본문으로 건너뛰기
This page is not translated yet. 한국어 is switched on so the navigation and URLs can be tested; the text below is the English original. Use the language selector in the header to switch back, or see the documentation home.
concept · Reviewed 2026-11-02 GA

Agentic Engineering agents

An agent handles one task and hands back a result. Agentic Engineering includes the full set, covering the work around a simulation from checking an input file before submission through to producing the deck that reports on a finished campaign.

Because Agentic Engineering is licensed in place of Advanced Modeling & Simulation, every agent below is available to you, including the three that come with that tier.

How to run one​

Type / in Rescale Edison to list the agents your workspace has, pick one, and give it a prompt or parameters. If you describe the task in plain language instead, the Orchestrator routes it to the right agent, or to a chain of them, and returns the result.

You can also open the Agent Library in the left navigation to run one directly, or add one as a node in Workflow Builder so it runs as part of a pipeline without anyone watching.

Anything that would change something asks you first. Applying a fix, resubmitting a job or editing a budget comes back for your approval rather than happening quietly.

Before you submit​

Job Input Validator​

Reads an input file and reports what would stop the job running, with line numbers and suggested corrections, so a run that would have failed never starts. Available as a control during job submission or as a workflow node.

Hardware Recommender​

Recommends hardware for a given solver and workload. The choice of coretype is what makes a run fast or slow and cheap or expensive, and it is often made from habit. See hardware for the same capability outside chat.

When a run fails​

Job Failure Summarizer​

Reads a failed job's logs and explains in plain language what went wrong and where, naming the file and line rather than returning a category. It reasons over the log text, so it handles solvers it has not seen before.

Job Troubleshooter​

Takes a failed job through to a running one: diagnoses the problem, proposes a fix, applies it with your approval, and resubmits, repeating if the next attempt also fails. Configured as a workflow node, it can pick up a failure and work it without anyone watching.

Finding what you need to know​

Knowledge Navigator​

Performs multi-leg retrieval across your connected data and returns answers with verified citations. When the evidence does not support an answer it says so rather than producing one. Other agents call it when they need to consult a manual or a specification. See knowledge bases.

Reporting on the work​

Jobs Report Generator​

Turns a completed simulation campaign into a PowerPoint deck with findings, charts and recommendations. Organizations can supply their own template, so the output arrives in the format the review expects.

Finance UI Agent​

Reports on billing, budgets, credits and deposits, broken down by workspace, project or member, which answers an end-of-month cost question without anybody exporting a CSV. See billing and reports.

Budget Control​

Monitors compute budgets, explains why spend changed, and flags reallocation opportunities before the quarter ends. It can also create and update budgets from a description. See how budgets work.

Making work repeatable​

Workflow Builder​

Builds a multi-step workflow with you rather than leaving you to assemble it in the editor. See workflows for what a workflow is and how a run behaves.

Agentic Project Builder​

Sets up an Agentic Project, which is how project context and knowledge carry across sessions and between the people on a team instead of living in one person's chat history.

Agents work together​

Agents call one another through the Orchestrator rather than duplicating work. Job Troubleshooter uses Job Failure Summarizer to read the logs before proposing a fix, and several agents call Knowledge Navigator when they need to consult documentation. One request can therefore involve more than one agent, and you do not sequence them yourself.

Solver coverage varies

Job Input Validator and Job Troubleshooter depend on authored skills for the solver in question, so coverage reflects what your organization has configured. If an agent does not recognize your solver it says so and prompts you to author a skill, which then applies for everyone in your organization.