Customer Services contribution to your business success

[Rob]

Hi all,

This topic is about your opinions regarding Customer Success department and its perceived contribution and help to your business. I would be very curious to know whether you would consider CS services as a major differentiator / competitive advantage when comparing with our competition.. I mean: Tech Support, Community and Insider Club, GSLearn courses, in-app AC onboarding, Help content and performance / help of your local Customer Success Manager (either direct GS or local GS Partner). If your business is served in any of the above-mentioned areas by a local GS partner please contemplate on it too but indicate that it's a local GS partner provided service.

The issue is stemming from a currently hotly discussed industry dilemma related to "AI hype" where AI is seen as "silver bullet" for everything... my worry is that replacing customer services with AI-driven approach would remove our uniqueness as a company when it comes to customer relationship. Having said that, I definitely see AI potential in analytics and resolution of some low-complexity, highly repetitive or trivial tasks but where should we stop? Where is this proverbial line in sand that separates perceived quality customer service and an automated "call center" or chat-bots?  

So, I am interested in perceived values of current CS vs fully automated service based on AI...

looking forward to your input, observations and suggestions...

Cheers

Rob

[jl_lt]

hi! i can only say "in thy question lies the answer"

[MitchD]

you don't even need AI to replace Australia's tech support (Central Innovation). Just need a script that repeats "we've never heard of this problem before" and "send us the file"

[rakurs]

From my own experience, the forums are an excellent place to seek advice, get advice, to find out how to dance around software limitations, but also to vent.

The wishlist is an excellent place to cope.

The roadmap is barren and does not inspire confidence.

The knowledge base, while "O.K." as is, should probably get many more articles.

I had official support from my local GS dealer (3D art, Croatia) and had largely positive experiences from contacting them. They are not developers, though, and whether the issues I brought up to them ever made their way to the developers and whether they'll be solved remains a mystery.

EDIT:

I have my own reservations about AI and its use in our field. While some aspects of using AI in architecture have appeal, most of them are downright scary, regarding technical consequences of overrelying on something that is almost guaranteed to be half-baked. All LLMs produce an immensely large amount of errors when presented with complicated tasks, and architecture doesn't at all seem like something straight-forward and basic. I mean, I tried the AI visualizer tool, and specified "no water, no oceans, no rivers" in prompts (or variations of those) and there was absolutely nothing I could do to prevent ocean vistas, lakes, ponds and rivers getting integrated into visualizations.

[Rob]

@rakurs thanks for the observations and feedback matey. The error-prone LLMs are worrying me too, especially the reliability of AIs when they interact with customer in paid service scenario. BIM software have become complex beasts, architectural / engineer's practice involves complex tasks, insurance and contracts are legal minefields... letting AI to rule this world of complexity is very risky. Still there is a use for it as I mentioned before, but where to stop? What should we leave to AI in our customer service and what should stay with humans?

Also, I can see a lots of industry activity in training/learn area led by AI, at the same time I am hearing from our customers that they would prefer in-person interaction... what do you think? could we just switch to AI-driven learn and training? Would you onboard your new staff by making them watch AI-compiled content? 

Common guys, keep it coming...

[Scott Boyd Turner]

The practice management software my company uses (Total Synergy) incorporates an AI chatbot that seems to be able to provide clear and concise answers to most queries, but this is also combined with strong online and telephone support.

Support is a major factor for us when looking at systems / software etc, this unfortunately has not been great from Graphisoft /resellers here in Australia, I did not receive any correspondence or telephone calls regarding the change from perpetual to subscription after more than 20years of maintenance. Hence version 28 will likely be the last version of AC I use as the rest of my office and our partner firms have all switched to Revit.

[Rob]

@Scott Boyd Turner Thanks for your feedback. This is very disappointing, providing that Australia used to be a "bastion" of our users' loyalty. I had spent 15 years of my professional life in Australia and when I was leaving a decade ago the situation was very different. Well, I will make sure that your words are heard...

[Eduardo Rolon]

Agree.

IMO this has a very simple answer.

When an existing customer gets in touch, 99.9% of the time it will be because they have a problem.

Based on that the best way to handle the issue will always be to have a human on the other end.

This has nothing to do with "simple" or "hard" problems since the problems you will have are from "new users" which do not know all the ways to get answers (and you want to keep them happy with their choice of AC) or from "expert" users that by definition have exhausted all other avenues to fix the issue meaning that it will be a critical one (or a brain fart problem).

Also "simple" and "hard" issues are definitions from the customer service side not from the customer point of view. For customers all issues are a big deal. Therefore when a user gets in touch the first step is to acknowledge the issue and help them, second is to point them to alternative ways for them to find a solution. Never force them to a specific way to handle a problem since with time they will need to call less.

