When it comes to software specific agentic AI, there are more then can be listed under the sun. Even I myself have gone through a myriad of AI software that I have used both personally and at work to help enhance software development process and experience.

Though, I am not writing this solely for my own musings, but as I have been asked by friends, colleagues, and mentees alike what AI tools, programs, and services should they be investing time in to help improve their standing within the tech space, specifically the novel field of agentic AI.

I’ve created a list from what I believe are specific agentic AI tools that are currently being used within both the personal and professional developer space, as a means to answer that question.

• Cursor: Cursor is the definitive IDE choice of any bleeding edge company today, due to the fact that it was the first IDE that had agentic coding built within, along with paving the way for many Agentic IDE tooling, such as the ‘.rules’ folder and integrated MCPs as we know today. Cursor also claims to have more than half of the current Fortune 500 who use its services; which should make it a no-brainer to atleast learn how to use.

• Langchain or CrewAI: So I put these two on the same bullet point because they are both LLM Multi-Agent orchestration frameworks written in python. I feel that out of the two Langchain is more popular due to being the early bird (such as most of the AI tools on this list), but I have also worked on projects that involved using CrewAI. These two enable more control over the specific agents within the orchestration by providing either a full-code solution (in python), or a low-code modular business solution.

• Bolt/ Bolt.new: Bolt is one of those tools that make you scratch your chin while saying “huh, interesting.” I can’t easily describe what Bolt does, besides being able to create web applications through the use of agentic AI. I mention this on the list because I have seen it be used by designers to give devs a blueprint of what they would want the final outcome of the wireframes and design systems they had built in Figma, into a PoC application with React. It’s an interesting tool, and looking at the website now it seems they have added more tool integrations. I’d say it’s best to be familiar with Bolt as a concept, but I wouldn’t expect you as a developer to know how exactly to use it.

• GCP’s Vertex Workbench: If you are serious about training LLMs for whatever use case is presented to you, whether that be for pleasure or for duty; you should invest time in learning Vertex. Yes, I know other cloud providers have AI training platforms that they provide, but Vertex is on another level. Vertex Workbench is the chef’s kiss because it comes preloaded with all the libraries that one would need for a managed VM instance given as a “Jupyter Notebook”-like environment. I have trained an LLM, created benchmarks, and performed evaluation training all in a single albeit powerful Vertex Workbench instance; that was all wrapped neatly in a .ipynb file and shared internally. I do understand that there is also Azure OpenAI and Amazon Bedrock, but in my opinion they are better at data-integration and data-governance respectively.

• Mintlify: Mintlify is one of those tools that abstract from the painstaking process of creating application documentation. It easily creates a documentation page that can be centered towards end users, or more robust for internal development and business personnel. I have seen it being used more frequently by businesses to help move along the process of internal tool documentation, but it also includes special tooling to enable that documentation into an MCP server. This enables other devs to have instant access to the documentation directly as they are working within the codebase, preferably through an Agentic AI IDE, such as Cursor, to enable quicker development flows towards that particular codebase.

• Copilot: For as much slack as Copilot gets, I don’t think it’s warranted or even deserved. Copilot is a Microsoft offering that comes in the complete package of other offerings within the MS-Suite of applications. I see it in the same vein as Gemini, sure its agentic modes aren’t going to be up to par with Cursor’s, but there is still value in learning it because Microsoft dominates most brick-and-mortar enterprises whether you like it or not.

I believe this list is an accumulation of whats currently hot within the development space today, but again this is from the overall general agentic AI software development space. If there are any additions you’d like to recommend or feedback you’d like to give, feel free to either write it directly here as a comment, or message me personally.