SAP Inside Track Berlin 2019
On the 31/08/2019 was SAP Inside Track in Berlin. As I write this I am on the train going back from Berlin to Heidelberg. A trip of six hours which I do not mind at all as I am sitting in the bar.
Berlin is a wonderful city. Last time I was there it was snowing, this time it was over 35 degrees centigrade but that doesn’t really matter. It is just a pleasure to walk around popping into various cafes and bars.
I had asked the organisers what hotel to stay in and they gave me a short list of three hotels. I choose the “weinmaster” because I liked the name. It turned out to e a very “arty” place which is no bad thing and also it had an adult’s only policy which I found interesting when I read about it, but sadly there was nothing rude/naughty about the place.
Berlin Hotel Wall
I travel on business a lot and one thing I notice is that the rule for hotels is that no two hotels can have shower controls that work in even a vaguely similar way. There seems to be a worldwide contest as to who can produce controls which are the most difficult to work out how to use them. This is a very competitive field but the Weinmaster wins hands down. It is even worse for me as I am pretty much blind without my glasses, I asked the receptionist about his and he said pretty much every single guest had brought the subject up. His quote was “someone put a huge amount of effort into designing this system … which doesn’t mean it is good”. Later you will see how this ties in with a talk given at SIT Berlin.
Maybe now is the time to actually start talking about the even itself. First up was the pre-event beers on the Friday night at a beer hall near to the event venue. That venue was the “SAP Data Space” which is sort of like a mini conference centre. It is co-located with the “SAP Data Kitchen” which sells food to nearby non-SAP companies. You do not generally think of SAP as being in the food provider business, though they do have coffee shops as I understand it e.g. one in Paolo Alto in California.
There were eight sessions, so here we go:-
HANDCUFFED & BLINDFOLDED – Soumalya Nash & Bart Van De Kamp (Accenture)
This being Berlin when you see the stage full of handcuffs and blindfolds it is all too easy to make certain assumptions like I did about my “adults only” hotel Once again there was nothing rude/naughty going on, the only reason the presenters wanted to blindfold and handcuff volunteers from the audience was – I hope – to stop them interacting with a computer system using a keyboard and instead use the “conversational AI” product from SAP and just tell the system what they wanted it to do using their voices.
In this case one person had to tell the system – via Alexa – to create a purchase order and someone else had to approve it. It was probably a bad example to pick as in my experience people how have to do manual repetitive tasks like creating purchase orders tend to master the process after a few weeks no matter how complex the SAP screens are. In this case the process took far, far longer than manual input would have done mainly because the “chatbot” kept talking and talking and refused to be interrupted until the volunteer screamed at it “SHUT UP!” to which it replied “we all have our limits” which I thought was quite a witty response from a machine.
Maybe there is something wrong with me but I don’t like talking to robots, which I have to do more and more as the years go by. The more realistic they are the less I like it. When I set up my trial HANA account on the CAL I had to set up an AWS account. The SAP side was all manual and involved about fifty thousand steps and took forever. The AWS side involved talking to a machine that sounded and responded almost exactly like a human and that process only took a few minutes but is disturbed me greatly. I knew I was not talking to a human no matter how much it sounded like one.
I don’t much like animals acting like humans either. With monkeys and aps it is just about OK, but I saw a certain type of dog in Ukraine which would walk around on its hind legs every so often and that gave me the creeps. It’s the same when a Parrot or similar bird says “Hello!” or sometimes even a full sentence. Just like a machine it has no idea of the meaning of what it is saying but in both cases it sounds like it does.
However, just in case you do not have some sort of horrible phobia about talking to robots like I obviously do, SAP are making great strides in this area.
There is a website with the URL cai.tools.sap where you can build a prototype of this sort of technology for free. The whole process did not seem complicated at all. You have to define some “triggers” – what is the intent of the user e.g. do they want to create a purchase order (or a monster) and once the robot has worked out what the human wants you need to define a corresponding “action” which could be a URL to an OData service in SAP to create a purchase order (or monster).
