Agentic AI Intro

An agent possesses core characteristics that allow it to act reliably and consistently on behalf of a user.

Characteristics of Agentic AI

• Leverages an LLM to manage workflow execution and make decisions
  • recognizes when a workflow is complete and can proactively correct its actions if needed
  • In case of failure, it can halt execution and transfer control back to the user.
• It has access to various tools to interact with external systems—both to gather context and to take actions
  • dynamically selects the appropriate tools depending on the workflow’s current state, always operating within clearly defined guardrails.

When to use Agentic AI

Suited to workflows where traditional deterministic and rule-based approaches fall short.

Example: payment fraud analysis

A traditional rules engine works like a checklist, flagging transactions based on preset criteria.

LLM agent functions more like a seasoned investigator, evaluating context, considering subtle patterns, and identifying suspicious activity even when clear-cut rules aren’t violated.

This nuanced reasoning capability is exactly what enables agents to manage complex, ambiguous situations effectively.

Where to use Agentic AI

Prioritize workflows that have previously resisted automation.

Complex decision-making

Workflows involving nuanced judgment, exceptions, or context-sensitive decisions, for example refund approval in customer service workflows.

Difficult-to-maintain rules

Systems that have become unwieldy due to extensive and intricate rulesets, making updates costly or error-prone, for example performing vendor security reviews.

Heavy reliance on unstructured data

Scenarios that involve interpreting natural language, extracting meaning from documents, or interacting with users conversationally, for example processing a home insurance claim.

Conclusion

Before committing to building an agent, validate that your use case can meet these criteria clearly. Otherwise, a deterministic solution may suffice.

Core Components

Model (LLM), Tools (external functions or APIs), Instructions (guidelines, guardrails).

Which LLM to use?

Each model has different strengths and tradeoffs.

Considerations

Task complexity, latency, and cost.

Recommendation

Use variety of models for different tasks in the workflow.

A simple retrieval or intent classification task may be handled by a smaller, faster model.

Harder tasks like deciding whether to approve a refund may benefit from a more capable model.

How to choose the model

Build prototype with most capable model for every task as a baseline.

Try swapping in smaller models to see if they still achieve acceptable results.

• observe if it fails.