I guess everyone who is reading this post is using ChatGPT to some extent, right?

Did you know there are other AI-powered models, equally powerful as ChatGPT, that can significantly boost your productivity in both personal and professional life?

In this post I’m going to provide my take on the most common AI models out there, based on my personal experience with them. I will also add a short section for each where I’ll tell you how you can leverage this AI engine as a product manager.

Now, unlike my other posts, this is not an evergreen post. With the rapid pace of AI advancements, much of what I cover here could become outdated within a month or two.

I’ll do my best to keep you all up to date in future posts, but my main focus is still going to be evergreen content that will serve you for many months to come.

With this disclaimer in mind – let’s go.

The main AI engines which provide a textual chat interface

I’m going to present here the key players when it comes to AI engines which are available to the public, while focusing on those offering a chat-like interface.

Let’s go over them.

ChatGPT

OpenAI’s ChatGPT is by far the most popular and widely used AI engine. As of this date, the most commonly used model of ChatGPT is o4. 

While there are also more advanced models, such as O3, these are more expensive and only accessible on the Pro tier (the pro tier is a new tier above the premium one).

The free and the premium tiers provide pretty much the same feature set. The difference is mainly around the quotas of usage. From a product management perspective – ChatGPT is a great example of PLG done right.

ChatGPT Main capabilities

• Information retrieval – you can ask ChatGPT about anything factual, and most of the time it will provide you with the right answer. It prefers reverting to its training set, but if it can’t find the answer there, it’s equipped with web browsing capabilities and will try to retrieve the answer from there.
• Content generation – you can ask it to generate posts, songs, short stories and more. It will do a decent job here, even though it’s not the best for that.
• Image generation – it can generate an image per request, by delegating your request to its ‘buddy’, Dall-E 3, which is another OpenAI product. 
• Data analysis and text summarization – It can analyze spreadsheets, PDFs, documents, and images, summarizing or reasoning about them to provide insights. The quality is good, though again – this is not the best engine for this task.
• Real Time voice chat – if you are not in the mood to type and you are a premium user, you can simply switch to voice mode and have a real time conversation with GPT. It’s super entertaining at family dinners or when you are with your buddies. You can also choose from various voices and even select the gender.

What Sets it Apart?

ChatGPT has some unique features for power or advanced users that make it shine even more. Here is the list:

1. Memory – As of this date – ChatGPT is the only engine which supports true memory. You can ask it to memorize facts or just notes about you or anything that you want it to remember and reuse in other conversations. This is very useful when you want it to answer in a different manner and/or know things about you that it can leverage in each new conversation without ‘teaching’ it again. For example – how critique should it be when you ask it for opinion, or to consider your profession when providing answers.
2. Custom GPTs – for repeated types of tasks – you can build your own GPT and either keep it to yourself or share it with others. This is very useful for your work. For example – I have a custom GPT that generates specs according to my template (I need to edit them, of course, but it saves me the boilerplate work). I also created a GPT that generates podcast episode previews from posts I write.

ChatGPT for product managers

ChatGPT is like a Swiss Army Knife. It can do a lot of stuff, and it’s doing a decent job for most of the tasks. However, for product managers, who don’t need image generation but do need strong research and analysis capabilities – there are better engines out there (and I’ll cover them in a second).

That being said – the ability to build custom GPTs (like I did) – is super strong, and I know many product managers are using it.

Claude

Anthropic’s Claude is another powerful contender in the AI space. Currently at version 3.5, Claude stands out primarily in two areas: content writing and coding. While it doesn’t have web browsing capabilities like some of its competitors, it compensates with exceptional analytical abilities and precise code generation.

Claude’s Main Capabilities:

•  Content Creation – Claude excels at writing tasks, producing high-quality, well-structured content that often requires minimal editing. It also excels at ‘getting into the head of a given persona’ and adjusting its writing style accordingly. At this time of writing – it does it much better than ChatGPT.
• Programming, Development and prototyping – When it comes to coding, Claude is considered the best. It can handle complex programming tasks, debug code effectively, and even help with quick prototyping. Here is a short & effective Youtube video that shows a great example in action.
• Image Analysis – While Claude can’t generate images, it can analyze them thoroughly, providing detailed descriptions and insights. It does it better than ChatGPT.
• Document Analysis – It can process and analyze various document types, and also data sets, and is considered better at this than ChatGPT.

