Walkthrough: From a Group Chat to a Plan

Use the walkthrough with the class

This is an interactive exercise for a shared screen. The class will take one messy group chat and apply the three instructions from the lesson to it, one at a time: extract, transform, and organize. Then the class will do all three in one instruction and compare.

A group chat is a good place to practice because everyone has one. Six people plan a Saturday at the lake in twenty messages, and by the end nobody can say who is driving or what time to show up. The information is all there. It is just in the wrong form.

The chat below is provided so the walkthrough can be done as written. If the class has a real planning chat it is willing to share on a screen, with names changed, use that instead and adjust the prompts. Do every step on the same chat, so students see how each instruction changes what is visible.

The chat

Paste this at the start of each new conversation, or keep one conversation open and send it once.

Here is our group chat for Saturday.

Maya: ok so lake saturday, who's in
Jordan: in. i can drive, 4 seats besides me
Priya: in!! i'll bring the cooler, someone else get ice
Theo: in. i have the speaker + the big umbrella
Maya: i'll do sandwiches. meet at my place 9:30?
Jordan: can we do 10, i work till 9
Sam: i'm in but only till like 3, i have work at 4
Priya: 10 works. ice??
Theo: i'll grab ice on the way
Maya: ok 10 at mine. sam we'll drop you back by 3:30
Jordan: does anyone have a second car, that's 6 of us w/ ellie
Ellie: i can drive too, 3 seats
Priya: sunscreen. someone. please
Sam: i got sunscreen and towels
Maya: parking at the lake is $10 a car, bring cash

Step 1 · Extract the people

Start with the simplest extract. Ask who is going.

Extract the people who said they are coming. List each person once.

Ask the class to check the list against the chat before moving on. Is Ellie on it? Ellie never said “in.” Jordan mentioned her, and then she offered to drive. AI has to reason about that, and the class should decide whether it reasoned correctly.

Step 2 · Extract what each person is bringing

Now a harder extract, because the answer is scattered across the chat and some of it changed along the way.

Extract everything someone said they would bring or do, with the person’s name next to each item. If something was asked for and nobody claimed it, list it separately as unclaimed.

Expect something close to this: Priya brings the cooler; Theo brings the speaker, the umbrella, and the ice; Maya makes sandwiches; Sam brings sunscreen and towels; Jordan and Ellie drive. Cash for parking is everyone’s job. Ask whether anything is unclaimed. It should not be. Ice and sunscreen were both requested and later claimed, and AI had to connect a request in one message to an answer several messages later.

Step 3 · Transform the times

The chat has times in several forms: “9:30,” “10,” “till like 3,” “work at 4,” “by 3:30.” Put them all in one format.

Extract every time mentioned in the chat and transform each one into the form H:MM a.m./p.m., with what it refers to. Where a time was proposed and then changed, show only the final time and note that it changed.

The meeting time should come out as 10:00 a.m., with a note that 9:30 was proposed first. Sam’s times should come out as leaving the lake by about 3:00 p.m., dropped off by 3:30 p.m., work at 4:00 p.m. Ask the class what would have happened without the rule about changed times. AI might have listed both 9:30 and 10:00 as meeting times, which is exactly the confusion the chat had in the first place.

Step 4 · Organize by person

Now organize what has been extracted into a table with one row per person.

Organize the plan by person. Make a table with the columns Name, Bringing, Driving, and Notes. Every person who is coming should have a row.

Look at the Notes column. Sam’s row should say something about leaving by 3:00 p.m. Jordan’s row should say he works until 9:00 a.m., which is why the meeting time moved. Ask whether the rides work: Jordan has four seats and Ellie has three, which is seven seats for six people. That fact is not written anywhere in the chat. AI worked it out from two messages, or it did not, and the class should look.

Step 5 · Organize by time

Same information, different organization.

Now organize the same plan by time. Make a schedule for Saturday from the first thing anyone has to do until the last person is home. Include who is involved in each entry.

The schedule should start before 10:00 a.m. (Jordan finishes work at 9:00; Theo stops for ice on the way) and should include the 3:00 p.m. departure to get Sam back by 3:30. Ask the class which table they would want on their phone on Saturday morning, and which one they would want on Friday night when they are packing. The answer is different, and that is the point of having two organizations.

Step 6 · All three in one instruction

Finish by starting a new conversation and asking for the finished plan in one instruction. The class should write it together, and it should name what to extract, what form to transform it into, and how to organize it.

Here is our group chat for Saturday. Turn it into a plan I can post back to the group. Pull out who is coming, who is bringing what, who is driving, and every time. Write every time as H:MM a.m./p.m. and use only the final version of anything that changed. Organize it as three short sections: Who and what, Rides, and Schedule. End with a list of anything the group still has not decided.

[paste the chat]

Compare the result with the pieces the class built in Steps 1 through 5. It should contain the same people, the same items, the same times. Then read the last section. Nobody said where at the lake to meet, or what to do if it rains. If AI lists gaps like these, ask the class where they came from: the chat did not contain them. AI knows what a plan for a lake day usually needs and noticed what was missing. That is useful, and it is also the kind of addition to read carefully.