AI Productivity Suites Explored (Part I): A Deep Dive into ChatGPT's Offerings & Plans
This mini series compares the three "AI Productivity Suites", namely ChatGPT, Copilot, and Gemini, starting with a deep dive into the plans on offer from OpenAI's ChatGPT
Writing for me remains very much an art, and this means that how inspired I feel at any point in time plays a big role in shaping my articles. This remains true even as most of my writings centre on Strategy and AI, both of which are business-related and practical topics. All of this to say that I took a couple of weeks break from writing as I’ve been on something of a creativity drought!
I’m glad to be back this week however, and exploring a topic that I get many questions about, namely, which AI productivity suite should I be adopting? To be clear, there are many, many AI tools in the market, but in my mind only three of them that are sufficiently broad and deep in terms of their capabilities to qualify being called a “suite”. These are the paid versions of OpenAI’s ChatGPT, Microsoft’s Copilot, and Google’s Gemini.
These three suites incorporate a general purpose AI chatbot, accompanied by a swathe of other features, such as multimodal capabilities (i.e., the ability to interact with and generate outputs not just in text but also in voice and images), plugins (i.e., accompanying apps that expand the utility of the base chatbot, the ability to customise your own chatbots (in the case of ChatGPT and Copilot) etc.
Yet with how rapidly the market and these AI suites in particular have been evolving, both individuals and companies have understandably struggled to keep pace and to fully comprehend the ins and outs of the various products. Furthermore beyond different tiers and plans, these AI suites have frequently undergone confusing rebrands. For instance, Google’s Gemini was previously known as Bard, it’s Duet AI tools will soon be rebranded as Gemini for Workspace, and Microsoft’s Bing Chat is now simple called Copilot, the same name used for more than a dozen other Microsoft products. Talk about mind boggling!
I’d initially set about writing this as a single article but soon realised that I would need to break up the content. This is therefore the first of four articles, the first three of which will explore ChatGPT’s, Copilot’s, and Gemini’s capabilities and limitations in-depth, followed by a fourth aimed at providing specific recommendations on which AI productivity suite depending on the type of user you are, as well as which decision-making factors are more critical for you.
OpenAI’s ChatGPT
OpenAI is the undoubtedly the current leader when it comes to Generative AI, and this reflects in the capabilities available through its free and paid offerings:
ChatGPT (Free): Provides basic chat functionalities in 50+ languages, is ideal for general inquiries and learning, and is the entry point into AI for many users;
ChatGPT Plus (US$20/month): Affords improved intelligence and accuracy and expanded capabilities via web search, image analysis, personalised chat experiences, and the ability to create customised chatbots for personal workflows;
ChatGPT Teams (US$30/month): Offers all of the capabilities of the previous tier and allows teams to more effectively collaborate and develop organisation-specific chatbots and workflows on the platform;
ChatGPT Enterprise (US$60/month): Equips users with all of OpenAI’s publicly-available capabilities, comes with enhanced service reliability, and affords full control over one’s own data.
To help you understand whether these services are right for you and / or your business, let’s now delve in more detail into each of them.
#1 ChatGPT
While our focus will be on the paid versions of each AI productivity suites, it’s worth describing at a high-level the capabilities of the free and basic chatbot from each of Microsoft, OpenAI, and Google.
The free version of ChatGPT, based on OpenAI’s GPT-3.5 model, was launched in November 2022, and was the groundbreaking, first publicly-available Large Language Model (LLM). It is a powerful general purpose chatbot and supports natural language conversations in over 50 languages.
However, while OpenAI has been continuously updating the capabilities of its free chatbot - for instance, it now hallucinates (i.e., makes up inaccurate answers) a lot less and is also much better at math and logic problems than previously - the free version of ChatGPT arguably now seems “quaint” in a market populated by super intelligent models such as GPT-4, Claude 2.1, and Gemini Ultra.
#2 ChatGPT Plus
At US$20 per month, ChatGPT Plus is the version of OpenAI’s paid subscription plans that is targeted at individual users.