One thing I did find funny was on the website was an apology about the name of the product changing. SAP renaming products? Surely Not! I could be totally wrong about this but I seemed to get the message from this presentation that SAP no longer calls their virtual assistant “co-pilot”. I am happy to be corrected on this.
DATABASES GO MULTIMODAL – Vitaly Rudnytskiy (SAP)
The first thing the presenter did was to give out of a pair of “developer socks” to an audience member as a prize. This sort of thing is very important.
Another fascinating fact is that the font in Fiori 2.0 is called “Font 72” on the grounds that SAP was formed in 1972.
Moving onto the main thrust of the talk people like me are used to Oracle databases which store data in a certain way (RDMS) with tables with foreign key relationships are so on and so forth. As it transpires other types of databases store data in a very different way and “multi-modal” databases can store the same data in lots of different ways all at once. I could be talking nonsense here but that is the impression I got.
As you no doubt know SAP has been buying a large number of various companies in the last few years, all cloud type organisations. When you buy a company you also get any technology they have developed and patented, whether you want it or not. As an example SAP got the JAM collaboration platform when they bought SuccessFactors which was of course not the reason they bought the company it was a sort of added bonus (or added liability depending on what you think of JAM).
In the same way when SAP bought “Callidus Cloud” as a side effect they got the rights to a database called “Orient DB” which is a multi-modal type of affair. This was demonstrated using the ever popular “Open Beer Model” (which keeps cropping up at SIT events) where you can see the same data in either a tabular form or a graphical form. What do I mean by a graphical form? That is where something like a beer or a brewery is represented by a ball with arrows linking it to associated data e.g. beer X is brewed by brewery Y in the city of Z.
In the same way the HANA database has multi-modal features. In this example a data set of Marvel superheroes was used. There are about six and a half thousand of them as it turns out.
To end off with Vitally gave out another present –this time a fluff (Teddy Bear) to one of the SIT Berlin organisers Oliver as he has a new son. The gentlemen in question actually turned up at the after event beers but he could not drink too many beers as he was only a few weeks old.
BLOCKCHAIN FOR FRAUD PREVENTION – Karina Krupina & Nena Becker (SAP)
This was the first of two sessions all about blockchain. In this one the story was all about a fictional person called Peter Purchaser who every time he has to buy something pretends he is putting the deal out to tender but in actual fact always arranges things so his mate Stefan Supplier gets the job. This is, of course, fraud.
The idea from SAP – and this is just a proof of concept, not an actual product as yet – is that the various suppliers contacted submit their bids and the details al get stored in a “blockchain”. Peter purchase cannot see the actual values until the deadline passes so he cannot tell Stefan what the other bids are, and after the deadline since the blockchain is immutable Stefan cannot sneakily lower his bid after the event.
Now it could be argued there are about ten billion zillion other ways to achieve the same goal, but I don’t suppose that is the point. This is just a possible use case for blockchain, which is often viewed as a solution looking for a problem.
WHAT IS BLOCKCHAIN AND WHAT IS A CONSENSUS MECHANISM? – Andrea Pham (SAP)
This maybe should have gone first as one person in the audience said he only understood the first blockchain presentation after listening to the second.
If someone thinks they have a use case for blockchain the first problem they have is that there are a very large number pf blockchain suppliers so it is very difficult to work out which one to pick. The analogy made was that of email providers – in the past there were a very large number of them as well and now this has dwindled down to just a few e.g. Google and Yahoo. So presumably the same thing will happen with the blockchain providers.
The definition of blockchain is “mathematical proof that something happened” as opposed to a person or organisation claiming that event happened.
One thing that is very clear, and was the main focus of the presentation is that blockchain does not seem to be all that secure, there are dozens of possible attack methods. My favourite one is the “Sybil Attack” because that reminds me of “Fawlty Towers”. In any event a large number of people have had a very large amount of money stolen from them via such attacks.
A “consensus mechanism” is a defence against these sort of attacks, and it seems that most of them do not work very well.
Now I could have got totally the wrong end of the stick here, but it seems to me that maybe blockchain.is possibly not the silver bullet which will solve all the worlds’ problems.
