the operation.
- Dispatch by operation. In the handler, identify the requested operation, validate every argument against the schema and map it to a specific internal client. Do not let a model-generated string become an unrestricted URL, query or command.
- Authorize at execution time. Check the caller, tenant, resource and operation inside the handler. An action-group schema describes what can be requested; it does not replace authorization or business policy.
- Return structured results. Include a stable success shape and an explicit error shape. Return only the information the next orchestration step needs, and avoid leaking credentials or internal stack traces.
- Test each operation independently. Exercise missing parameters, wrong types, unauthorized resources, downstream timeouts and duplicate requests before allowing the agent to call the group.
Keep fulfillment handlers small. The agent may call the same operation more than once while it plans, retries or corrects parameters, so handlers should be idempotent where possible and should record an operation or request identifier for deduplication.
When Lambda-first code is the better choice
Deterministic transactions
Use application code when a transaction must always follow a known order, such as validating an account, reserving inventory, charging a payment method and writing an audit record. A model can help interpret the initial request, but it should not be allowed to reorder irreversible steps.
Strict contracts and provider choice
A custom orchestrator lets you normalize several model providers behind one interface, enforce a house prompt format, or change models without changing the tool contract. You can require a typed tool call, reject anything that fails validation and choose a fallback model according to your own policy.
Custom guardrail sequencing
If policy checks must occur before retrieval, after tool output and again before a side effect, explicit code makes those checkpoints visible. You can also stop after a fixed number of model turns or require human approval for a particular outcome.
Rank #3
- 【Great Compatibility】This Katerk 1/4 inch hex shank bit holder is specifically designed for 1/4 inch hex shank drill bits. It's compatible with most 1/4 fast hex handles, hex sockets, various electric screwdrivers, and handheld screwdrivers. The bit holder makes it a valuable addition for any handyman.
- 【Secure and Safe】Built with a secure backup nut design, each drill bit holder securely locks onto your bits, ensuring they stay firmly in place. Additionally, our bit holder incorporates a high-quality steel ball rolling design that holds up to several kilograms of weight, ensuring your various drill bits don't fall off.
- 【Easy One-Handed Operation】The bit holder for impact driver allows you to change bits single-handedly, simplifying your workflow. Its multi-color design further allows for quick identification of the drill bit you need.
- 【Compact and Convenient】Thanks to its compact size, this 1/4 inch bit holder is easy to carry around. The bit holder allows for easy attachment to various tools, making this a convenient addition to your construction accessories. The Katerk bit holder is cast from high-quality alloy material, promising a long product lifespan. Despite its rugged strength, the bit holder remains lightweight, making it portable.
- 【Cool Christmas Gift For Men Stocking Stuffers】 This screwdriver bit holder, driver bit holder, impact bit holder, can be given as a gift to your loved one, especially for anyone involved in construction or electrical work. It's a must-have for stocking stuffers for men and women, tools gifts for dad, tech gadgets for men, gifts for dad, gifts for him, gifts for husband, gifts for boyfriend, cool gadgets for men, and cool gifts for dad.
Testing and replay
Because transitions are represented in your code, unit tests can feed a known state and assert the exact next outcome. Persisted state can support replay of a failed run, deterministic simulation of tool errors and regression tests for prompt or model changes.
When Step Functions should replace an agent loop
Use Step Functions when the workflow has operational behavior that should not depend on a model continuing to respond. Typical signals include:
- A wait for a person, an external callback or a scheduled time.
- Retries with different delays for different failure classes.
- Compensation or rollback after a later task fails.
- Parallel tasks whose completion must be joined before continuing.
- Execution that can exceed Lambda’s 15-minute function limit.
- An audit requirement for every transition, approval and terminal state.
A useful division of labor is to let a model classify the user’s goal and select one of several approved workflow types. Once selected, Step Functions executes the transaction with fixed states, while Lambda performs individual tasks. The model does not receive permission to skip an approval state or invent a compensation path.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →A practical hybrid architecture
The hybrid pattern handles ambiguity at the boundary and control in the transaction:
- Interpret. A conversational agent extracts the user’s goal and asks for missing non-sensitive parameters.