ChatGPT Plus: Features
Subscribers of ChatGPT Plus, are able to access GPT-4, the company’s best-in-class and market leading LLM. Since its launch nearly a year ago, GPT-4 continues to outperform all other LLMs across a wide variety of tasks from writing and reasoning, through to summarising and coding. If you’re interested in learning more about how LLMs are evaluated, feel free to refer to this article.
ChatGPT Plus also offers a number of other capabilities that are now considered ‘hygiene factors’. On reflection, I find it more than a little wild that these features are now seen as being table stakes because most of them only become accessible to the public no more than 6-9 months ago!
Searching the web;
Analysing and generating images based on text prompts;
Supporting voice instructions and conversations;
Chatting with and querying documents.
Beyond these capabilities, ChatGPT Plus also offers a number of unique features:
Custom Instructions: Supports personalisation of interactions with ChatGPT by allowing users to provide context about their personal and professional life, and to specify response preference (see this article on how the Custom Instructions feature can help to improve the quality and relevance of your ChatGPT conversations). For instance, my preference is always for more detailed answers and for ChatGPT to respond in British English;
Advanced Data Analysis (ADA): Enables users to analyse data, upload and download files, and interpret and run Python code. For most users, the core utility of ADA will be to have it play a role similar to that of a data analyst or junior data scientists to analyse and visualise data, all using natural language instructions.
Plugins: Applications that expand ChatGPT’s capabilities and allow the chatbot to interact with third-party applications. Given the overlap in capabilities between plugins and GPT Builder, OpenAI has announced that it will no longer be possible for users to install new plugins or create new conversations with existing plugins from 19 March, with a full phase out expected in April.
GPT Builder: Enables users to create their own customised versions of ChatGPT, known as “GPTs” to support their personal workflows. GPT Builder allows users to guide the behaviour of the chatbot in three ways: with pre-defined prompts, by providing access to a knowledge base (this is essentially documents that you upload and allow the GPT to reference as part of its response), and by accessing external applications via APIs. You might need some training to build a GPT, but technical knowledge is not required as nearly all instructions can be provided using natural language. The bots that I have built for myself are now a big part of my personal workflow and include an email assistant that drafts notes in my tone and style, and a bot that summarises daily news events relating to AI.
GPT Store: Offers access to the tens of thousands of publicly available GPTs that have already been created by other users. As you would expect there is a vast range of quality on display with the 150,000+ (as of early Jan 2024) public GPTs. User feedback however, is a fairly accurate indicator of utility, and I have found some truly impressive public GPTs out there. Examples include DesignerGPT, a GPT enabling users to design, build, and host websites, and Logo Creator, a GPT that can create amazingly good logos. All GPTs are currently free-to-use, but OpenAI has announced that a commercial model alongside revenue sharing (similar to Apple’s AppStore) will soon be coming into place.
There are however three considerations that potential ChatGPT Plus subscribers should be aware of before committing:
Message cap: There is a message cap of 40 messages every 3 hours when using the GPT-4 model after which users either need to switch to the free GPT-3.5 model or else wait until the cap resets itself. My workaround is to make sure I use the GPT-4 model only for queries that require its specific intelligence or capabilities, and to use the GPT-3.5 model for run-of-the mill tasks. This limitation is partially addressed in ChatGPT Teams and fully eliminated for ChatGPT Enterprise users.
Data privacy: OpenAI has become clearer about its data privacy and retention policies over the past year. ChatGPT Plus offers users the option to switch off “chat history and training” so that data in your conversations will not be used to train OpenAI’s models. However, such chats will still be retained for 30 days, which depending on a company’s data security policies, may or may not be acceptable. Any conversations that have been saved as part of chat history and which are subsequently deleted will also continued to be retained on OpenAI’s servers for 30 days thereafter. The only OpenAI plan which allows full control of one’s own data is ChatGPT Enterprise.