In both cases AI is not the main solution but something that it is optional, an alternative when the user does not expect a live person to be available.

[Rob]

@Eduardo Rolon 

"Also "simple" and "hard" issues are definitions from the customer service side not from the customer point of view. For customers all issues are a big deal. Therefore when a user gets in touch the first step is to acknowledge the issue and help them, second is to point them to alternative ways for them to find a solution. Never force them to a specific way to handle a problem since with time they will need to call less."

totally agree.. this is a brilliant point, thanks.

[victorbaboi]

Hi! I would split the kind of issue the support department deals with in 3 categories: user (lack of) knowledge, software bugs and business workflow.
To provide a bit more context - I've worked as a BIM manager for many years, dealing with both resellers in my country. For the past year I've started my own company that combines programming and Archicad and bought my subscription from Graphisoft directly for some time. So I've seen this side also - hint, local support won.

For the first one - user knowledge (or rather lack there of) - this may be assisted by AI, however as you pointed out LLMs are prone to hallucination, especially when pushed to answer technical questions. In this area, support people (and I'm referring to internal / local resellers / the forum and Graphisoft Support itself) are invaluable and should only be supplemented by AI as a "first line of defence" / smarter way to search.
Something like an autoresponder generated by AI that gives it it's best shot, before an experienced support person takes a look at the ticket. If it was something simple that the AI was able to fix, all the better, problem solved! If not, the user should know that there's a person at the other end and won't have to fight an unwinnable battle with a bot.
Also for this point (and I'm not understanding the business case Graphisoft is making) - what were previously video courses available on youtube are now locked behind a paywall. Instead of providing all the possible avenues for a user to learn and extend his skills witch in the end translate into a paid subscription and more knowledgeable user base. I understand the need to pay for teacher's time in live courses and whatever certificate you get in the end, but all the materials should be available for study - as part as your continuous subscription if not free for anyone who wants to learn the software. And I see this as relevant for support, because this information would cover probably 90 to 95 of problems user face.

For the software bugs - this point is simple to answer: is the AI is able to understand a problem, fix it in code, recompile, test and deploy an update? Ok. But then we wouldn't pay a software subscription, but an AI subscription that makes software. This is perhaps coming in the future, but not just yet.

And the last category, what I've called business workflow - let's say I have a licensing issue or a misspelled invoice I cannot register with my accountant - would AI be able to modify said invoice? Or provide a temporary license in case of some strange/unknown/whatever happened here to the licensing system? This is where local support from resellers excels, as they understand the local business constraints and have a lot more context than Graphisoft itself. They may not be able to fix bugs, but they are able to help in so many other ways with a simple phone call rather than a 3 weeks long mail chain.

Working with AI daily myself, I see it as a great and tireless help provider (it may help me understand the issue if I'm able to explain clear enough what the issue is). But in the end it's not a replacement for an in-depth help system. So in this context (and sorry if I sound harsh) - I would translate "I am interested in perceived values of current CS vs fully automated service based on AI" as "do you guys need the support team or can we fire them?". Yes, we need them. And the programmers and the testers.

In short: do I see an use for AI inside Archicad? of course! Do I see it as replacing the entire support system? In no way!

And where would I draw the line? My simple answer: is the AI able to resolve the issue faster than a human? Great! No? Then let the human to do it.

[Rob]

@victorbaboi  good points, but customer service is not just technical support (although this service is probably most used). I would be interested in your take on learning/training/onboarding procurements... this is my imagination: you could ask AI to compile a 1-hour training session for your staff (or yourself) on a given topic in a second. Or another one - any new staff could be trained / onboarded based on their initial knowledge level of Archicad by AI-assembled material coming from GS... Now, should it be in person or AI or both? If both where is the line? Should we invest in a self-improving technology that would give you repeated and personalised access to learning/training/onboarding, best deals / sites for custom content, add-ons, GDL developers etc... obviously without human interaction. Would it be ok? Would it dent your relationship with Graphisoft? ...or you wouldn't care as long as it's fast and reliable.

[victorbaboi]

@Rob  indeed, I missed the fact that you were referring to the entire ecosystem, but as you pointed out, most of the interaction with Graphisoft is with tech support. And this is where I see more and more industries switching from call-canter operators that at least have the ability to elevate some issues - to an AI system that gets harder and harder to distinguish from a Turing test.

And the issue I have with this is not related to AI itself, but going to a point where at each step the interaction is driven by pre made scripts. And guess what! those scripts have bugs and corner cases in them just like any software. And if you don't provide for the ability to recover from such issues (by blindly trusting the hype of the solution that will fix all of your problems), the user will, in the end, choose from what he has available - from finding work-arounds to, in the end, stopping altogether from using your service.

