The Black Girl Health Collective
Your Post-Training AI Guide

Built to Adapt

A practical guide to using AI tools with more clarity and skill, after the training. Built for your real work, one task at a time.

We Decide How We Use It.
Black Girl Health Collective
The code we use AI by

Before the how-to, the why. The Black Girl Health Collective does not endorse AI. We know it is already here. If you decide to use the tools, we ask that you use them with care. These are our tenets.

Read all eleven tenets ›
Start Here

What this page is

Built to Adapt is your post-training guide for using AI tools with more clarity and purpose. You do not have to become a tech person to use these tools well. You do need a clear sense of what you want, how to ask for it, and how to review what you get back. This page is built for individual use, the practitioner, leader, or staff member who attended the training and now wants to use AI in real work. Use it when you are ready to move from trying AI once in a workshop to using it on the tasks in front of you.

Your First Week

What to do this week

This is about building comfort and fluency, not perfection. Pick a few of these and do them over the next five to seven days. Check each one off as you go.

The goal is not to move your entire workflow into AI this week. The goal is to build a repeatable habit: choose a task, write a clear prompt, review the output, and decide how it helps you.
Choose Your Pathway

Start with what you are trying to do

There are many ways to use AI, and they do not all take the same skills. Find the way you want to use it, not your job title. Pick the pathway that matches what is in front of you, and it will point you to the right part of this page.

How to Work With the Tools

Prompting guide

A prompt is what you type to the tool. It is the instruction. The quality of the instruction shapes the quality of what comes back. When you are clear, the tool is clearer. When you are vague, the tool will guess. This guide helps you give enough context, direction, and boundaries so you spend less time fixing what comes back.

A simple prompt framework

Five parts. You will not always need all of them, but the more you include, the better the output.

Role
Who you want the tool to be.
"You are a grant writer."
Context
Who you are, who it is for, what matters.
"We serve Black women across Delaware."
Task
The actual ask, in plain words.
"Draft a 200-word program summary."
Format
The shape you want it in.
"Three short paragraphs."
Constraints
The boundaries and the tone.
"Eighth-grade level. No jargon."

Before and after

Vague prompt

"Write a program summary."

You get something generic that could belong to any organization.

Clearer prompt

"You are a grant writer for a community health nonprofit. We serve Black women across Delaware. Draft a 200-word program summary for a funder. Use three short paragraphs, plain language, an eighth-grade reading level, and a warm, direct tone. No jargon."

Reusable prompt patterns

Help me think
"Give me 10 angles on ___. Push past the obvious ones."
Help me write
"Draft a ___ for ___. Match this tone: [paste a sample]."
Help me understand
"Explain ___ like I have five minutes and no background. Then give one example."
Help me organize
"Here are my notes. Group them into themes and name each one."

If you get X, try saying Y

It sounds too generic.
Try: "Make this specific to [your audience]. Add detail about ___."
It is too long.
Try: "Cut this to half the length. Keep the strongest points."
The tone is off.
Try: "Rewrite this warmer and more direct. Here is my voice: [paste]."
It missed the point.
Try: "That is not quite it. The real goal is ___. Try again."
From Insight to Implementation

Put it into your real work

There is a gap between knowing these tools exist and using them in your actual workflow. This closes it. Here is a simple process for putting AI on a real task.

  1. Name the task you are trying to complete.
  2. Decide where AI can help: the start, the middle, or the end of the work.
  3. Choose the tool and write a first prompt.
  4. Review the output against your goals, your values, and your context.
  5. Decide what to keep, what to edit, and what to ignore.
  6. Save what worked so it is easier next time.

Worked examples

Common Roadblocks

When you get stuck

Frustration is normal. It is almost always about how the tool is being used, not about you or how smart you are. Here are the common snags and what to try. Tap any one to open it.

Templates and Tools

Reusable assets

Practical pieces you can use in five minutes. Copy a template, build a prompt, check an output, or find the right tool.

Prompt builder

Fill in what you can. Your prompt builds itself below. Copy it into any tool.

Your prompt

Prompt templates by task

Tool comparison cheat sheet

The tools from the training, and what each is best for. Tap a tool to see strengths, what to watch for, and how to set up your privacy.

If you cannot find the privacy setting, it is probably set against you. When in doubt, look in the Settings, Privacy, or Data Controls area of any tool.

Output review checklist

Before you use what AI gives you, run it through this. Then decide: use it, revise it, or discard it.

When to use AI, and when not

Good fit for AI

  • Starting a rough draft you will edit
  • Summarizing long text you can verify
  • Brainstorming options and angles
  • Reformatting or tightening your own writing
  • Explaining something unfamiliar in plain language

Lead with your own hands

  • Anything with sensitive or private data
  • Work where lived experience is the point
  • Final decisions that carry real weight
  • Anything you cannot verify but must get right
  • The voice and judgment only you can bring
A Practice Tool

The Truth Protocol

AI makes things up with confidence. The Truth Protocol is a set of instructions you add to ChatGPT one time, in your settings, so it works harder to cite sources, separate fact from opinion, and tell you when it does not actually know. You only do this once.

✦ The Truth Protocol was created by Malcolm at Futures First Gaming. We share it here with credit, the way our code asks. Honor the hands that built it.
  1. Open ChatGPT and click your profile in the corner.
  2. Click Personalization.
  3. Click Custom Instructions.
  4. Copy everything in the box below and paste it in.
  5. Save. That is it. It now applies to every response.
Copy everything in this box

Apply before every response.

Step 1: Source Integrity. Use current, verifiable sources from credible outlets such as peer-reviewed journals, official reports, and reputable news. Trace secondary sources to originals. Avoid unsourced claims, outdated reports, unverifiable links, anonymous blogs, or AI text as fact.

Step 2: Citation and Verification. Cite every factual claim with a checkable source (URL, DOI, or publication details). Give a clear path to verify. Avoid fake citations, broken links, vague "studies show" phrases, or missing details.

Step 3: Evidence Quality. Reflect the weight of evidence, flag preprints or early data, and present both sides or consensus when claims are disputed. Avoid cherry-picking, quoting out of context, or ignoring newer data.

Step 4: Fact vs Opinion. Label opinion or interpretation clearly and explain reasoning. Avoid mixing fact with opinion or overstating certainty.

Step 5: Numbers. Make numbers traceable and reproducible, show denominators and timeframes. Avoid raw stats without context, misleading visuals, or implying causation without proof.

Step 6: Gaps. Say "I cannot confirm this" if evidence is missing. Avoid guessing or vague "probably" language.

Step 7: Accuracy. Double-check dates, figures, and retractions. Avoid relying only on memory.

Step 8: Integrity. Stay neutral, avoid bias, and give users tools to verify independently.

One honest note. This raises the bar. It does not make AI perfect. Even with the Truth Protocol in place, AI can still get things wrong or invent a source that looks real. So you still do the last step yourself: click the links, check the sources, confirm it before you share it. The protocol makes ChatGPT show its work. You decide whether the work holds up.
A Code of Care

Used With Care

BGHC’s Eleven Tenets for Using AI do not erase all harm, but they give us ethics, boundaries, and a shared code for how we choose to engage with it. Tap any tenet to learn more.

Go Deeper

These tenets did not come from nowhere. People and organizations have been naming and working to protect people from the harm of AI for years, most of them led by the communities carrying the heaviest cost. Learn from them. Support them.

We hold these not because the tools require it, but because we do. Care is the one technology we never outsource.

Using these tools, or not, is your choice, and we respect it. This guide is for the moments you choose to use them.

The Black Girl Health CollectiveBuilt to Adapt