Beyond Prompt Engineering: The Evolution of AI Products in 2024

Beyond Prompt Engineering: The Evolution of AI Products in 2024
Photo by Matt Ridley / Unsplash

Do you remember when you first tried ChatGPT? I recall the early days when it was easy to do silly, fun things, like create a song about technology in the style of Eminem, but harder to write emails and articles reflecting our natural voices. While AI applications have launched nonstop in the past two years, many were unimpressive and expensive, lacking the full set of features needed for extensive use.

Recently, I started testing exciting AI products. They’re more mature, with AI features integrated but not the entirety of the application. Finally, I’m seeing products blend AI and non-AI features, enabling users to complete certain tasks wholly within the tool, whether the features are AI-powered or not.

Not to mention all the new features launched by the AI behemoths:

  • OpenAI has rolled out Advanced Voice Mode more broadly. They also launched their new advanced reasoning model “o1,” and a couple of weeks later, the canvas feature, a UI mode for editing and refining “documents” ChatGPT creates.
  • Claude supports highlighting within artifacts for editing. Artifacts can now be published, shared, and remixed. This is great for non-tech folks to make Claude’s tools accessible without figuring out how to host software, and remixing is a fun way to encourage creativity.
  • NotebookLM launched with a fun podcast feature that got extensive media attention. It’s the first AI application that lets you access static RAG solutions for free, and the hallucination rate has been lower than competitors, making it a great tool when accuracy is important. Not sure what RAG is? I explain it in more detail later.

The New Face of AI: No Coding Required

Generative AI is no longer confined to tech circles. It’s becoming commonplace, and non-tech people are recommending it to friends in the same way Google search was recommended in the early 2000s.

Here’s how I describe the Generative AI products landscape:

llm model comparison (11).png
  1. Conversational Assistants: General chat-based interfaces like ChatGPT for various problems.
  2. Embedded Assistants: AI features seamlessly integrated into your existing tools, like Google Docs or Microsoft Excel.
  3. Domain-Specific Assistants: Tailored AI solutions for specific industries or tasks. Think Harvey for legal and Datarails for finance.
  4. AI-Native Software: Products built with Generative AI at the core that don’t require expert prompt engineering skills and may offer but don’t rely solely on a conversational interface. For example, Gamma (over PowerPoint) for presentations.

My Picks: Game-Changing AI Tools

To keep you at the forefront of AI innovation, I’ve tested the latest consumer products. Here are some standout apps that impressed me:

  1. Gamma: Imagine PowerPoint with AI powers in all the right places. It’s transforming how people create presentations and websites. My favorite features: (1) suggested AI image prompts tailored to my content and theme and (2) features to change content like “make it more visual” and “try a different layout” without dragging a dozen blocks around a slide.
  2. Cove: This app re-imagines the user interface for AI applications. It offers a canvas view (like Miro) for building and sharing editable artifacts, like tables or lists, making it a great tool for a visual way of moving across information - like planning and brainstorming. I love how it auto-completes tables as I add rows and columns.
  3. Lex: My newest writing tool. It’s like Google Docs got a major AI upgrade, making the writing process smoother and more intuitive. But be warned - if you’ve optimized writing to a conversational assistant like Claude, it’s a different setup. There is a learning curve, but I see a lot of potential and love the ability to add custom “checks” to my writing. I’m still testing this app and hope it can replace Grammarly.
  4. Claude Projects / NotebookLM: These tools are transforming research and content creation through their static RAG features. RAG improves GenAI response accuracy using provided documents for context instead of training data. It’s a great way to provide reusable details. NotebookLM excels at accuracy and reduces AI hallucinations, and it’s free. My preferred tool is Claude Projects, which is great for building simple applications and has better features and a more intuitive UX.

What This Means for You

The rapid advancement of these tools means it's easier and more impactful than ever to leverage AI in your professional life. Whether you're crafting presentations, writing reports, or conducting research, there's an AI tool to enhance your workflow. I haven't even covered enterprise tools or features.

Stay Ahead with Mellonhead

Registration has closed for “Getting More From Generative AI,” but I’m offering the same curriculum to companies wanting to provide GenAI training or managers wanting to level up their team members’ skills. If you want to equip your organization with the skills to thrive in this AI-driven world, I can help.

Ready to explore how AI can transform your work? Let's chat about how Mellonhead can bring AI expertise to your team.