How to Use GPT-5, Claude & Cursor Together

8/10/2025

Unlocking Genius: How to Use GPT-5 with Other Tools like Claude & Cursor

Alright, let's talk about the seismic shift happening in the world of AI & software development. It’s no longer about just one tool; it’s about creating a symphony of them. The recent launch of OpenAI's GPT-5 has been a game-changer, but the REAL magic happens when you start weaving it together with other powerhouses like Anthropic's Claude & the AI-native code editor, Cursor. Honestly, it feels like we've been handed a new set of superpowers.

This isn't just about making things a little faster. We're talking about fundamentally changing how we build, write, & even think about problems. It's about creating a personalized, souped-up workflow that leverages the unique strengths of each tool. So, grab your coffee, & let's dive into how you can start conducting this AI orchestra.

The Star of the Show: What's the Big Deal with GPT-5?

So, GPT-5 finally dropped in August 2025, & the hype was, for once, completely justified. This isn't just a minor upgrade from GPT-4. It's a "significant leap in intelligence," according to OpenAI, & based on what we're seeing, that's not an exaggeration.

Here's the thing: GPT-5 is a unified system. It has a smart, quick model for most of your day-to-day queries & a deeper, more thoughtful reasoning model for when you need it to "think hard about this." A router in the background figures out which one to use, which is pretty cool. For the real power users, there's GPT-5 Pro, which has even more extended reasoning capabilities.

The performance benchmarks are just nuts. It's setting new records in math, coding, & even health-related queries. We're talking 94.6% on the AIME 2025 math competition without any tools & a massive 74.9% on the SWE-bench for real-world coding tasks. It's also multimodal, meaning it can understand images & visuals, scoring an impressive 84.2% on the MMMU benchmark for college-level visual reasoning.

But beyond the numbers, it just feels more useful. It’s better at following instructions, reducing those weird "hallucinations" we sometimes see, & it’s less of a sycophant, giving you more honest answers. For developers, this is HUGE. GPT-5 shows massive improvements in generating complex front-end code & debugging large repositories. It even has an eye for design, understanding things like spacing & typography.

And the best part for developers? It's already being rolled out in public preview for GitHub Copilot. If you have a paid Copilot plan, you can access GPT-5 right within VS Code. This direct integration is where things start getting really interesting.

The Collaborator: Claude's Unique Strengths

Now, while GPT-5 is a beast, it’s not the only player on the field. Anthropic's Claude brings a different, but equally powerful, set of skills to the table. Claude was built from the ground up with a focus on safety, reliability, & ethical alignment. This makes it a great choice for tasks where you need a high degree of trust & accuracy.

Like GPT-5, Claude 3 is multimodal, able to process images & audio alongside text. But Claude's real secret weapon is its "Integrations" feature. This is a game-changer. It’s built on something called the Model Context Protocol (MCP), which is basically a common language that lets Claude talk directly to other apps & tools.

Think about what that means. You can connect Claude to your Google Workspace, Jira, Confluence, Zapier, Slack, & a ton of other services. It doesn't just access the data in these apps; it understands it. Claude can analyze sales data from your CRM, summarize a long thread in a Confluence document, or even create Jira tickets for you based on a conversation.

This makes Claude less of a simple chatbot & more of a proactive work assistant. It's like having a team member who can go out, gather information from all your different systems, synthesize it, & present you with a comprehensive report, complete with citations. For complex research or project management tasks, this is an absolute lifesaver.

While a business might use a tool like Arsturn to build a custom AI chatbot for their website that provides instant customer support 24/7, you can think of Claude's integrations as a way to build a personalized AI assistant for your internal workflows. Arsturn helps businesses create no-code chatbots trained on their own data to engage website visitors & answer questions. Similarly, Claude uses its connections to your internal data to provide you with deeply contextualized answers & automate tasks.

The Conductor's Baton: Cursor, the AI-First Code Editor

This is where it all comes together. If GPT-5 is your star performer & Claude is your expert collaborator, then Cursor is the conductor's baton that lets you orchestrate them both. Cursor is an AI-powered code editor built on top of Visual Studio Code, which means you get to keep all your favorite extensions, themes, & keybindings.

