Are you a ChatGPT business or Copilot company? (It’s a trick question)

With Microsoft’s decision to open up access for all businesses to add Copilot to their Microsoft suite, many companies will be wondering whether they should pay for AI directly available in Word, Outlook and PowerPoint or persist with trying to get people to use ChatGPT well.

And the company behind the AI large language models used in both platforms — Open AI — hasn’t made this a simple choice.

Until recently, if you wanted your business to be using the paid version of ChatGPT (now called GPT Plus), each individual needed a licence and there was limited sharing within teams.

This was likely a misstep.

Even the most basic SaaS app has an option for a set number of seats to encourage SMEs to buy-in to their platforms — but not the most powerful AI language model in history.

Because everyone had to sign up individually, many of the clients we work with have one, maybe two people with a paid licence (for the lofty sum of about $30 a month), while everyone else uses common or garden variety ChatGPT.

It’s amazing how many senior leaders of businesses, who are instrumental in the procurement and investment decisions of their companies, have not seen the difference between paid and free ChatGPT (and in some cases, haven’t tried AI at all).

In fact, much of the reluctance and suspicion we see about the use of AI comes down to people having poor experiences with the off-the-shelf ChatGPT, which is less reliable, more prone to hallucination, and struggles to keep ‘memory state’ given limits on tokens.

That’s a fancy way of saying, it’s good … but not great.

A team version designed for SMEs changes that dynamic.

ChatGPT for Teams was finally released in January, nearly a year after the first paid licences were made available.

Now, you can move all your people on to GPT-4 with a centralised dashboard for admins, and a single procurement pathway.

Most importantly, you get additional ‘context’ room, which means you can upload larger documents or spreadsheets than the free version can handle, and get it to analyse the content or pose it questions about the text.

It can now remember more for longer, so your conversations can also be far more detailed and illuminating.

You get the latest DALL·E 3 model for image generation, which can be used directly in the chat —give it a couple of dot points, for example, then say ‘based on this information, please visualise this as an infographic for my PowerPoint’.

It comes with the advanced data analysis plug-in built in, allows web browsing, can view and describe images, and can respond to verbal requests.

You can keep your data separate from training models by default.

And best of all, you can create customised small GPTs for your context and share them across the team.

This means you can take a very specific piece of information and turn it into a little ChatGPT of its own.

For example, if you are a government department or company with a hefty style guide, make a Style Guide GPT so you can ask ‘please review this text and check if anything conflicts with our style’.

What about price?

ChatGPT for Teams costs slightly more than GPT Plus – an extra $5/per person each month (about $35 in total), which is probably worth it for the simplified procurement and GPT sharing features.

But that price now makes ChatGPT for Teams not far off the additional cost of upgrading a Microsoft Enterprise licence to include Copilot, so let’s look at that option in more detail.

Copilot has the advantage of being directly embedded in the tools most people use each day — Word, Outlook, PowerPoint, Teams, and (although still in development) Excel.

That alone reduces the challenge of rolling out AI technologies as it simply becomes a little button that can be clicked whenever needed to provide writing support, ideas or summaries, rather than needing to jump platforms.

At the same time Copilot is a bit more intrusive than I, at least, was expecting.

When I start a document, before I even write a word, Copilot is offering to help. It pops up and offers to coach me on my emails. It’s kind of the AI to be so worried about the dozens of emails I write a day, but not necessary most of the time.

More helpful is the option to improve the searching and summarising tasks that dog anyone who has years of emails, Teams messages and notes scattered across their channels.

Again, it’s not perfect, but it can read through a long email chain and summarise key points, or take a pretty good transcript from a Teams meeting and allow you to query what was said, by whom and how this contributed to different outcomes.

A key concern by business leaders is that by supercharging search it will surface documents that shouldn’t be shared — that’s not been my experience to date, but anything that helps me wade through my OneDrive files is a positive.

But all this functionality comes at a cost.

A Microsoft Business Plan able to be upgraded to Copilot is $32.90/per user per month while Copilot is a further $44.80 per month with a minimum 12-month commitment.

For the best value, you need to be a serious Microsoft user, who primarily uses Outlook for email or Teams for communications, and who has documents in OneDrive and SharePoint, but if it saves each team member an hour or more a month, it could quickly become worth the cost.

So ChatGPT for Teams of Copilot for Microsoft — which should you take?

The long answer will come down to whether you want an AI boost in productivity with the lowest possible risk (Copilot) or an AI burst of speed that can make your business more competitive and creative (ChatGPT for Teams).

Each of these requires a different implementation strategy, different training and different expectations for the time-to-value for your users.

The short answer, though, is ‘why not both’?

Don’t wait on the development of an organisational roadmap to get pilot groups moving on AI.

Put the back office on to Copilot for anything with a standard operating procedure.

Move your innovators and experimenters onto ChatGPT for Teams.

And train everyone — right down to the most junior staff member — to be flexible about working across AI platforms as that’s the best way to future proof your investment.

Can we help? We work daily with clients to help them better understand the fast-changing AI landscape, and can train your teams in what AI can do, how to build their confidence and the protocols you need to have in place to experiment on AI safely.

If you would like to chat about your company’s circumstance, please reach out to Ruth on rcallaghan@canningspurple.com.au