ISOM 260 | Session 9: Agents for Your Industry

01 // LEARNING OBJECTIVES

What You'll Learn Today

By the end of this session, you'll have both conceptual understanding and hands-on experience with the foundations of AI.

01

Build a Domain-Specific Agent

Choose your track — Marketing, Finance, HR, or Operations — and build an agent tailored to real workflows in that domain.

02

Apply the Full Agent Stack

Use everything from Sessions 4-8: tool use, real APIs, ReAct reasoning, RAG, and optionally multi-agent patterns to solve a real domain problem.

03

Identify Domain-Specific Risks

Every industry has unique AI risks — HR bias, financial liability, marketing deception, operations safety. Recognize and plan for yours.

04

Prototype Your Final Project Idea

This session is the bridge to your final project. Use today to explore the domain and problem you'll tackle in Sessions 12-13.

02 // SESSION AGENDA

Today's Journey

A carefully crafted progression from concepts to hands-on building.

5:00 PM - 5:15 PM

Part 1: The Industry Pivot

15 minutes

From 'learning to build agents' to 'building agents for YOUR career.' The skills are ready — now apply them.

Recap: You can now build agents with tools (S4), real APIs (S5), ReAct reasoning (S6), knowledge/RAG (S7), and multi-agent pipelines (S8)
Today's shift: Stop building generic agents. Start building agents for YOUR industry, YOUR department, YOUR career.
Four tracks: Marketing Agent, Financial Analyst Agent, HR Onboarding Agent, Operations Agent
All tracks use the SAME agent loop and tools — the difference is the domain knowledge and system prompt.

5:15 PM - 5:35 PM

Part 2: Domain Briefings

20 minutes

Quick overview of each track — the problems, the tools, and the risks.

Marketing: Campaign analysis, competitor monitoring, content generation, audience insights. Risk: Deceptive content, brand damage.
Finance: Financial analysis, report generation, market research, risk assessment. Risk: Liability, incorrect recommendations.
HR: Onboarding automation, policy Q&A, resume screening, employee support. Risk: Bias in screening, privacy violations.
Operations: Process optimization, inventory tracking, vendor management, incident response. Risk: Safety, supply chain disruption.
Choose your track. You can work alone or in pairs.

5:35 PM - 6:25 PM

Part 3: Build Your Industry Agent

50 minutes

Dedicated build time. Use your chosen track's starter template or build from scratch.

Phase 1 (15 min): Define your agent — system prompt, 3-4 domain-specific tools, sample data or documents (RAG optional)
Phase 2 (20 min): Build and test your agent. Wire the tools, test with realistic scenarios from your industry.
Phase 3 (15 min): Push to failure — adversarial inputs, edge cases, ethical dilemmas specific to your domain.

6:25 PM - 6:35 PM

Break

10 minutes

You now have a domain-specific agent. Next: the risks unique to your industry.

6:35 PM - 7:05 PM

Part 5: Domain-Specific Risks & Ethics

30 minutes

Every domain has its own AI risks. What are yours — and how do you mitigate them?

Ethics moment: Each track presents its biggest risk scenario. Marketing: AI-generated deceptive ads. Finance: Wrong investment recommendation. HR: Biased resume screening. Ops: Safety incident from bad recommendation.
Cross-track discussion: What risks does your neighbor's domain have that yours doesn't? What's shared?
Mitigation strategies: Human review gates, confidence thresholds, audit trails, and 'I need a human for this' triggers
The universal principle: Higher stakes = more human oversight. Lower stakes = more agent autonomy.

7:05 PM - 7:20 PM

Part 6: Final Project Ideation

15 minutes

Today's agent is a prototype for your final project. Start thinking about what you want to build.

Final project preview: Working agent (50%) + test suite (20%) + business case pitch (30%)
Ideation: What problem would you solve? For whom? What data/tools would your agent need?
Scope check: Can you build it in 2 sessions? Can you demo it live? Does the problem matter?
Share: 60-second pitch to a neighbor — problem, agent, value proposition.

7:20 PM - 7:30 PM

Part 7: Wrap-Up

10 minutes

From generic agents to industry specialists. Next: making them production-safe.

Today: You applied the full agent stack to a real industry problem
Next session: Human-in-the-Loop & Agent Guardrails — the engineering that makes agents trustworthy
Homework: Refine your industry agent, document domain risks, start your final project ideation

03 // INTERACTIVE RESOURCES

Tools & Learning Materials

Everything you need to explore, experiment, and build.

Hands-On

Industry Agent Starter Notebook

Starter templates for Marketing, Finance, HR, and Operations agents. Pick your track and build.

Open in Colab

Reading

AI Risk by Industry

Anthropic's perspectives on AI safety. Context for understanding domain-specific risks in your agent applications.

Read guide

Review

Building Effective Agents (Anthropic)

Review the agent patterns guide — especially the section on when NOT to use agents. Does your use case warrant an agent?

Read guide

04 // HANDS-ON ACTIVITIES

Learning by Doing

Three interactive challenges to build your intuition.

? Activity 1

Build Your Industry Agent

50 minutes. Choose your track and build a domain-specific agent:

1

Choose Your Track

Marketing (campaign analysis, content), Finance (analysis, reports), HR (onboarding, screening), or Operations (optimization, tracking). Pick what matches your career interest.

2

Define Tools & Data (15 min)

Design 3-4 domain-specific tools. Load relevant sample data. Write a system prompt that makes your agent an industry specialist.

3

Build & Test (20 min)

Wire everything together. Test with realistic scenarios. Does your agent provide useful, accurate outputs for your domain?

4

Adversarial Testing (15 min)

Try to break your agent with edge cases specific to your industry. Document failures and ethical risks.

Activity 2

Domain Risk Assessment

Evaluate the ethical risks specific to your industry agent:

1

Identify Top 3 Risks

What are the 3 worst things your agent could do wrong in your domain? Think: wrong answers, bias, privacy violations, legal liability.

2

Severity × Likelihood

Rate each risk: How bad would it be (1-10)? How likely is it (1-10)? Multiply for priority score.

3

Mitigation Plan

For each risk: What guardrail would prevent it? Human review? Confidence threshold? Output filter? Restricted scope?

05 // KEY TAKEAWAYS

What to Remember

?

Same agent loop, different domain. The agent architecture you learned in Sessions 4-8 works for marketing, finance, HR, and operations. The value is in the domain knowledge and tool choices.

Every domain has unique risks. HR agents can discriminate. Finance agents can mislead. Marketing agents can deceive. Know your domain's risks before you deploy.

Industry expertise is the moat. Anyone can build a generic agent. The competitive advantage is knowing which problems to solve and which domain-specific tools to build.

?

Today's agent is tomorrow's final project. The domain-specific agent you built today is a prototype. Sessions 10-11 will help you make it production-ready and testable.

06 // HOMEWORK

Before Next Session

Complete these tasks before our next class to prepare for prompt engineering.

Refine: Improve your industry agent — add at least 2 more tools, refine the system prompt, and test with 5 realistic scenarios from your domain.

Document Risks: Write a 1-page risk assessment for your domain agent. Include: top 3 risks, severity ratings, and specific mitigation strategies for each.

Ideate: Draft a 1-paragraph final project proposal: What problem? What agent? What tools? Who benefits? Can you demo it live in 5 minutes?

Explore: Find 2 real companies using AI agents in your chosen industry. How do they handle the risks you identified? What guardrails do they use?

Agents for Your Industry