Cursor's whole reason for being is to integrate AI deeply into the coding workflow. It's not just about autocompleting a line of code. It's about understanding your entire codebase. You can highlight a chunk of code & tell it, in plain English, what you want to change. You can ask it questions like, "Is there a bug in this file?" & it will give you a context-aware answer.

It has features like:

Intelligent Code Refinement: It suggests cleaner, more efficient ways to structure your code.
Inline Code Generation: You can generate new code from a natural language prompt without leaving your editor.
Error Correction & Debugging: It can spot & fix errors in your code automatically.
Code Base Questions: You can literally chat with your entire codebase.

The most powerful feature of Cursor, in this context, is the ability to bring your own AI models. Under the hood, Cursor uses a mix of powerful models, but it also gives you the option to plug in your own API keys. This means you can have GPT-5 (especially via the Copilot integration) as your primary code generator & autocomplete engine, while also being able to call on Claude for specific tasks.

Weaving It All Together: A Practical Workflow

So, what does this look like in practice? Here’s a hypothetical workflow for a developer building a new feature:

The Big Picture (Claude): You start your day in Claude. You ask it to "review the project requirements in Confluence, look at the recent customer feedback in Intercom, & summarize the outstanding tasks in Jira related to the 'new feature' epic." Claude goes off, does the research, & comes back with a concise brief, saving you an hour of manual digging.
Scaffolding the Code (Cursor + GPT-5): Now you jump into Cursor. Based on the brief from Claude, you start scaffolding the new feature. You use a prompt like, "Create a new React component with the following structure..." & let GPT-5, via its deep integration in Cursor (likely through Copilot), generate the boilerplate code. As you're coding, GPT-5's powerful autocomplete is predicting your next moves, often suggesting entire blocks of code that are exactly what you were about to type.
Complex Logic & Refactoring (Cursor + Your Choice of Model): You hit a tricky piece of business logic. You're not quite sure the best way to implement it. You highlight the relevant code in Cursor & ask, "Refactor this to be more efficient & handle edge cases A, B, & C." Here, you could even experiment. You could try the prompt with GPT-5 & then, if you want a second opinion, switch the model to Claude (if you've configured it) to see how it would approach the problem.
Connecting to Other Services (Claude): Your new feature needs to integrate with a third-party API. The documentation is a bit dense. You can feed the documentation URL or PDF into Claude & ask it to "explain how to authenticate & make a 'create user' call, & provide a Python code sample." Claude's ability to digest & synthesize large documents makes it perfect for this kind of task.
Customer-Facing Interactions (Arsturn): As you're building out the front end, you think about how users will interact with it. If they have questions, how will they get answers? This is where a tool like Arsturn comes into the picture for the business itself. The business can use Arsturn to build a no-code AI chatbot, trained on all the new feature's documentation, that can be embedded directly into the app. This chatbot can then handle customer questions, provide instant support, & even help with lead generation by guiding users through the new feature. It’s about creating that seamless, instant connection with your audience.
Final Polish & Debugging (Cursor + GPT-5): You're almost done. You use Cursor's AI-powered debugging tools to scan for any potential issues. GPT-5's improved debugging capabilities help you spot a subtle bug in your logic that you might have missed. Finally, you use Cursor's natural language editing to add comments & documentation to your code, ensuring it's easy for your team to understand.

The Future is a Conversation

What we're seeing is the evolution of development from a monologue, where we tell the computer exactly what to do, to a dialogue, where we collaborate with AI partners. Each of these tools—GPT-5, Claude, & Cursor—has a unique voice in that conversation.

GPT-5 is the brilliant, lightning-fast coder who can turn an idea into reality in seconds.
Claude is the wise, experienced researcher who can connect the dots across your entire organization.
Cursor is the integrated environment that brings them all together, allowing you to seamlessly switch between them without breaking your flow.

The key is to not think of it as "GPT-5 vs. Claude" but "GPT-5 & Claude." By understanding the unique strengths of each model & using a flexible platform like Cursor to wield them, you can build a workflow that is far more powerful & efficient than anything that was possible before.

It’s an exciting time to be building things. Hope this was helpful & gives you some ideas for your own AI-powered workflow. Let me know what you think