Typical Berlin Street
DESIGN THINKING IN “ACTIVATE” METHODOLOGY – ANNE JOHNSON & JULIANA HENDEL
No doubt you have heard about “design thinking” a million times before and know that “activate” is the successor to the ASAP methodology when it comes to implementing SAP projects – for example there are some radical new concepts like using “sprints” to delivery missing functions in short bursts as as opposed to loads of functions at once in a “waterfall” phase.
This was an interactive presentation where the audience had to split into teams of three and come up with crazy ideas to improve a beer themed web site which had most of its functionality filled by SAP standard but still had some gaps which needed to be addressed.
The presentation ended with some psychology of why this sort of design thinking worked e.g. when someone “owns” an idea they are more attached to it then if it is assigned to them. In my experience when someone gets the idea they “own” something e.g. a certain program then they guard it to the death and will not let anyone else come near it.
Neither presenter works for SAP, and the next observation is mine alone – if there is some sort of methodology to address gaps in the SAP standard then the positive thing is that hopefully some people at SAP understand there are gaps rather than the “everything vanilla / if you have a special requirement it is bound to be covered by standard SAP functionality either now or one day far off in the future” message I hear at major SAP events.
BUILDING THE RIGHT IT BEFORE YOU BUILD IT RIGHT – TUDOR RISUTIA
As you may possibly know not every IT type project goes successfully and when the audinec was asked for a show of hands as to who had been on a disastrous project many hands went up, including mine.
This is not just related to SAP – though they have come up with some products in their time which were pretty much dead on arrival e.g. SEM, Web Dynpro Java, XApps, ESOA and so on – for example 80% of mobile apps do not make money but that does not stop ten billion people creating new ones every day.
This ties back to the shower system at my hotel. Someone spent ages building the system and it cost the hotel a fortune and it is unusable by a human.
Tudor explained the concept of PRETOTYPING as opposed to PROTOTYPING. This is the concept of seeming to build a prototype without actually building one – the rhyme is “faking it before making it”.
He gave some famous examples. The most pertinent is that of IBM who were considering building a speech to text machine many years ago when most people hated typing. Instead of a real machine they had someone behind the wall listening to what the testers said and typing it in so it appeared on the screen as they said it.
That was a lot easier to build than any sort of speech to text software. Ironically as it turned out the focus group initially liked the idea but then decided it was a load of old rubbish as it would make the office really noisy, give you a sore throat etc..
The reason I think this is pertinent is that IBM abandoned that speech to text thing as a bad idea 20 years ago, and yet today the very first presentation was all about conversational AI which was a real prototype speech to text thing (albeit with added AI) which the volunteers from the audience also quickly learned to hate. So in some ways there really is nothing new under the sun.
The “Mechanical Turk” was another example from the 19th century. Its inventor claimed it was a machine that could play chess but in real life there was a chess expert dwarf inside with big feet hiding inside the machine moving around the chess pieces. Nonetheless it fooled everyone at the time.
Fast forward to the present day and Amazon offer a service called “The Mechanical Turk” which they describe as “artificial artificial intelligence”. Here the fake prototype involves calling an AI/ML service which is actually one or more people pretending to be computers that can do AI/ML.I love the idea, and what a wonderful thing to be able to say when someone asks you at a party what you do for a living i.e. impersonate a computer.
A non-SAP thing that was shown was “teleport” where instead of drawing the screens on a computer screen like you do in BUILD you draw them on an actual physical whiteboard with a marker pen just like you do in meetings currently. Then the application scans the whiteboard and generates code from what you have drawn.
NOTHING ABOUT DEVILS – SOREN SCHLEGEL
As it said in the title this session was not about devils – it was about ABAP daemons instead and how they relate to the internet of things. I was happy to hear this as I have often struggled to find a real life use case for the things.
In case you do not know yet a Daemon is a background process, and an ABAP Daemon is no different – it is like a batch job that runs forever until programmatically stopped, one that comes back to life it it “dies” via a short dump or an application server shutdown or some such. If one application server goes down they can jump out and into another one and keep going.
The presentation started by noting there were ten billion zillion different IOT providers, all claiming they are better than all the others. As with the blockchain companies presumably eventually this this will get whittled won over time by business failures and mergers and acquisitions.
