Copilot’s new feature, Tuning, allows organisations to effectively teach the AI model their own playbook. Photo: Michelle Kroll.
Copilot – Microsoft’s popular AI assistant – has been evolving fast, but its latest capability may be a game changer, experts say.
Still in early access, Tuning allows organisations to shift Copilot from a clever generalist into a powerful digital colleague, capable even of adopting a company’s tone.
It takes Copilot’s use of large language models up a notch, drawing on personalised, organisation-specific data to make it better at performing repetitive tasks to your standards.
While not suited to every function, it has clear value in many, says Mojo Up director Daniel Buchanan who doesn’t see any business that won’t benefit from it.
“One good example is a lawyer or procurement officer creating contracts,” he says.
“Copilot out of the box is quite conversational, and if you give it a 100-page contract and ask it to make a new one, you’ll probably get 20 pages back because it summarises documents. Obviously, you can’t be doing that with contracts.
“With Tuning you can feed it templates and examples to tell it what it must look like or which elements it must contain. It’s an intuitive learning model, which means you can rate elements like the tone and vocabulary, improving its ability to communicate in your organisation’s style.”
The more you train Tuning, the better it matches your desired tone.
Daniel says it’s achieving what was attempted 15 years ago with RPA (Robotic Process Automation) – technology that used bots to automate repetitive, rules-based computer tasks such as data entry, form filling and report generation. While RPA increased efficiency and reduced errors, it never quite “got there”.
“This [Tuning] is the evolution of that – it’s not hard to set up and the benefits are huge, the potential is huge,” he says.
At a time of heightened sensitivity around resource spending, the technology can help do more with less, if used wisely.
“Of course, you wouldn’t let AI make contract-based decisions, but it means you don’t need a human to do all the boring drafting,” he says.
“You can see how this could be useful for everything from contracts to quotes to invoices, tenders to proposals. Everyone from tradies to accountants, every department from finance to corporate, has the kinds of repetitive tasks ideal for automation.”
Mojo Up subject matter experts are already working with agencies and businesses to identify low-hanging opportunities where Tuning can be practically implemented, and guide them on producing work that aligns more closely with company standards.
According to Daniel, early adoption is key.
“With this sort of tool, the quicker you can demonstrate its value to business, the more likely it’ll succeed,” he says.
“For larger organisations, it’ll be a case of setting up small project teams to continuously set up and drive these solutions across the business because even if you set up Tuning today, you’ll want to refine it over time as the tech evolves.”
For more information, contact Mojo Up.