Claude’s limitations

• It doesn’t have memory. As mentioned before – Claude doesn’t retain context between conversations. While it recently introduced the concept of ‘Projects’, which retains the context within a given project, it can’t recall instructions and context that span between projects, such as some facts about you or the style you want it to maintain in its writing.
• It doesn’t connect to the web. You want to analyze a url for you? Tough luck. It can’t do it. Yes, if you are determined you can use some extensions and ‘teach it’ how to do it, but for most people – this requires too much effort. I guess this is just a matter of time, but for now it’s a strong downside.

Claude for product managers

Most product managers I know work mainly with ChatGPT. While the custom GPTs feature is amazing, for many of the other PM responsibilities, Claude will probably be a better fit.

I’ve seen some amazing stuff being built by Cursor AI (an AI development environment) when connecting it to Claude. I’m talking about full functioning prototypes built by product managers with no engineering or coding background.

Thus, for product managers I’d definitely recommend Claude over ChatGPT when it comes to:

• Prototypes development or interactive demos (in validation phases)
• Data analysis (analyzing and summarizing users feedback for example)
• Content generation (writing posts or composing important emails)

Perplexity

Perplexity brings something unique to the table – it’s essentially an AI-powered search engine. What sets it apart is its ability to perform deep web access, making it particularly valuable for research tasks.

Perplexity’s Main Capabilities:

• Research Assistant – While ChatGPT will search the web if it has no choice, it will always prefer relying on its training set. Perplexity, on the other hand, ‘lives’ on the web. It can dig deep into web content and provide comprehensive, up-to-date information.
• Limited Image Handling – While it can work with images, both generation and analysis are handled through a deferred process (via GPT or Claude, depending which of them you connected to Perplexity, if at all).
• Limited Memory – It maintains some context within conversations, though not as extensively as ChatGPT.

What Sets it Apart?

• Perplexity can function as a search engine. In fact, I replaced my Google search engine with Perplexity. It means that each time I write something in the search bar – it invokes Perplexity rather than Google. It’s a different approach on how to retrieve information from the web, and I personally believe it’s the future.

Perplexity limitations:

• Its built-in models are a bit limited. While Perplexity maintains its own LLM models, they are not as powerful as OpenAI’s or Anthropic’s ones. You can ask Perplexity to use ChatGPT or Claude instead, but from my experience, it will significantly slow down the time to get a response (from 2-3 seconds to 20 seconds or something like that). If you have the time – great! If not – then better stick to its built-in models (which are good enough for most search queries and will also probably improve over time).

Perplexity for product managers:

• I think all product managers should replace their default search engine to Perplexity. If you don’t feel comfortable saying goodbye to Google, you can do it in one dedicated browser on the machine that you use for work. Perplexity will help you with all your research tasks about competition and other product research related tasks.

NotebookLM

Google’s NotebookLM takes a different approach. It’s designed to work with a specific closed list of sources and excels at creating structured content from them. It means that unlike other AI models, it doesn’t suffer from hallucinations, because it’s basing its ‘factual’ world only from the sources you provided in each notebook.

NotebookLM’s Main Capabilities:

• Closed list of sources – you can add pretty much any url, Youtube video, Google Drive files and file you’ve uploaded from your machine. It can digest dozens of such sources in each notebook.
• Once you’ve uploaded all of your sources, you can leverage its chat interface to ask pretty much anything you want about the information contained within. It will provide accurate answers and with references to the statements it writes in its summary.

What Sets it Apart?

• Auto-podcast Generation – A unique feature that can turn written content into podcast format. This is an amazing feature that you must try. It will blow up your mind, especially now that they added for you the ability to ‘join’ live to the generated podcast.
• FAQ Creation – Specialized in generating comprehensive FAQ sections.
• It’s free!

NotebookLM’s limitations:

• Tedious process for adding sources. I couldn’t find an effective way to provide it with a large amount of sources at once. For example – for me it’d be ideal to provide it with a link to the knowledge base I’ve written over the years, have it crawl & index it, and then to provide FAQ or answer any question about my content. Sadly, it can’t do it (for now).
• Its chat answers, while accurate, are not at the same level of depth or insight as I’ve seen with ChatGPT or Claude. This may change once they integrate Gemini 2.0.