So relating to the rest of the ecosystem (outside tech support) - I worked for years with the local reseller - or resellers, as of some time, in Romania. And all the things related to licencing costs, available products and so on went supper smooth, without the need for an AI. It's not like there are hundreds of product versions with a myriad of options to choose from. Last time I looked there were 3 or 4 licencing choices on the GS webstore, mostly related to time. What would AI bring on top of a well designed software description?
The pain came when the reseller told me: I cannot invoice you for this product, as it's only sold by GS. And here I went on to deal with purchasing software (effectively importing it) from another country, and dealing with the legal and accounting hassle. Would an AI have helped in such issues? No. Allowing the local teams to work with the local community and having GS focus on improving the software - yes, it would have. So hopefully this answers your question regarding the seals part. And yes, replacing the whole thing with and AI (no matter how smart) would not just dent the relationship, it would... replace it altogether.

Now, in regards to the training, as I said previously, I see AI as an amazing tool for self leering. But only as part of a broader set of tools - tutorials, wikis, forum, training courses and even manuals, that work along-side the BIM leads and BIM managers. But not as a replacement for any of them. And this is because, in the end, all the AI knows is what it's fed and it mostly takes for granted - and I'd like to avoid situations where the users interprets literally what's been said to him and goes on and applies it, only to later realize that the one small trick used in a special case is not a good practice in general. And this is how I see enough users interacting with such AI systems (aka "your brain on GPS").

A (somewhat) short point about the onboarding process - at least in my role as a bim manager, I spent my time explaining to new users the way in witch ArchiCAD itself works and how the template is designed, always starting from core principles (like layer combinations and views). Only once I made sure this is understood, and the user has the ability not only to use what's given to him but to modify and adapt it to his needs, other tools like internal wikis, manuals and so on come into context. And I only see the AI taking over the "chores" at this point, by providing the users with the exact steps to do a task or where to find a menu. I'm not sure it will be able to replace the core interaction, as when it will be able to, we will all need new hobbies, including the users 😉 .

So in the end I'd like to add two points about this AI planning, that are sure to come up:
- what would a Graphisoft trained AI be able to provide to the end user on top of current models, like chat gpt and perplexity? Given that LLMs are hyper based on text, what would be new besides being trained on the manuals and forums (witch are publicly accessible, regardless of the copyright and moral issues)? Having a window inside Archicad with limited text based context is not really a selling point in what is, mostly, a visual medium.
- do you view such a system as and addon, on top of the current subscriptions? Or as part of them, as a ... you know, a service provided as the subscription? The latter case would be great - something to try and, if proven useful, actually use. But if it's just another paid addon... I would take a hard look at the competition (and I'm referring to Open AI and such, not Revit), as we as users already pay for such AIs.

Hopefully it answers some of your questions.
Regards,
Victor

[Rob]

@victorbaboi  Hi Victor, I appreciate your curiosity but at the moment I am not interested in any productisation or particular customer's end solutions. My curiosity is based on Microsoft research published on LinkedIn... the guy who published it there says (among other things): AI isn’t just replacing jobs. It’s reshaping how work is done - task by task. More importantly, the future of AI & work isn’t binary (safe vs. doomed). It’s modular. Fragmented. Evolving.

Link: https://www.linkedin.com/posts/linasbeliunas_is-ai-coming-for-your-job-microsoft-just-activity-73545...

It is not about firing people and replacing them with AI technology, it is a complex issue . Yes, there are areas where AI would fail (as you suggested in your examples) but my question here is about kind of standard customer service - it is about whether you value human customer service, whether it contributes to your business success by being human, does the human factor contribute to your loyalty to GS or it is the area that you wouldn't care if it was serviced by human or AI. May be I am repeating myself here but I want to stick with the essential points.

[MASz]

Some relationship. I've had some absolutely terrible experiences reporting bugs over the years and I have little confidence in Graphisoft at this stage.

Increased focus on AI is just twisting the knife. 

As others have pointed out, I don't think Archicad has much of a future in Australia. I think it's more to do with Archicad's stagnation and Revit's dominance. When so many commercial projects are on Revit and Archicad can't interface with that at all, then you can see why it's going in this direction. 

[Rob]

@MASz now, let's stick with the topic here and try to be constructive. I understand that the situation in Australia is not rosy, but why would you consider focusing on AI as a wrong move... just to clarify - are you referring to AI in products or customer services?

...and why was your bug-reporting experience terrible?

[MASz]

AI has limited usefulness, and the rest is rather scammy. The hype around the topic is just insane and a number of things are becoming clear about it. Many CEOs have drunk the kool-aid on this and are exposing just how disconnected they are from reality. Cute little club of mediocrity where they are pressured by each other and investors to mention AI every chance they get, just nonstop inanity.