- Constrain. The agent can choose only from tools that return a safe, typed plan or a workflow name. Exclude irreversible operations from free-form selection when possible.
- Authorize. Lambda or the workflow service checks identity, tenant, resource ownership, limits and approval requirements independently of the model.
- Execute. Step Functions or application code performs the fixed sequence, with idempotency keys, explicit retries and compensation.
- Report. The executor returns a concise status and machine-readable details. The agent may turn that status into a natural-language explanation, but it must not claim success unless the executor reports it.
This arrangement preserves a conversational experience without making a probabilistic planner the control plane for money movement, account changes, deletion or other irreversible effects.
Reliability, retries and failure handling
Agent-driven failure modes
A managed agent can choose an unsuitable tool, omit a parameter, misunderstand an error or repeat an operation while replanning. Schemas with precise descriptions, bounded enums and clear error responses reduce these risks, but they do not eliminate the need for execution-time checks.
Rank #4
- Long Nib and Deep Hole Marker: Our mechanical carpenter pencil with 45mm nib is designed for easy marking of deep holes or narrow areas. These construction pencils are the great choice for woodworking tools, construction tools, carpenter tools, contractor tools, wood carpentry tools and architect tools
- Extra Refills in 2 Colors for Versatile Marking: The construction mechanical pencil comes with 12 extra 2.8mm refills, including 6 red and 6 black refills. The black refill is suitable for light surfaces, while the red wax is perfect for dark surfaces. Our carpenter mechanical pencil makes sure that you'll have an ample supply for extended use
- Built-in Sharpener: Our construction pencil comes with a built-in sharpener to ensure the mechanical pencil tip is always sharp and ready for use. Never buy an extra pencil sharpener again. A great tool for any woodworker pencil, contractor pencils. The refill can easily be extended or retracted with a simple click of the pencils mechanical, allowing you to work more efficiently and accurately
- Portable Clip Design: Our deep hole construction pencil features a portable clip design, easy to carry and attach to your pocket or tool box, so that you can keep the carpenter pencils mechanical close at hand, making it a convenient tool to have on the go. Great gifts choice for carpenters
- Stronger Pencil Lead: The black refills are made of lead, sturdy and smooth. The red refills are made of wax, clear and light. These marking pencils are much thicker and stronger than normal pencils during the marking process of construction work, suitable for various surfaces, such as glasses, metal, boards, floors, walls, furniture, etc. The written marks can be easily wiped with a wet paper towel when needed
Lambda failure modes
Lambda adds ordinary distributed-system concerns: cold starts, downstream throttling, transient network errors, duplicate delivery and partial completion. Use timeouts that are shorter than the surrounding workflow, classify retryable errors, and make writes idempotent. Do not hide a permanent validation error behind automatic retries.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Workflow-level recovery
For a multi-step transaction, record which side effects have completed and define compensation for each reversible step. A Step Functions catch or retry branch can then act on a known state rather than asking a model what to do next. For a custom orchestrator, persist the state represented by START, MODEL_INVOKED and TOOL_INVOKED transitions so a worker can resume safely.
Security and governance differences
Bedrock’s managed runtime supplies facilities for permissions, encryption, monitoring, prompt stages and optional guardrails. You still have to scope the IAM permissions of the agent and fulfillment function, protect sensitive data in prompts and tool responses, and decide which actions require confirmation.
In a Lambda-first design, equivalent controls are your responsibility. Centralize authorization rather than trusting a model-produced role or account identifier. Validate resource ownership, redact secrets from logs, set network egress deliberately and preserve an audit record of the model request, selected tool, authorization decision and result. Apply the same controls to tools called through an agent and tools called directly by application code.
Cost, quotas and performance expectations
There is no universal winner on cost, speed or accuracy. Bedrock usage depends on the selected model, input and output tokens, the number of planning turns and Region quotas. Lambda cost depends on memory size, duration and invocation count. A hybrid workflow adds orchestration and state-service charges, while a custom loop may spend more engineering time and testing effort.