NotebookLM for product managers:

• In theory, NotebookLM could excel at analyzing customer interviews and extracting common insights. The problem is, as I just noted, that the model behind it is not as strong as its competitors. When trying to do that – I was quite disappointed with the results. ChatGPT and Claude did a much better job at generating such a summary in a requested format. I guess this will change in the future, so keep this tool close to you.

Gemini 1.5

Google’s AI model shows promise but is still finding its footing. 

To be honest – I didn’t spend a lot of time with it, as I wasn’t impressed with the answers, and all my needs were covered by ChatGPT, Claude and Perplexity.

That being said – I’d like to put a very big disclaimer on this. While I was not impressed by Gemini 1.5 in any way – Google announced Gemini 2.0 a few weeks ago. They released a very thorough Youtube video covering their new model’s capabilities, and… it is very impressive. I couldn’t access it because it’s still in beta and not available for me (last time I checked) – but according to some initial tests – it beats GPT’s 4o and also provides amazing image generation and real time voice chat capabilities.

Therefore, I won’t cover Gemini 1.5 here, as it will soon be superseded. I’ll cover 2.0 when it becomes available for me.

One final note about Gemini 1.5 for product managers:

Here you can see a great example of PLG done wrong!

When I was trying to generate images using this engine, it only allowed me to generate images of anything which is not human, and informed me that if I want to generate images of humans I’ll need to be on the pro version. The PMs behind this behavior need to rethink their strategy and learn something from OpenAI’s product team. True PLG lets you experience all features, while putting limitations on the quantity of the usage, rather than hiding  some of the features behind a paid tier. This way – users know exactly what they will be getting once they upgrade. For more details on this – read my post about PLG here.

Grok

Grok is the AI model released by X (ex Twitter). While not as strong as its competitors when it comes to reasoning, data analysis and content generation, it has one very strong superpower – it has direct access to all the real time data of the X social network.

I haven’t spent a lot of time with it, but my peers who have tried it told me it’s an excellent source if you need real time, up to date, information.

Grok’s Main Capabilities:

• It can provide you with answers about your queries the same way ChatGPT and Claude does. However, unless your questions are aimed towards what’s happening right now, based on X network – then expect inferior experience than its competition.
• Image Generation – Can generate images, though the Grok’s engine is not considered one of the best right now. There is an advantage for this engine, though, and I’ll discuss it in a second.

What Sets it Apart?

Grok’s super powers are around the fact that it’s much less restrictive. It doesn’t have the ‘safety’ limitations that the other chat engines have so you can ask about more controversial stuff.

Aside from that – its image generation engine can generate ‘deep fake’ images of known people, something that other engines won’t allow you to do.

Grok for product managers:

At its current state, I don’t see much usage for Grok when it comes to product management, unless the current way you consume your industry news is based on the X network.

Summary

Each of these AI engines has its sweet spot. If you’re primarily doing research, Perplexity might be your best bet. For coding and content writing, Claude often provides the best results. ChatGPT remains the most versatile with its memory features and custom GPTs, while NotebookLM excels at working with specific sources. Gemini 1.5 shows promise but might need some time to mature fully.

Remember, the key is not necessarily to stick with one tool but to understand which tool works best for your specific needs. Sometimes, you might even want to use multiple tools in combination to achieve the best results.

Conclusions for product managers

My recommendation for product managers depends on the budget your company is willing to allocate to you for improving your productivity.

If your company only allows you to be on a premium tier with one vendor – then I’d recommend going premium on ChatGPT, as it’s the most versatile engine out there at the moment.

However, if your company is more lenient on the AI budget – I’d definitely go premium on Claude, Perplexity and leave ChatGPT as the third priority. To be honest, though, each premium tier for any of the vendors is currently $20/mo. Hence, for $60/mo you can boost your productivity by an order of magnitude. For most companies – this should be a no brainer decision.

By leveraging the unique strengths of each tool, you can tailor your approach to the specific challenges of product management.

Good luck!

—

If you liked this post – don’t forget to ‘like’ it so I’ll know you found it useful.