← All technologies Draft technology card · TEC-0005
Front
TEC-0005AI development and agent integrationArchitecture

AI agent and agentic workflow

An AI assistant that can take several steps and use approved tools, instead of only giving one answer. Its permissions must be tightly controlled.

Where it creates value

Best suited to work like this.

Research workflows, controlled data gathering, service-desk assistance, document preparation, and orchestrated administrative work.

What it does

The intelligent task.

Plan multi-step work, call tools, retrieve context, maintain task state, request approval, and adapt actions based on results.

Typical value

Coordinates multi-step knowledge work and tool use, reducing manual handoffs while preserving defined review and approval points.

Where it performs well: Can coordinate multiple systems and reduce manual handoffs across a multi-step task.

Value maturity · editorial assessment Demonstrated

Agentic workflows are showing value in bounded, supervised processes, while dependable autonomy, controls, and operating models are still developing.

  1. Emerging
  2. Demonstrated
  3. Scaling
  4. Established
Back
TEC-0005 · How it works in practice

AI agent and agentic workflow

Read the card type together with the use case. A component can be central in one solution and enabling in another.

01 · Inputs

Goal, instructions, tool catalog, permissions, context, intermediate results, and approval rules.

02 · Intelligent task

Plan multi-step work, call tools, retrieve context, maintain task state, request approval, and adapt actions based on results.

03 · Outputs

Completed actions, tool calls, drafts, structured results, audit logs, or requests for human decisions.

Common application patterns
  1. Research workflows
  2. controlled data gathering
  3. service-desk assistance
  4. document preparation
  5. and orchestrated administrative work
Enablers

What must be in place.

Bounded goals, least-privilege tools, explicit approval points, robust error handling, observability, cost limits, and safe stopping behavior.

Where it struggles

Limitations and drawbacks.

Errors can compound across steps; poorly bounded agents may take unintended actions, expose data, loop, or create difficult-to-explain results.

Human role

People remain accountable.

People authorize sensitive actions, supervise exceptions, review outcomes, and can pause or override the workflow.

Continue exploring

The card is the simple front door. The website can hold the full system behind it.

Use the technology record to connect applications, processes, enablers, risks, controls, evidence, and implementation experience without placing all of it on the printable card.

Browse technologies