Workflow integration: Unlike Microsoft’s Copilot and Google’s Gemini which provide full integration with Microsoft 365 and Google Workspace respectively, ChatGPT is obviously a standalone platform. This can be a downside for some users, especially those whom are more used to working entirely within the Microsoft or Google ecosystems. Personally, I have not found this to be a major issue, especially since I now find the majority of my workflows now taking place within the confines of tools such as ChatGPT and Perplexity (a chatbot that specialises in AI-enabled web searches and whose capabilities you can read about here).
Service reliability: In the initial months of 2023, ChatGPT frequently experienced service outages due to the popularity of its services. Since then the company has invested significantly in expanding server capacity, and incidences of unplanned downtime have fallen. Responses may sometimes be more sluggish during peak periods.
ChatGPT Plus: Verdict
In general I am a big fan of ChatGPT Plus. ADA alone has been a significant game changer for the analytical side of my work, not to mention the ability to build personalised bots in the form of GPT Builder.
With OpenAI being the clear leader in the Generative AI game, subscribing to the paid versions of ChatGPT (Plus, Teams, Enterprise) means that you’ll always have access to the latest features (if this is something that is important for you). I’m also constantly blown away by OpenAI’s ability to ship new features on such a frequent basis, and in doing so, never failing to impress.
As an individual user, as long as you are comfortable with your chats being stored for a minimum of 30 days on OpenAI’s servers and are not precious about integration with an existing productivity suite (e.g., Microsoft 365, Google Workspace), then ChatGPT Plus is a definite no-brainer for you.
For those thinking about whether to upgrade from the free version, the difference in intelligence and capabilities between GPT-3.5 and GPT-4 is like night and day. I can only assure you that once you’ve tried GPT-4, there’s really no going back!
#3 ChatGPT Teams
ChatGPT Teams costs US$30 per user per month (or $25 per user per month if paid on an annual basis) and is the version of OpenAI’s paid subscription plans that is targeted at teams and small businesses. A minimum of two users is required for a subscription.
ChatGPT Teams: Features
This tier includes everything in ChatGPT Plus, and additionally also includes:
Higher message cap limit: Offers a GPT-4 message cap of 100 messages every 3 hours (as opposed to 40 messages every 3 hours). A message cap of 100 messages every 3 hours equates to a whooping 30 messages per hour, and is really about as much as even a peak user could ask for!
Team-specific GPTs: With ChatGPT Teams, users are able to create and share custom bots (GPTs) that are specific to their team’s workflows. For instance, a Marketing team might create a specific chatbot that is able to replicate the tone and style of their company’s brand for use in marketing campaigns, while a Finance team may develop and share a GPT that is able to support with the creation of monthly performance reports for each business unit.
Data excluded from model training: All chat data will not be used to train OpenAI’s models by default. This is as opposed to ChatGPT Plus, where users have to specifically switch off “chat history and training” for their chats to be excluded from OpenAI’s training data. All chats will continue to be retained on OpenAI’s servers for 30 days.
ChatGPT Teams: Verdict
ChatGPT Teams essentially does not have bring a significant upgrade in capabilities beyond those of ChatGPT Plus. The main reason for signing up for ChatGPT Teams, is if you want to collaborate with other individuals within your team (or you are a team leader that wants to encourage this), in particular leveraging GPTs that are tailor-made around your team’s workflows.
The increase in message caps is a cherry on the cake, as is having your team’s chats excluded from the data that is used to train OpenAI’s models (as this is something that can already be enabled in ChatGPT Plus). Still, the latter is a nice-to-have as having to consistently switch off “chat history and training” when working with confidential or sensitive data can be quite annoying and is something that can easily slip the mind.
The value-for-money with ChatGPT Teams is not quite as clear-cut as it is with ChatGPT Plus. That being said, I’ll go out on a limb here and say that the incremental $5 per user per month (or $10 per user per month if payment takes place monthly rather than annually) isn’t likely to break a company’s bank.
#4 ChatGPT Enterprise
ChatGPT Enterprise is the Rolls Royce of ChatGPT plans and comes with a commensurate luxury price tag. While the price is not advertised on OpenAI’s website (you need to contact their Sales team directly for this), the latest intelligence that I have from October 2023, places the cost at $60 per user per month. As of last reporting, a business would need to commit to at least 150 seats and to a 12-month contract.