Free tools Windows power users keep installed
One-click scans. No signup required.
Measure the topology you actually deploy: average and worst-case model turns, tool-call count, token volume, Lambda duration, retry rate and the percentage of requests requiring human intervention. Bedrock runtime quotas are per model and Region; check the account’s assigned values and request an adjustment when the documented process allows it. Do not infer throughput from a different model, Region or account.
Best Value
- Milwaukee Ink all Fine Point Marker, Black, 4 Per Pack
- 4 per pack Features Clog Resistant Marker Tip Writes through Dusty, Wet and Oily Surfaces Durable Marker Tip for Writing on Concrete, OSB and Rough Surfaces
- Clog resistant tip writes on dusty, wet and oily surfaces and is optimized for rough surfaces such as OSB, cinderblock and concrete
- Hard hat clip- attaches for easy access
- Quick dry time with reduced smearing and marking
Lifecycle: Classic agents and AgentCore
AWS documentation states that “Amazon Bedrock Agents (now Amazon Bedrock Agents Classic) is no longer open to new customers.” Existing users may have operational reasons to maintain a Classic deployment, but a new customer should not treat it as the default long-term foundation.
AWS’s current platform direction, AgentCore, provides a serverless agent runtime along with memory, observability and tool gateways. Its developer guide describes consumption-based pricing with no upfront commitment or minimum fee; regional prices and service availability still need to be checked for the deployment location. If AgentCore does not provide the control or provider abstraction you need, a Lambda and Step Functions architecture remains a viable alternative.
Migration questions to answer
- Which Classic features are essential: action groups, knowledge bases, memory, traces, guardrails or return control?
- Can each tool be represented as a stable, independently authorized operation?
- Which conversations require durable memory, and what data-retention policy applies?
- Which side effects must move behind a deterministic workflow?
- What telemetry and replay capability is required before cutover?
Keep the tool contract separate from the agent runtime so that replacing Classic with AgentCore or a custom orchestrator does not require rewriting every backend integration.
Common implementation mistakes
Putting business logic in the prompt
Prompts can describe intent, but they are not a substitute for authorization, transaction boundaries or validation. Enforce those rules in Lambda or the workflow service.
Making one giant tool
A broad “execute anything” operation is difficult to secure and test. Split capabilities by resource and risk, describe required parameters precisely and expose preview operations before commit operations.
Assuming a successful model response means success
Only the downstream executor can establish whether an API call completed. Return an explicit status and have the conversational layer report failure when the executor reports failure.
Ignoring duplicate calls
Retries, timeouts and replanning can produce repeats. Use idempotency keys and check whether a side effect already occurred before applying it again.
Choosing Lambda for a waiting workflow
Do not keep a function open while waiting for a human or external callback. Move the wait into Step Functions or another durable mechanism, then invoke Lambda when the event arrives.
A decision checklist
- Choose managed agent orchestration when the request is open-ended, tools are optional, clarification is conversational and the result can be safely constrained.
- Choose Lambda-first orchestration when you need a custom state contract, model abstraction, exact guardrail order or deterministic tool selection.
- Choose Step Functions when waits, callbacks, approvals, retries, compensation or long execution are first-class requirements.
- Choose a hybrid when a model should interpret intent but code must authorize and execute the transaction.
- For a new Bedrock deployment, evaluate AgentCore or a custom design instead of starting with Agents Classic, which is closed to new customers.
Bottom line
Bedrock Agents and Lambda solve different parts of a tool-using workflow. Bedrock supplies managed, model-driven planning; Lambda supplies controlled execution. Step Functions supplies durable, explicit workflow semantics. Start by classifying the workload: let an agent handle ambiguity and tool choice, but put authorization, idempotency, irreversible effects and long-running control in code or a workflow engine. That boundary produces a system that is conversational where it should be and deterministic where it must be.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