MQTT is a communication protocol designed for remote locations, and it has a very small code footprint which sits inside your “thing” – which in the demonstration was a chip about half an inch long.
MQTT requires a broker and that is where the ABAP daemon comes in. You create a Z class based on CL_ABAP_DAMEON_EXT_BASE and use that to set the Daemon going programmatically (or indeed switch it off) and to publish and subscribe to events.
You can monitor which ones are active with TCODE STMDAEMON.
So far so good. I never got around to asking him (though I had plenty of time to do just that, I was too busy drinking) how that compares with the ABAP Channels concept which also uses a publish and subscribe mechanism. The ABAP channels are bit more form based and the Daemons seem to be all code but they seem to do the same sort of thing i.e. react to events (messages) and forward them on to subscribed applications.
APACK – Sebastian Wolf (SAP)
I wrote about this in my blog about SIT Oslo so there is no need to repeat myself – just follow the link below
I will say this is an open source project and I subscribe to updates which happen almost every day, so this is a fast evolving product.
TWO BEER SESSIONS
The first one was a BBQ right in the garden of the venue, with great food, loads of beer and wine, and an ice cream to finish off – and all for free! Just to be clear there was no charge for the event, food and coke and water all day, breakfast and lunch and a BBQ at the end, and there was also some SAP stuff to listen to during the day, so what is not to like about this?
Eventually though the SAP Data Kitchen staff wanted to go home for some reason as opposed to just keep giving us more bottles of wine until the wine cellar was emptied which was, admittedly, our plan. They eventually managed to get rid of us and we migrated down the road to the “Lemke” beer hall where we had the pre-event drinks the night before.
Berlin Lemke Beers
Then I had a beer at the hotel bar when I got back, however that is probably not vitally important information for you to have.
It was a wonderful event. Sadly I have to return to Australia in a few weeks, so will not be here (Germany) for the next one but if I was I would be there like a shot. If you get the chance I would encourage you to attend the next one.
Thanks for the in depth summary of our little event, Paul. You covered the gist of it pretty well with beer, handcuffs, food, and more beer, and some technical sessions on top. One think that wasn't for adults that comes to mind was the SAP teddy bear from Vitalij. Glad you liked it, and thanks for the recommendation. Hope to have you back sometime soon.
Thanks a lot for the feedback.
Though I’m a bit sorry to hear that you didn’t like talking to robots. ? But our main idea was to highlight this new technology offering from SAP in the domain of the modern user interface. I hope we could share some information about it.?
Just one more thing, SAP is not actually renaming their old product CoPilot to Conversational AI, they recently acquired the platform Recast.AI and then renamed it to SAP Conversational AI, and then they are now migrating the CoPliot into Conversational AI. Thus Merging two products into one.
We hope to meet you again in another SAP event. We liked the SIT Berlin a lot, and would surely recommend all SAP developers to try to attend it if feasible next year. ?
Just one further question - SAP's Firori strategy was to have a "co-pilot" section in the standard layout of all future applications. After the merging of the two products is complete I presume the name "Co-Pilot" will be no more and instead there will be a "conversational AI" section in the standard Fiori layout?
The digital assistant that will available as per the standard Fiori 3.0 layout, will still be called CoPilot digital assistant, but this component will run on the SAP Conversational AI platform, in the back.
The following high level diagram represents the basic working principal.
Thanks for joining and your very valuable feedback.
In addition to Soumalya's input also to highlight that in future the SAP Co-Pilot will be named as Co-Pilot. The reason behind this is that’s not a separate product anymore, which also means no separate license. SAP will provide standard Co-Pilot skills, in case of extension then the SAP CAI Platform is required (thus also the license).
Initially the focus will be on the digital assistant functionalities, but soon also the person-to-person functionalities.
thank you for your blockchain commentary and it seems to me that it's a little bit more than just a solution looking for problem, esp after reading this: https://data-economy.com/sap-invests-e200m-in-new-ai-and-blockchain-technologies-hq/ which i found here: https://twitter.com/dahowlett/status/1178062320263684097, so is it too early to register for the next year SIT yet?