A place to tinker with AI
Image from Figma plugin “Craf-AI Art Generator” using prompt “robot playing in a futuristic playground with laptop”
“Immersion is the key to developing mastery in any field.”
- Robert Greene
Incubyte is gearing up to accelerate our client-product journey powered by AI (batteries included)!
We help clients integrate AI into their products, and proactively use AI tools to augment our abilities as Software Craftspeople.
As AI evolves, so do we. And so, we're constantly playing with it.
We wanted to share our learnings and interesting outcomes with everyone. As a result of our playing around, we came up with AI Playground (clever wordplay, eh!).
AI playground is a no-limits experimental zone for humans at Incubyte to explore the boundless possibilities of AI.
*We use our continuously growing AI capabilities for our clients, with their permission of course.
We built Code Critique front-end using ChatGPT. Isn't that awesome!!
Backed by OpenAI APIs, Code Critique is a simple tool that can help you find bugs, understand, or improve your code.
P.S. Don't judge us on the UI. It's ChatGPT and not us. (Thank lord we don't have to blame other humans anymore)
Use Code Critique to get explanations for a piece of code, find bugs and let it improve the code for you.
By clicking on this button, you are consenting to send data to OpenAI
Between the two extremes of AI replacing humans and AI being just an addition, what’s the reality?
Watch this short discussion between our cofounders, Rushali and Sapan; they discuss the pros and pitfalls of using AI in software development. They also touch upon the fears around job loss. Watch out for Sapan’s tip on using AI - trust, but verify.
AI takes away software jobs.
Well, not yet. While the capabilities are immense, the best results come when AI compliments the work done by humans.
Check out this playlist that captures how Incubees use the AI tool GitHub CoPilot to augment their work quality while reducing time and effort.
AI takes away software jobs.
Well, not yet. While the capabilities are immense, the best results come when AI compliments the work done by humans.
Check out this playlist that captures how Incubees use the AI tool GitHub CoPilot to augment their work quality while reducing time and effort.
Our Incubee, Visrut, developed a quick use case where Generative AI is utilized to summarise PDFs.
It’s tiring to sift through long PDFs, especially when you’re running short of time.
Here’s another way how Incubyte is using AI to reduce cumbersome tasks.
.png)
Long legal documents, ebooks, or resumes - it can summarise it all,
while making perfect sense (which is quite essential, wouldn't you agree?)
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Aliquam dapibus, purus eu scelerisque ullamcorper, augue dolor maximus risus.
Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.
Go fast, Co-pilot helps you go faster by simply removing a lot of typing. Just imagine you could type 2-3 times faster. The long test names, or common methods, are automatically suggested so you can get a lot done faster.
Saves time from browsing/searching. For instance, imagine writing a variable called ssnRegex and co-pilot suggesting a right regular expression
Save time by eliminating trivial but typing-heavy activities with ChatGPT. For example, you can create a prompt like "Create a POJO to encapsulate OpenAI v1/completion endpoints response in Java, use Jackson."
Leads to clean code, in order to to get suitable suggestions, we need to name things clearly, though this may mean our naming may lean towards implementation and not domain
A comment can generate a SQL query! Isn't that awesome?
Measure twice and cut once. The code must be double-checked before accepting as a solution. AI is known to hallucinate
Legal implications, generated code is coming from someone's server. It may be a copy of existing, copyrighted code! Also, we don't know yet, how much of our code is being sent to OpenAi
Hard to do TDD. When practicing TDD, Copilot may generate way more production code than you need. Keeping the scope of suggestions limited to the test in question is hard.
Incorrect code generation, the suggested code may not be the best possible solution.
