ChatGPT is a large language model developed by OpenAI that uses deep learning techniques to generate human-like responses to natural language inputs. It has been trained on vast amounts of text data, making it capable of answering a wide range of questions and carrying on conversations on a variety of topics.
The advantages of ChatGPT include its ability to automate customer service and support, provide personalized assistance, and generate more engaging content for social media platforms. However, there are also potential threats, such as the possibility of malicious actors using the technology to spread disinformation and manipulate public opinion. Additionally, the widespread use of ChatGPT could potentially displace human jobs in customer service and support industries.
“Please note that the above information is also generated by ChatGPT.”
Some of the prompts or queries we’ve explored are:
template building (prd / gtm ) “write a product requirements document for a new feature that allows travelers to book experience like cruise and local sightseeing for a product manager for a world travel guide company”
“Write a go to market strategy for that feature, launching in San Fransisco “
prd writing for fillers “Expand on the metrics, go two levels deep”
condensing text “Rewrite this shorter and punchier”
marketing copy “Write a compelling call to action to try this feature, write it with a fun attitude”
draft messages “Write a slack message to the design team telling them nicely to revert to the original design”
meeting agenda and strategy “Give me a minute by minute agenda for a product review“
performance review “Write a professional performance review for an engineering lead who has helped us meet the launch deadline, give them some compliments, but in a compliment sandwich explain he sucks as a people manager, keep it optimistic.
Writing user stories is a breeze with it. The prompt allows you to be rather precise about the project you are working on, its objectives, etc. After that, request that it output the project’s user stories. Based on the output, you can then have it become more detailed. It’s quite cool; I advise experimenting with it.
They are, but it’s great at generating a bunch of them in a very short amount of time. Also can give you ideas of stories that you may not have thought of yourself.
I haven’t tried acceptance criteria, but I imagine it can yield pretty good results too.
That’s my experience in a nutshell. 60% solid output, 40% that’s anywhere from mildly inaccurate to junk. Throwing out the junk has proven much quicker than writing the good stuff from scratch.
Really helpful for competitive research/business cases. It’s like using google but 10000x better.
Right now, one of my big projects is reimagining our entire user experience across our platform. Obviously, this is a big lift requiring a lot of resources, so I need to make a business case for it. I told Chat GPT “Tell me about how the quality of product’s user experience impacts sales of that product. Please include specific numbers and studies, with links to each.” ChatGPT summarized results from gartner, Forrester, UserTesting, and one other place. Leadership loved it and green lit my project.
This is something that would’ve taken me a week (not 40 hours; but a couple hours over 5 days) to collect information for, then another day-ish to frame it up into something readable. Instead, it took ChatGPT 1 minute, and then I spent another hour or so reviewing the sources and adding to misc. decks/documentation.
Let’s say you’re building a nutrition app and you want to implement a feature that allows you to track calories. You can simply input “write me a user story that describes the ability to count calories…”.
You feel like the output isn’t specific enough? Simply ask it to be more specific. E.g., “rewrite this user story based on the assumption that I have a daily goal count that I need to work forwards.
Need some ideas on acceptance criteria? Ask it to write acceptance criteria as well.
Also, HUGE DISCLAIMER - assume anything you input into ChatGPT will be used to further train the algorithm. So, make sure to it e not inputting anything my that’s “confidential”
For a non-work-related example, try using the following prompt:
Create a spreadsheet for me that includes the name of all films that have won the academy award for best picture in the last 20 years. Include the year the movie was released, director, lead, IMDB score, rotten tomatoes score, and metacritic score as separate columns.
I was very surprised that it could create spreadsheets. But lo and behold.
Warning: It won’t code for you. Or it might, but it would be incorrect. The llm needs to be led in the identification of selectors and pathways, text recognition, loop management, etc. Basically, you must write all of the pseudocode. Also, GPT3 won’t help you if you don’t know how to scrape (unless you’re attempting to learn).
It was only used by one of our devs to provide some functionality for shits and grins.
He had it complete the task in three steps (grab this thing, do that thing, and write something here), after which he pieced it together, tested it, and it was successful.
That’s exactly what I said, it would not code for you, and if it did, you would have to re-write it… Anyways, I’m glad that it has been helpful to your dev in some way…
Helps plan workshops or collaborative activities – seriously, it writes a great agenda and helps come up with a great first draft of how workshops should run
Helps a ton with marketing copy. A lot of “can you make this more concise?” or “can you make this copy feel a little more conversational/professional/marketing-speak?”
Helps with micro-copy ideas
Can write really accurate user personas (again, more of a first draft you can polish)
Can explain how to code things to me (for side projects, not work related)