This article was originally published on LinkedIn.
There's a growing gap between companies treating AI as an experiment and companies rebuilding their operations around it. The difference isn't budget — it's approach.
Morgan Stanley, Klarna, and Lowe's aren't just running pilots. They're using AI to rewrite how financial advisors work, how customer service scales, and how inventory moves. And they're doing it systematically, with clear metrics and guardrails.
Earlier this year, OpenAI published AI in the Enterprise: Lessons from Seven Frontier Companies. The lessons are unusually concrete — and surprisingly accessible to businesses that don't have billion-dollar R&D budgets.
If you're running on NetSuite and experimenting with Claude or ChatGPT via the AI Connector, you already have what you need to implement the same - or very similar - strategies. This post walks through their seven lessons and shows you how to apply them in your own environment — intelligently, pragmatically, and with finance at the core.
Lesson 1: Start with Evals
Case Study: Morgan Stanley
Morgan Stanley is one of the world's largest investment banks and wealth management firms, known for its data-driven decision-making and regulatory rigor. Naturally, they started with rigorous "evals" — structured tests that measured how AI models performed in specific contexts like summarization and translation. This gave them the confidence to scale AI responsibly across the firm.
For businesses running on NetSuite:
- Test Claude's variance explanations against your finance team's commentary from prior months.
- Measure reconciliation accuracy by comparing AI-prepared Trial Balance checks with human-prepared ones.
- Score Claude's summaries of Income Statements for accuracy, relevance, and clarity.
By building evals into your monthly close, you'll know when Claude is trustworthy — and when to step in.
Lesson 2: Embed AI into Workflows
Case Study: Indeed
Indeed is the world's leading job search platform, connecting millions of employers and job seekers through data-driven matching and recommendation tools. They didn't just add AI as a helper tool — they embedded it directly into their product, using GPT to power job matching and explain recommendations.
For businesses running on NetSuite:
Rather than trying to embed AI directly inside NetSuite, use custom agents to bring AI into your existing workflows. Think of these agents as lightweight, specialized processes that (1) pull the right data, (2) analyze it with an LLM, and (3) deliver clear, actionable outputs where people already work.
- Monthly Report Summarizer Agent: On a schedule, export your Income Statement and Balance Sheet, have the agent generate a concise executive summary (key trends, risks, wins), and send it as an email or shared doc.
- Variance Analysis Agent: When new comparative Income Statements are available, calculate period-over-period changes, draft variance explanations, and flag unusual movements for review.
- KPI Monitoring Agent: Pull saved searches for key metrics (e.g., gross margin by subsidiary or DSO). When thresholds are exceeded, generate context and send a short "insight alert."
- Close-Package Agent: During close, chain tasks: export reports → run tie-out checks → generate the flux analysis → draft the executive summary → distribute results for review.
These agents don't require changing NetSuite's interface. They simply add an intelligent layer around your existing processes so insights appear in the tools and communications your team already uses.
Lesson 3: Start Now and Invest Early
Case Study: Klarna
Klarna is a global fintech company based in Sweden, best known for its "Buy Now, Pay Later" platform that serves more than 150 million consumers worldwide. Their AI assistant now handles two-thirds of all customer service chats, saving $40 million annually. They achieved this not by waiting for "perfect" AI, but by starting early and compounding improvements over time.
For businesses running on NetSuite:
- Use Claude to generate draft board reports from Income Statements and Balance Sheets.
- Ask it to prepare plain-language explanations of cash-flow trends.
- Generate draft commentary for budget vs. actuals.
Even if early results require refinement, each cycle trains your teams and builds an AI knowledge base. Over time, those small wins compound into transformational outcomes.
Lesson 4: Customize and Fine-Tune
Case Study: Lowe's
Lowe's is one of the largest home improvement retailers in the world, operating more than 1,700 stores across North America. They fine-tuned GPT on their product catalog, improving accuracy by 20% and error detection by 60%.
For businesses running on NetSuite:
- Incorporate industry-specific terminology from your chart of accounts (e.g., SaaS metrics, wholesale terms, retail store codes).
- Leverage historical management commentary so outputs match your leadership's tone, format, and KPIs.
- Include internal KPIs beyond GAAP (e.g., gross margin by location, CAC, same-store sales).
The result: outputs that feel like they were written by your finance team — not a generic AI.
Lesson 5: Get AI in the Hands of Experts
Case Study: BBVA
BBVA (Banco Bilbao Vizcaya Argentaria) is a multinational banking group headquartered in Spain, known for early digital transformation and innovation in fintech. They rolled out ChatGPT Enterprise to 125,000 employees and let them build their own custom GPTs. In just five months, employees created 2,900+ AI-powered tools, reducing project timelines from weeks to hours.
For businesses running on NetSuite:
- Let analysts create prompts for flux analysis or cash-flow forecasting.
- Enable controllers to design AI-driven reconciliation workflows.
- Give CFOs board-ready executive-summary templates they can refine.
When the people closest to the process drive adoption, AI doesn't stay in IT's sandbox — it delivers value where it matters most.
Lesson 6: Unblock Your Developers
Case Study: Mercado Libre
Mercado Libre is Latin America's largest e-commerce and fintech platform, often called the "Amazon of the South," serving over 100 million active users. They built Verdi, an AI-powered development platform that helps 17,000 engineers build apps faster, reducing backlog and accelerating delivery.
For businesses running on NetSuite:
- Automate repetitive data-fetching and report-prep tasks so developers can focus on reusable agent patterns and orchestration scripts.
- Standardize how data is pulled, structured, and passed to AI so new agents can be built and tested quickly.
Once your teams and systems are ready, the next step is to scale ambition — to move from experimentation to automation.
Lesson 7: Set Bold Automation Goals
Case Study: OpenAI
OpenAI itself uses AI to automate internal Gmail-based support workflows, handling hundreds of thousands of repetitive tasks every month.
For businesses running on NetSuite:
- Automate full close-package reporting: Income Statement → Flux Analysis → Executive Summary → delivery.
- Use AI agents to reconcile the Trial Balance against supporting ledgers.
- Trigger risk alerts when KPIs cross thresholds, with agents drafting potential explanations and action items.
The future of AI in NetSuite isn't just "analysis on demand" — it's agentic automation, where AI proactively monitors, analyzes, and reports without waiting for a prompt.
Conclusion: A Playbook for NetSuite AI
AI in the Enterprise shows that success with AI isn't about plugging in a model and hoping for the best. It's about adopting a new mindset: test, embed, customize, empower, and automate.
- Evaluate before scaling.
- Embed AI into daily workflows.
- Start small, compound wins.
- Customize for your business context.
- Empower the experts.
- Unblock developers.
- Set bold automation goals.
Do that, and AI won't just help you analyze NetSuite reports — it will fundamentally transform how your finance team operates. The next generation of NetSuite finance teams won't just use AI — they'll work alongside it. The sooner you get started, the faster that future arrives.