I have been fascinated by the AI inventions and the new technology popping up daily. LLM talk recenty that is popping up in all technology corners, and I’ve started to move past my impostor syndrome of data science work and started to really get hands on with Python and the cool things that you can do with this new LLM technology. LLM is “Large Language Model” and is the secret sauce of how things like chat GPT and the like can generate responses to you, and chatGPT leads the game right now with how many “parameters” the LLM has been trained on. This can be thought of what it’s been taught, but training a model and bending it to your will is out of scope for what I hope to talk about in this post. Overall the advancements have impacted technology in an awesome way, with Code Generation and auto documentation and all of the yet-to-be discovered ideas that are possible when something has been developed that can respond so quickly and thoughtfully. I have been a developer for 20 years and worked with Salesforce for 10 years, and I can say without a shadow of a doubt that this time right now is unlike any other in my dev history, where someone has the tools readily available that can make them as prolific as someone who has been doing it twice as long. This isn’t a shortcut or cheating or anything, you still have to know the basics, but let me paint a picture to show you what I mean. Circa 2020: Developer sits and develops a feature, let’s say an Aura component. Pre- GPT days if you got stuck, your options were limited. :: search google:: You were at the mercy of Stack Overflow and Salesforce Forums that would be a russian roullete of answers to your problem. Memes and other tropes were made about the rudeness of people and unwillingness to help, and other people came out and helped EVERYONE (shout out to sfdcfox), so it was a mixed bag. You got your idea to try, and you plugged it in to your code, tested, and tried to move forward. It meant you had to be tenacious about it, and if you didn’t have a senior resource around, you could really feel like you were on an island if you had to fix a bug. The anxiety that goes along with the search for your answer, only is worth it because of the satisfaction of FIXING it, which eventually always happens. Circa 2023 (Today): Developer sits and develops that same feature, and with a different perspective. The options to ask chatGPT about how to approach, asking chatGPT for a snippet, or using Co-Pilot to auto fill the code, are something that just didn’t exist before. But now, I never have to leave VSCode. I can use the extensions I have and the integrations they offer to have multiple inputs to AI – based assistance, and it doesn’t take long before that gives you a confidence about your approach. The thought process now may be something like: 1) I dive into the problem more quickly knowing that the anxious panic of googling for answers for potentially hours isn’t my only option 2) I can plug into chatGPT something like “Show me an example aura component js controller that will get URL parameters” — 9/10 this is going to be a copy and pastable answer 3) I can use co-pilot to start a code block with a comment that says what I want to do, and Co-Pilot suggests a code block, and I can hit “Tab” to accept it or peruse other options. 4) I can copy and paste the code I wrote into ChatGPT if it doesn’t work and get a working sample back (most of the time) 5) I can use VSCode extensions to interact with ChatGPT and get Optimization, Comments, Test coverage, and more. It doesn’t take much convincing that the 2nd option is supercharged and awesome, and that is why I say todays atmosphere for developers is unlike any other. Sounds too good to be true. Whats the catch? The catch is that you have to know how things work, how to structure questions to the AI, and how to test what it gives back. There will no doubt be hustlers out there who use ChatGPT to come off as experts on something they know nothing about, only to be left in the dust when it gets to a point where they don’t even know how to articulate the problem or scenario and are out of AI-friendly assists. I am definitely still saying that there is no better time to START, but also saying that it makes you more efficient and able-bodied to do the work, not magically a wizard. Time will tell how people evolve to AI and working alongside it (who knows, it may be outlawed like they did in Italy)… but one thing is for certain if you don’t know how to leverage AI for your development work, you will be worse off than if you did. Also if we take a look outside the box of salesforce for a moment, adopting this AI assisted development style in your work is something that opens up doors to other technologies and other platforms, which is a MUST for developers to have knowledge of to stay relevant. My Python output exploded with AI assisted tools, and the amount of devops related automation, and cool script ideas that were able to become prototypes in a very short time is insane. Honestly it’s moving so fast now, that new tech comes out every day and projects are piling up. (Side projects, nothing with a deadline). It’s a good problem to have but it’s crazy to think that some ideas get left behind to start new ones when the original idea was something in and of itself a great idea. For example my Content Generation tool I mentioned in the beginning of the post… it was crazy how much time I spent creating that tool that would generate 1 or 2 sentences, and that was cutting edge at the time. Now today, you can “literally” take a picture of your refrigerator insides and it will process the images, figure out what they are, and figure out recipes based on what is in your fridge. In one second!!. That is insane! But that is the type of advancements that are happening that open up these new doors. Whether its Generative (Text generation, answer generation) or Image/Video processing, or any other area of the new AI craze, there are ideas waiting to be discovered as soon as people realize it’s possible. So from a catch perspective, the catch is you have to keep up to date with it, and you have to be willing to dive in on the development side, in order to fully appreciate what is happening when these companies like openAI release updates. I suppose the catch also is that you have to be fascinated by it for me to actually think that someone would be willing to spend so much time learning it. It would be a lie if I said I did anything other than study AI the last month or so, but it was not a chore, it was fun. If you don’t get excitement out of it, then I wouldn’t think that any of this would resonate and it’s probably not relevant to you. While I strongly discourage ignoring AI, that is your choice if you don’t want to embrace it and I don’t think you will be unemployable if you ignore AI. It will help you in many many ways though, and the pros outweigh the cons by a huge amount. How can I get some exposure to Salesforce Development and AI? I have decided to write some blog posts about it to hopefully teach some people about it, and also do what I can to support the Salesforce community and learning the topics. It is something that will revolutionize all kinds of jobs, but luckily I am not part of the camp that thinks it will steal them. I think as with any other situation in a competitive job market, the people who choose to embrace it and learn it will be lightyears ahead of those who don’t, and people who have these skills in their toolbox will be a no-brainer to select for companies over someone who doesn’t care to embrace it. I am still learning more everyday myself, and something that makes a knot in my stomach is the fact that the stuff I am learning today may not even be relevant in a couple of months. It’s too hard to say though, but I actually personally dealt with this fast moving knowledge being a problem which I won’t bore you with here. Basically I thought content generation a year ago would be the “Next Big Thing”, spent months and months working on something, and then was 1-upped by chatGPT in the worst way. That was kind of depressing but it made me even more eager to learn all that I could about AI and how it might work for a Salesforce Developer in their favor, and I am happy with the progress. If you keep checking my blog, I will do my best to put into words the last month or 2 of hacking away at learning this, and some of the prototyping and cool shit that came out of the AI obsession. If you aren’t intrigued yet, then here is an AI generated cat… which has been one of my favorite parts of learning AI. The image generation is fantasticly creative and if you want some information on generating images and prompts check out this post about Stable Diffusion.
Salesforce Development and AI – No Better Time to Start
