This is how I'd read a folder full of json files and turn them into a dataframe. There's probably a more efficient way that Pandas can do, but this gives a good level of fine control.

a.json

{
    "name": "A",
    "chats": [
        {
            "dt":"2020-07-02 10:32:54.876422",
            "from":"me",
            "message":"yo"
        },{
            "dt":"2020-07-02 10:35:54.876422",
            "from":"them",
            "message":"🍆⁉"
        },{
            "dt":"2020-07-02 10:36:54.876422",
            "from":"me",
            "message":"no"
        }
    ]
}

b.json

{
    "name": "B",
    "chats": [
        {
            "dt":"2020-07-02 10:32:54.876422",
            "from":"me",
            "message":"yo"
        },{
            "dt":"2020-07-02 10:35:54.876422",
            "from":"them",
            "message":"🍳🥪"
        },{
            "dt":"2020-07-02 10:36:54.876422",
            "from":"me",
            "message":"👏 yes 👏"
        }
    ]
}

json_slurper.py

import os
import json

import pandas as pd

dataframes_of_chats = []

for json_file in os.listdir("."):
    print(json_file)
    if ".json" in json_file:
        the_file_object = open(json_file, "r", encoding="utf-8")
        file_contents = json.load(the_file_object)
        print(file_contents)
        messages = file_contents.get("chats")
        correspondent = file_contents.get("name")
        chats_with_x = pd.DataFrame(messages)
        chats_with_x["correspondent"] = correspondent
        dataframes_of_chats.append(chats_with_x)

all_the_chats = pd.concat(dataframes_of_chats)

print(all_the_chats)