Spaces

First-week checklist for an AI agent at a space organisation

This checklist walks a space research or technology organisation through its first week with an AI agent answering chats. It starts with published facts and keeps sensitive topics with people.

Runs in your browser. Nothing you enter is sent anywhere or saved.

Before you start

The first week with an AI agent is about trust, both yours and the public. You are teaching the agent what your centre is happy to say, and you are learning where the questions really come from. A calm, narrow start is much better than switching on everything at once.

Work through the list in order. Start with the public line, where information is already published, and leave press and technical topics on hand-over from day one. Keep a note of what surprised you, because that note will shape your written answers in week two.

Tick each one as you go. 0 of 9 done.

[ ] Write one page of public facts: sites, hours, tours, events and education offers.
[ ] List the topics that are never discussed in chat.
[ ] Name a person for each hand-over: press, security, contracts, outreach.
[ ] Connect your messaging accounts and run the agent in suggest-only mode.
[ ] Test with ten sample messages, including a tricky technical one.
[ ] Switch on replies for the simplest questions only, such as hours and booking.
[ ] Read every handed-over chat at the end of each day.
[ ] Check the launch and event calendar for the coming month.
[ ] Hold a short team review on day seven and write down three changes.

A sensible shape for the week

Use the first two days for writing and testing, not for live chats. Days three and four are for suggest-only mode, where your staff see what the agent would send. Days five to seven are for live replies on simple topics, with a person reading the results every evening.

If a scrubbed launch or a school holiday falls inside the week, move the start. A calm week teaches you more than a chaotic one.

Signs the week went well

You do not need numbers to judge it. Look for these signs.

  • Staff say the messages reaching them are the ones that truly need them.
  • Replies match what your website says, word for word where it matters.
  • Hand-over notes tell the press office or contracts team what they need at a glance.
  • Nobody has been surprised by something the agent said.

FAQ

Do we need to tell the public we use an AI agent?

It is good practice to say so in your first message or on your contact page, and some rules may require it. Being open helps people trust the replies and know when a person has joined.

How much written material do we need before starting?

One clear page of public facts, plus a short list of excluded topics, is enough for week one. Add more as real questions show you what is missing.

What if the agent gets something wrong in the first week?

Correct the written reference, tell the sender plainly, and note it for review. Keeping the agent to published text and handing over anything unclear keeps mistakes rare and small.

More free tools for space research and technology programs

Read the guide: Answer visitor, student and supplier questions in chat, and keep technical and security matters with your people.