Beyond the elimination of the GPT-4 message cap (i.e., users essentially have unlimited GPT-4 messages), there are no step up in capabilities from ChatGPT Teams to ChatGPT Enterprise.
What organisations do get with ChatGPT Enterprise however, is absolutely piece of mind and reliability:
Full control over data: Chat data is naturally excluded from the training of OpenAI’s models. The organisation in question is able to control all aspects of its chats and chat data, including critical to eliminate the 30 day data retention period that is common to all other ChatGPT plans.
Service reliability: The service levels for ChatGPT Plus and Teams are occasionally throttled during peak periods, resulting in either very slow response speeds or a full interruption of service. This should not happen for ChatGPT Enterprise customers as their gold-plated package naturally brings gold-plated service levels.
Platform security: ChatGPT Enterprise comes with SAML SSO. In non-tech speak this basically means that users can sign on to ChatGPT using their existing enterprise sign-on credentials, which tend to come with relatively heavier protections and makes it so that employees do not need to remember yet another password.
Platform analytics: Company admins are provided with a dashboard to analyse platform usage at the aggregate level - they are not allowed to view individual chats for privacy reasons - which can be helpful for understanding when and how employees are using the technology. A CIO I’ve spoken to shared with me a screenshot from his ChatGPT Enterprise dashboard. My view is that these analytics will merely be interesting (rather than useful and / or insightful) for all but the most data-centric companies.
Account management: Naturally, client organisations also get access to a dedicated account manager and priority support for any requests.
ChatGPT Enterprise: Verdict
It should come as no surprise that ChatGPT Enterprise is really only affordable to the largest and most profitable organisations. For these companies, especially those in sectors such as financial services (e.g., hedge funds, investment banks, wealth managers), where data security and confidentiality is paramount, the incremental $30 per user per month does make sense.
For the rest of us however, ChatGPT Enterprise is likely an overkill. After having spoken to a large number of organisations about their AI strategy and plans, I know of no more than a small handful that are ChatGPT Enterprise customers.
Given the intensifying competition coming from Microsoft and Google, my prediction is that ChatGPT Enterprise will always remain a niche product and may even ‘die out’ at some point, potentially to be replaced by a new tier on top of ChatGPT Teams which promises more control over organisational data, but without the rest of the bells and whistles in ChatGPT Enterprise.
Summary
For individuals looking to up their productivity or creativity, my unwavering advice is to go for ChatGPT Plus. It’s not by any means a massive investment, and once you’ve taken the leap to upgrade from the free version and try out its capabilities, I promise you that there really is no going back.
The decision for teams and businesses is usually more complex. Data privacy considerations take on more significance, which argues in favour of Microsoft’s Copilot (something we’ll explore in more depth in the next article in this series).
Even if you are a steadfastly loyal Microsoft 365 or Google Workspace customer however, my advice is to not rule out ChatGPT plans for your team members. Certain functions and teams, by nature of their workflows and / or having highly innovative and AI-savvy team members are likely to benefit significantly more from having access to ChatGPT’s feature-rich plans. I can assure you that offering these individuals and teams the option of ChatGPT Teams instead of enforcing a blanket “we shall only use Copilot because all our other systems are on Microsoft” policy, will pay substantial dividends in terms of your organisational productivity, insights, and creativity.
What's your take on OpenAI's ChatGPT suite? How has it changed the way you work? Drop your thoughts in the comments and join the conversation on the future of AI productivity tools!
Justin Tan is passionate about supporting organisations and teams to navigate disruptive change and towards sustainable and robust growth. He founded Evolutio Consulting in 2021 to help senior leaders to upskill and accelerate adoption of AI within their organisation through AI literacy and proficiency training, and also works with his clients to design and build bespoke AI solutions that drive growth and productivity for their businesses. If you're pondering how to harness these technologies in your business, or simply fancy a chat about the latest developments in AI, why not reach out?