Back to Blog

How to Build Data Workflows with AI

Use ChatGPT, Claude, and other AI assistants to build with People Data Labs

Vinay Rajur
Vinay Rajur
10/30/26
7 min
AI assistants like ChatGPT and Claude are making it easier than ever to build with data.
What once required hours of writing code and troubleshooting API integrations can now start with a simple prompt. Whether you're enriching a prospect list, researching companies, or building a data-powered application, AI can help you get from idea to working code much faster.
But there are some important things to get right.
Without the right context, AI assistants can generate code that uses incorrect API parameters, references nonexistent fields, or consumes more data credits than expected.
At People Data Labs (PDL), we've seen customers run into these issues, and we've built tools to help avoid them.
In this post, we'll share four recommendations for building reliable data workflows with AI, and how to get started with PDL.
1. Start with your use case, not code
One of the biggest benefits of AI-assisted development is that you don't need to start with a technical implementation. You can start by describing what you're trying to accomplish.
For example, maybe you want to:
  • Enrich a list of prospects with job titles, contact information, and company details.
  • Find companies in a specific industry or geography for market research.
  • Identify professionals with particular experience or skills.
  • Automate a recurring data enrichment workflow.
PDL offers two primary types of APIs to support these use cases:
  • Enrichment APIs: Find and enrich a specific person or company using information you already have.
  • Search APIs: Discover people or companies that match criteria you define.
Not sure which API you need? You can ask your AI assistant to help you figure that out too.
The important thing is to be clear about your goal, what information you already have, and what you want the output to look like.
2. Give your AI assistant the right documentation
This is probably the most important recommendation we can make:
Don't assume your AI assistant knows how an API works. Give it the documentation.
AI assistants don't always have the latest information about an API, and they can sometimes make incorrect assumptions about supported fields, parameters, or functionality.
At PDL, we've made our developer documentation easier for AI assistants to access and understand.
Every page in our docs is available in Markdown, a text-based format that's particularly easy for AI tools to work with.
There are a few ways to use this:
  • Markdown pages: Add .md to the end of any PDL documentation URL to get its Markdown version.
  • Copy page: Use the copy option in our docs to share documentation directly with ChatGPT or Claude.
  • llms.txt: Give your AI assistant an index of our documentation so it can find relevant pages itself.
For example, you can start with a simple instruction:
Use https://docs.peopledatalabs.com/llms.txt to find the relevant PDL API documentation. Only use documented endpoints, fields, and parameters. If you're unsure about something, ask rather than guessing.
This helps ground your assistant in the actual API documentation and can save a lot of troubleshooting later.
3. Test your workflow before spending credits
AI assistants make it easy to experiment, but you probably don't want to use production API credits every time you test an idea.
That's why we recommend starting with a Sandbox.
PDL offers free Sandbox APIs that let you build and test API integrations using synthetic data, without consuming credits.
Note: PDL offers free Sandbox APIs that let you build and test API integrations using synthetic data, without consuming credits.
Our Sandbox APIs use the same general structure as our production APIs, so switching between them is straightforward.
For example, here's the difference between the Person Enrichment endpoints:
Production: api.peopledatalabs.com/v5/person/enrich

Sandbox: sandbox.api.peopledatalabs.com/v5/person/enrich
You can ask your assistant to build against the Sandbox first, test the integration, troubleshoot issues, and then prepare a production version once everything works.
One thing to keep in mind: because the Sandbox uses synthetic data, it's intended for testing your integration, not evaluating real-world data coverage or match rates.
4. Build in a few guardrails
Once you have a working prototype, there are a few important things to consider before scaling up.
Here are our recommendations:
  • Start small: Test with 5–10 records before processing your full dataset. Review the results and make sure everything works as expected.
  • Control your costs: Set explicit limits on API requests and records returned. For enrichment, use likelihood scores to control how confidently records are matched.
  • Use bulk APIs: PDL's Bulk Enrichment APIs support up to 100 records per request, making larger workflows more efficient.
  • Protect your credentials: Keep API keys out of your prompts and source code, and follow your organization's policies when sharing data with AI tools.
  • Review before running: Make sure you understand what the generated code does, how many requests it will make, and when it will stop.
You don't need to know what every line of code does, but you should understand how your workflow behaves before running it against production data.
And remember, your AI assistant can help you review all of this too.
Want to Try It Yourself?
We've put together a more detailed guide to building PDL API workflows with AI, including:
  • Ready-to-use prompts for Person and Company Enrichment and Search.
  • Common mistakes to watch for when working with AI-generated code.
  • A code review checklist to help you catch potential issues before running your workflow.
Whether you're building your first integration or looking to speed up an existing workflow, the guide can help you get started with more confidence.
Making Data Easier to Work With at PDL
We believe AI assistants will play an increasingly important role in how people discover, access, and use data.
And as the way people build with data changes, data providers need to evolve too.
Good data isn't enough on its own. It also needs to be well-documented, accessible, and easy to integrate into the tools and workflows people are already using.
At PDL, we're continuing to invest in making our data easier to work with, whether you're writing your own integrations, experimenting with AI-assisted development, or building production-scale applications.
Our AI-readable documentation, free Sandbox APIs, and flexible enrichment and search tools are all part of that effort.
Ultimately, we want you to spend less time figuring out how to access data, and more time putting it to work.
Get Started
Happy building!
About the Author
Vinay Rajur
Vinay Rajur
Product Marketing
People Data Labs
LinkedIn Logo
How to Build Data Workflows with AI | Blog | People Data Labs