For many follow-up teams, the hardest part of the day isn't the difficult conversation. It's the stack of messages waiting behind it. When every reply has to be written from scratch, good responders slow down, and seekers wait.
A free AI assistant can help with the part of the work that doesn't require a person: getting a solid first draft on the screen. The listening, discernment and prayer still belong to your team. Used this way, AI gives responders more time for what only they can do.
Free versions of tools such as ChatGPT, Claude and Gemini all work for this. Here's how to set one up in three steps.
1. Set the guardrails before you start
Agree on three rules as a team before anyone opens a tool.
A person sends every message. AI drafts. Your responder reads, edits and decides. Nothing goes to a seeker unreviewed.
Protect people's stories. Never paste names, locations, phone numbers or other identifying details into an AI tool. Remove them before you ask for a draft, and check the tool's settings so conversations aren't used for training where that option exists.
Know when not to use it. Anyone in crisis or serious distress gets a person from the first word. Flag those conversations for a trained team member, or the proper authorities, right away.
2. Teach the assistant your voice, then draft
Start each session by giving the assistant some context. You can paste in a prompt like this:
"You are helping a Christian ministry's follow-up team draft replies to people who message us online. Our replies are warm, short (under 80 words), plain and personal. Each reply names what the person shared, answers briefly and honestly, and ends with one open question. Don't use religious jargon. Don't include links unless I ask. I will edit every draft before sending."
Then, for each message, paste in the seeker's words with identifying details removed and ask:
"Draft a first reply to this message: [message]. Give me two options, one slightly more personal than the other."
If a draft misses the mark, tell the assistant why: "Too long," "Sounds formal," "They're grieving, so be gentler." It will adjust.
Bonus Strategy: Build a custom GPT (Open AI), Gem (Gemini), or Project (Anthropic) that retains the details of your voice and context so that you don’t have to repeat that first prompt in every session. The downside to this is that most of these features are paid. If cost is a concern, stick with the individual session strategy above.
3. Personalize, then build a shared library
Before sending, every responder should run a quick check. Does this sound like me? Does it respond to what this person actually said? Is there anything here about faith or Scripture I'd want a leader to look at? Then add a personal line only you could write.
Over time, save the prompts that work best in a shared document. A small library of tested prompts for common situations, like prayer requests, doubts or first-time questions, helps new volunteers get up to speed and keeps your team's voice consistent.
Your next step
Choose one responder to try this for a week on routine first replies only. Compare how long replies take and how often seekers write back. Then decide together whether and how to expand it.
Technology works best in ministry when it gives people more room to be present. Let AI handle the first draft so your team can give each seeker their full attention.
