2025-08-02

chatbot usage examples

tools such as chatgpt expose a natural-language interface to a language model and an extensive knowledge corpus. they can replace web searches by extracting and synthesizing information that would otherwise require manual effort. openai identifies its main audiences as software developers, content creators, businesses and enterprises, and students and educators.

prompts

what is x always
create a short question whose answer is the preceding response
create a list of all questions that i or you explicitly or implictly asked in this chat
create an image that represents me as a person
given a doodle, colorize it
rewrite the following text for clarity
write in the prevalent writing style of the 1960s

use cases

learning and research

scientific calculation
converse to clarify unfamiliar concepts

language

multilingual translation
grammar correction
rewrite sentences for clarity
serve as a conversation partner in a different language
translate application localization files to additional languages
explore language nuances

writing

prime with questions about a topic, then generate a story based on this knowledge
detect potential faux pas
identify content worth responding to
detect content to drop to keep things shorter
create an outline of thematic directions (limited use)
collect text of your sent emails and have it adopt your writing style

tips for writing with chatgpt:

dont rely on it for personal relationships. use it only to explore topics or nuance, but use your own ideas, words, and the full emotional context as you have it
take its words as opinionated hints, not authority
read responses to extract cues, not to reuse text

coding and technical

create domain-specific calculators
write a mathematical function, then convert it to code
migrate code from an old api to a new api after being provided with two examples
translate source code between languages
improve variable names
add comments
generate command lines
convert mathematical equations to code
generate a list of commands for renaming multiple files

data and text handling

extract information
text reformatting
convert text in images to text

daily life and planning

given a detailed timeline of life, propose future goals
fitness plans
recipes
personal finances
scheduling
estimate food calories
upload product manuals and ask questions about the product

entertainment

interpret song lyrics
identify media from partial details
comedy and interactive stories
role playing
adopt the speaking style of a specific persona
recommend music, video channels, fashion, and gifts (note: bias)

summarize all assumptions made in this conversation
list all unresolved questions from this chat
identify implicit constraints or preferences in my prompts
convert this exchange into a faq format
list all named entities in the text
identify my cognitive biases based on this conversation
guess what i might be trying to accomplish, even if not explicitly stated
invert this idea and explore the opposite

projects

projects with instructions and project files are useful because they allow iterative development towards a goal by regularly updating project files and instructions with the current development state.

project file examples

collect documentation of a software library or a product you are researching as a single markdown text file

llm agent automation

especially when programming, applying suggested changes can be inefficient. exchanging updated results or error messages is also not automated.

some ideas and potential solutions:

applying patches
- ast-based search and replace; replace entire functions
- fuzzy string matching (problem: llm diffs are unstable and imprecise)
filesystem modifications
- generate necessary shell commands
state analysis
- connect to a debugger that automatically sends state data
- use a controllable sandbox via text instructions