COPILOT: GAPS IN THE MICROSOFT SUITE
Author: Jonathan Stuckey
Audience: SharePoint solution designer, Project manager, Change manager, Subject-matter experts.
We take a few minutes to discuss the glaring gaps in the Microsoft Copilot suite. The following is brief look at the critical types of user output, that is not addressed in Microsoft's Generative AI suite (Copilot) for Microsoft 365. Where Copilot drops the ball, there are already a swarm of niche players rushing in.
Gaps in the Microsoft Generative AI suite
While Microsoft is busy throwing stuff in their new stream of revenue-generating tools, you should take note that you will certainly have to have a collection of such tools from various vendors - if not for just for the short-medium term - if you want to do something more than email, meeting notes and text-oriented documents
The weaknesses in the suite come from the general approach adopted, but hanging the AI framework on a ChatGPT derivative with a limited scope. Microsoft's Copilot generative AI suite is deliberately broad-spectrum, which is consistent with their long-term goal of "user productivity" for everyone, and if we apply Pareto Microsoft's target is enabling the 20% of capability that 80% of users can utilise for general tasks.
The 80% which is made of uniqueness of roles, practices, processes and industry constraints - well that's always been down to the organisation solve - and so you are going to be waiting a long-while for MS to drip-feed in the enhancements which will support the gaps.
What do I mean by Gaps?
Well, there are various classes of Generative AI, and Copilot leverages a couple of big ones, but then it plays lip-service to the others. For example, if you want...
a photo-realistic picture of someone running through a crowd, while everyone looks baffled - all-good.
Yes! Copilot Chat and embedded 'Visual Creator' tool will do it.
to create UML or BPMN based diagrams from user described tasks and activities?
Hmmmmm, its back to ye-old Visio / PowerPoint tool set or get a 3rd party targeted, well developed Generative AI tool.
rapidly create something 'less engineery', like an information flow diagram, or document decision tree based on end-users description of process?
Yeah, Nah. Again, you are back in ye-old MS world of Visio and manual DIY. Or DiagramGPT etc
what about some visually compelling roadmap diagram, or project swim-lane representation for business investment case?
Nope. The phrase "cats-chance..." comes to mind. Crack open Excel, Project, Visio... or Miro, LucidChart
consolidate lots of research notes, draft articles, white-papers and some meeting recordings to create an audio podcast, or avatar based video discussion?
Errr no. You are talking never, never land from Microsoft. Look-up NoteGPT, NotebookLM etc
Microsoft has the same issue evident for anything which has a degree of sophistication in the content's development or design. Particularly anything which is flow-based, or diagramming oriented. Same goes for cohesive rich-media generation, or even depth of UX design. It's just not there.
Hence the explosion of .AI vendors creating, some frankly amazingly good services and tools e.g. examples off the top of my head...
Table: software vendors offering AI driven drawing and diagramming
What does this mean in the market?
The massive continual investment by software industry means the state of the tools and services today are only getting better. Rapidly. This will force specialisations and stand-out options for capabilities and services supping things like Generative AI over: Solutions design, Project management, Information Architecture, Engineering drawings, Architectural blueprints...
We have already started to spawn a massive number of start-ups focusing on better automated experiences using Generative AI over the '80%' of needs.
Once you have saturation of a niche, then comes the classic consolidation model: The massive number of similar vendors with so-so offerings either fail, get bought-out, or grow to be the new power-houses.
Eventually platform-vendor launch into mergers and acquisitions, where these power-houses are acquired and bolted into bigger suite and we (the end users) get a dumbed-down version of slow-releases of improved tools as the new capabilities are digested, broken-down and then repackaged.
What about Microsoft and Copilot
Microsoft grows its software portfolio by merger and acquisition. They do hundreds of these a year. The speed of development in AI and Generative AI space will hinder natural internal development (ironically - considering they offer AI enabled Software Development tools and Platform) - so M&A it is, but they'll need to do it quicker.
Where we would expect to see the type of capability delivered is from: Visio and PowerPoint products, probably via Copilot or 'Designer' integrations, and this has not yet materialised. In fact, there hasn't been a peep about this since May 2023, and there's nothing on other official channels (e.g. MS Community: MSOffice and Vision, Microsoft Copilot help & learning, Microsoft 365 Roadmap | Microsoft 365) either. The only (recent) announcement from Microsoft on Visio in it was on 5th December 2024 - on the Ignite 2024: Are you ready for Office 2016/2019 end of support? | Microsoft Community Hub.
So, assuming I'm waiting until MS gets its act together, I will take a foundation-less punt, based on wild speculation that:
PowerPoint has some awesome features coming! hmmmm no. So, unlike them if they do.
The Visio product team guys haven't seen the news and are scrambling to release? Very possible
Copilot Chat and Visual Creator (nee Designer) will drop a bombshell in May/June? Maybe... no.
Microsoft buy another tool and brand it Copilot Drawings! ... Well, this would fit the pattern.
Microsoft forgot about this part of their market and we just have to wait while the Copilot team get around to them in Q4 2025. ...the cynic in me says I can see this happening
And what about OpenAI?
Well, OpenAI's ChatGPT has taken a smart route developing a hosted pass-through model for users generating extensions, add-ons and new products which extend the core models provided. There are already options for these sorts of gaps published by independents and 3rd parties - but that's not a revenue stream available for Microsoft as its indirect - as pass through - and that's not Microsoft's preferred model.
The OpenAI approach, before DeepSeek R1(v3) spoiled their party, was to extend their models and get consumer and multi-vendor support to fill in the missing capabilities. But when you look at competitive benchmarks none of the big AI platforms talk about stats on NL translation for Diagramming or Drawing production or on complex richmedia generation. They all benchmark on Language Processing, Maths and Code.
I guess we will have to wait a little while yet to see anything here.
What does that mean for you (and me)?
If you need tools from Generative AI that deliver specialise sophisticated outputs i.e. drawing, diagramming, schedules, roadmaps, rich-media generation etc, then you are going to have to do the diligence on the range of smaller vendors / providers, keeping in mind that there will be mass consolidation soon, and a large number of these could disappear.
If it's me doing the looking I would view established vendors, with 'good enough' options as may be better than what looks like 'the best' application ever.
Be sure about what you are looking for from the Generative AI tools which you'll use to fill the Copilot gaps. Microsoft is playing catch-up, they are moving pretty fast but it is a big lumbering beast and a waiting game may take too long, if you want all you Generative AI in once place.
It is important to remember that the speed of change in this space when considering options - so don't take too long when selecting what you need - this is for right-now, and not for always.
Recommendation
Trial some of the preferred options in the marketplace, in parallel. There are already strong leaders in most of the sectors specialising in support for critical professions or roles, that would use these tools.
Ensure you have robust quick-assessment criteria to check against, i.e. include things like regional support, standards (existing, developing) support, privacy policy, data-training controls etc.
Check to see if the vendor provides basic, acceptable / standard functionality which will insulate you from vendor collapse or market consolidation:
content can be exported to common formats e.g. png, svg, pdf etc
you can access definitions i.e. includes a Mermaid definition or similar JSON-esque export
allow for the fact your artefacts that need a long-life will need to be re-written to new format
If the vendor you favour doesn't allow for an exit path, it's not a good choice.
Keep an eye on the pricing though. Many vendors are trying to recoup their research and development costs, and they are not sure about the right market price for their services at this point.
Close
If you have questions and want to understand more about the gaps with Microsoft's Generative AI suite (Copilot), and how to address them - give us a call: hi@timewespoke.com
About this article: No Generative AI was used in the creation of this article.
About the author: Jonathan Stuckey
Comments