Take for instance Mark Zuckerberg's post about "Personal Superintelligence": https://www.meta.com/superintelligence/
It's complete nonsense and reminds one of the whole failed push towards a Metaverse.

If anything, AI has more been a convenient excuse for companies to fire employees. The recent Atlassian debacle reflects pretty badly on the leadership of the company.

I have seen how Graphisoft's CEO talks and I have nothing positive to say about that. Failure stems from the top.

As it stands, the way that Graphisoft handles bugs has many problems that I have already explained on this forum many times. In short: lack of transparency, lack of action, far too many problems and the way they get handled sometimes is insane. I had a big issue with the 3d window performance and developers were trying to gaslight me on it: https://community.graphisoft.com/t5/Modeling/AC27-3d-window-performance-broken-on-Windows-developers...

Eventually they did actually acknowledge it was a bug and they fixed it, but that was after months of me hounding them. It is already hard enough to break through to developers, and now what, we want to put up more roadblocks in the form of AI? The discussion should be about how to improve customer services rather than replacing it with AI.

If Graphisoft is serious about improving this, a big part is improving the bug handling process. Implement the bug tracker, have better processes around handling bugs, actually fix them, don't leave them for multiple versions, or better yet - discover and avoid bugs in the first instance without the need for testers. This would could a huge amount of my customer service needs. The difference is night and day with some other developers. I send them a bug, they investigate it on their end, acknowledge that it is a bug and sometimes deliver a fix extremely quickly, thanking me in the process for reporting it. This creates the kind of relationship where I am happy to support the developer.

As for AI products in Archicad, I see no value in them. There are some tasks that could be automated, but that's not there yet and it is probably can be done better through new features instead.

I have tried to be constructive for many years about Archicad, but it has been a futile exercise. I am just repeating myself at this point. Many users have tried to be constructive and have submitted tonnes of feedback that has not been acted on. I don't even see the point of this thread. Graphisoft is on a clear path of enshittification and fleecing its customers through increased subscriptions. 

[Rob]

@MASz 

"I have tried to be constructive for many years about Archicad, but it has been a futile exercise. I am just repeating myself at this point. Many users have tried to be constructive and have submitted tonnes of feedback that has not been acted on. I don't even see the point of this thread. Graphisoft is on a clear path of enshittification and fleecing its customers through increased subscriptions."

Although it is not in my remit to comment on company's licensing policy, I appreciate that there have been tonnes of feedback, but what I am trying to do here is to make it more structured with a focus. Simply some of the feedback is affecting several departments and my role is to establish Voice of Customer program that could channel this feedback more effectively with the clear strategic priorities. At the same time every customer's feedback needs to be in a closed loop of action transparency for you... obviously there will always be a delay that is given, providing the number of possible involved decision-makers and projects that are already in making at our end. What I can promise though is that the feedback will find the appropriate ears and eyes. Actually that is the essence of VoC. BTW I am really amused by peculiarity of "enshittification"  word, it should definitely make it to Oxford Dictionary. 🙂

edit: checked the dictionary and it is an actual word!! well done.

[Eduardo Rolon]

Don't forget the "Trust Thermocline".

https://every.to/p/breaching-the-trust-thermocline-is-the-biggest-hidden-risk-in-business

Doctorow has written more on the issue and the Wikipedia entry has a good summary

https://en.wikipedia.org/wiki/Enshittification

[jl_lt]

Great article. As valid now as when you first mentioned it, when there was still a chance of not crossing it

[Rob]

@jl_lt 

you know that I don't like one-liners full of wisdom here.... pls contemplate in more detail. Every opinion counts as long as it's clearly communicated.

[jl_lt]

i would have to say that you are already trying to steer the conversation to the direction you want to hear. AI does that too, that is why one has to try to eliminate bias from the very questions oneself is asking.

but here it goes, in the form of nuggets of wisdom:

-id rather talk to a useful AI than a useless human,

-if the human is actually going to (try to) solve the problem, you better get to talk to him/her as soon as possible. For examples of how to do it correctly, check out how they do it at www.domain.com helpcenter;  non intrusive AI bots coupled with expert human help.

-you already have that expert human help in these very forums. threat them with lots of love.

-all your written help content, tutorials and learning materials should be completely free to all.         

-more education = less deviation 

-for guided turorials, certifications, courses and personal tutorials you can charge as much as you want. 

-Let the AI help with the learnimg process.

-All the AI in the world wont be able help with a faulty, out-of-date software. 

-as much as we'd like to deny it, and the AI assurances for the contrary, architecture as a profession is on the verge of an existential thread.