Give It Your Toolkit | Just Add AI

See the Difference

You want code that sends reminder emails to customers with overdue invoices.

Without your toolkit

You: "Write code that emails customers who have overdue invoices."

AI guesses at your system:

import OverdueMailer from 'acme-invoicing'; // Error: Module 'acme-invoicing' not found client.sendOverdueNotices(); // Error: 'client' is not defined

AI invents functions that don't exist. The code won't run.

With your toolkit

You: "Here are functions I have:"

getOverdueInvoices() â†’ Invoice[]
getCustomer(id) â†’ {name, email}
sendEmail(to, subject, body) â†’ boolean

"Write code that emails each customer with an overdue invoice."

AI uses your real functions. The code actually works.

Why This Works

AI doesn't know about your code, your company's systems, or your private APIs. When you ask it to write code without context, it guesses — and often invents functions that don't exist.

But when you tell AI what tools you have available — just their names and what they do — it can assemble them into new solutions. You're giving AI the building blocks; it figures out how to put them together.

The Efficiency Secret

You don't need to share how your code works inside — just what it does and how to call it. This means you can describe far more tools in the same space.

What you share	Space needed	Functions you can describe
Full code implementations	~150 words each	~10 functions
Just signatures & descriptions	~25 words each	~60 functions

This approach actually produces better results — AI focuses on what things do, not how they work inside.

What to Share vs. What to Skip

Function names — what you call them
Inputs — what goes in (and types)
Outputs — what comes back
One-line purpose — what it does
Constraints — limits, errors it throws

Skip This

Implementation code — how it works inside
Internal logic — algorithms, optimizations
Database details — unless directly relevant
Full documentation — brief is better
Every edge case — just the main behavior

How to Ask

"Here are the functions I have available: [list them]. Write code that [does what you need]."
"My system has these APIs: [describe them]. Build me a script that [your goal]."
"Using only these tools: [list], create code to [task]."
"I have access to: [services/functions]. Combine them to [accomplish goal]."

The Technique

Describe the tools you have — not how they work, just what they do and how to call them. AI becomes a composer, assembling your existing pieces into new solutions that actually work with your system.

When to Use This

• You have internal APIs or libraries AI hasn't seen
• You want code that integrates with your existing system
• AI keeps inventing functions that don't exist
• You need to combine several of your tools into a new workflow
• You're building automations using your company's services

Think of It Like This

Imagine hiring a chef for a private dinner. You don't teach them how to grow tomatoes or raise chickens — you just show them what's in your pantry. "Here's what I have. Make something great."

That's this technique. You show AI your ingredients (functions, APIs, tools), and it creates the recipe (new code that combines them).

The chef doesn't need to know where you bought the olive oil. They just need to know it's there and what it's good for.