It’s the holidays, so I’m currently scouring sites like The Wirecutter to get the inside scoop on gifts. Which noise-canceling headphones are worth the money, and which will fall apart in a week?

It made me think: someone should write a “holiday gift guide” to AI chatbots. So here’s ours.

AI chatbots are not all built the same, and even the best ones aren’t best for every use case. I have my favorites (hi, Claude!), but I still think carefully about which chatbot to use when I’m approaching a specific task. 

Want to become an AI power user? Join our AI Prompt Writing workshop in December. Use code BESTLLM for 20% off a workshop seat or BESTPROMPTS for 25% off membership.

A quick side-by-side guide to the top AI chatbots

It’s easy to fall back on your one go-to tool for every task – I always have Claude open in at least one tab. But when you start choosing a tool that’s best equipped for the problem you’re solving, your outputs will get stronger. 

My AI SWAT team includes three LLMs – ChatGPT-4, Claude by Anthropic, and Perplexity. I pay $20/month (each) to access the paid versions of GPT and Claude, and use the free version of Perplexity. 

‍

I’ve experimented with others including Chat GPT-3.5, Poe by Quora, Pi by Inflection AI, Bard by Google, and Bing Chat. Below is a quick side by side comparison of all these LLMs. For more details, check out our full LLM guide that we update weekly.

The best AI chatbots for each use case 

The best LLM for thought partnering 

Claude by Anthropic

Our unfiltered take: Using Claude feels like having a natural conversation, and its capabilities to process large documents and handle different file types make it great for productivity tasks that involve documents. 

My pro tip: I particularly like uploading project briefs for new products or tests we’re running and asking Claude for feedback as my different stakeholders: our CFO, our CEO (my boss), and the wider team that reports to me.

Example use cases:

• Critique a decision you’ve made‍
• Example prompt: I have two job offers – one for the first product manager role at a fast-growing Series A startup and one for a senior product manager role managing a small team at a Series D, large SaaS firm. Help me weigh the pros and cons of each role, play devil’s advocate, and make a recommendation.
• Review a list of potential options and help you prioritize‍
• Example prompt: Act as my strategic advisor and help me prioritize potential acquisition channels. I’m evaluating five options: 1) 30-day free trial, 2) a student pricing tier, 3) LinkedIn ads, 4) Facebook ads, 5) newsletter ads, and 6) free events. Help me narrow this list to top 2 options that could drive $250,000 in incremental revenue. I have a monthly marketing budget of $10,000 per month
• Give you feedback acting as a specific persona (your manager, team, CEO, board)‍
• Example prompt: Here’s an internal memo I’ve written that explains the next three features we’ll prioritize on our product roadmap. Give me feedback as though you are the CEO, head of engineering, and chief product officer at my company.

The best LLM for data analysis 

ChatGPT Plus (GPT-4) by OpenAI

Our unfiltered take: GPT-4 is the most advanced LLM on our list right now, but also charges $20/month. Its voice is more robotic and matter-of-fact than Claude’s conversational tone, but its beta features including advanced data analysis make it the best choice for more analytical challenges. 

My pro tip: I use GPT as a quick data analyst-level way to get insights on quantitative (usually spreadsheet) data. However, I don’t recommend GPT for text data analysis – when I want insights on survey responses or Zoom chat transcripts, I turn to Claude. Also: don’t forget to anonymize your data before you upload it and remove any PII (personal identifiable information like email addresses or names).

Example use cases:

• Analyzing sales or marketing data‍
• Example prompt: The attached spreadsheet has sales data from the last 3 months, and contains the job title of everyone who made a purchase. Please summarize this data, and tell me our top 3 and bottom 3 job titles based on number of purchases. 
• Analyzing survey data‍
• Example prompt: The attached spreadsheet includes anonymized survey responses from our most recent AI workshop. Calculate NPS for the overall workshop and then calculate NPS by job title. Share the data in a table and round NPS to the nearest whole number.

The best LLM for research with citations

Perplexity AI

Our unfiltered take: Perplexity is the gold standard in conversational search engines. It excels at research that requires the most up-to-date information with citations, providing robust in-line citations (which ChatGPT just added in its latest update) and additional sources at the top of the page.

My pro tip: Enable “co-pilot” mode, which makes responses better and prompts Perplexity to ask you follow up questions (e.g. “what industry are you looking for benchmarks in”) to provide a more personalized response.

Example use cases:

• Industry benchmarks‍
• Example prompt: I want to understand industry benchmarks for SaaS free trial – what should I expect for conversion rate to free trial and conversion rate to purchase among free trial users? Include data for opt-in vs. opt-out free trials.
• Product research‍
• Example prompt: We’re developing a new product to help employees cope with burnout at work. What are some trends around employee burnout over the last year?

The best LLM for content writing

ChatGPT Plus (GPT-4) by OpenAI

Our unfiltered take: While LLMs are great for brainstorming and rough drafts, I still haven’t found one that I completely trust to write creative, nuanced copy - so at least for now, these emails are being written by a person. But GPT-4 excels at generating templated copy for simple content in a repeatable format. 

My pro tip: Create a template for content that you repeat and run regularly. At Section, we share a weekly email with our members that outlines all the courses coming up for the week. We used to fill it out manually, but now we have a GPT template that pulls in the course information right from the course calendar on our website. 

Example use cases

• Template for recurring emails‍
• Example prompt: Please help me build an email template for our weekly email that goes out to Section members. This email goes out every Sunday, and shares the upcoming courses for the week. The format is as follows: [Use an existing email to define the format, and put instructions, i.e., pull the list of courses from our course calendar, in brackets.]
• Template for recurring social posts‍
• Example prompt: Please help me build a template for a monthly LinkedIn testimonial post highlighting a positive review from one of our customers. Use our testimonial database [URL here] to source the content. The format is as follows: [One short sentence highlighting the customer and what they liked about the product.] [One short sentence that pulls a quote directly from their review.]

The best LLM to compare all other LLM results

Poe by Quora

Our unfiltered take: Poe gives you access to multiple LLMs, which is great for someone who’s familiar with different AI tools, but potentially confusing for anyone just starting out. Once you get comfortable using multiple AI tools, Poe makes it easy to switch between them all from one platform. 

My pro tip: I often get asked how to deal with hallucinations, and one of my go-tos is to give the same prompt to a few different LLMs and see how different the responses are. Poe makes this extremely easy.

Example use cases and prompts will vary by the specific tool within Poe. 

–

That’s a roundup of some of my favorites, but there are plenty more in our full LLM guide for you to test out. 

Want to get more efficient using AI? Take our upcoming AI Prompt Writing workshop. Join our AI Prompt Writing workshop in December. Use code BESTLLM for 20% off a workshop seat or BESTPROMPTS for 25% off membership.