Technology Blogs by Members
Explore a vibrant mix of technical expertise, industry insights, and tech buzz in member blogs covering SAP products, technology, and events. Get in the mix!
cancel
Showing results for 
Search instead for 
Did you mean: 
STaur
Explorer
729

Note : You can skip to IBM TWC Skills Section if you want a short and quick understanding of the skills developed by IBM with the The Weather Company data.


Evolving Customer Experience and SAP CoPilot


Over the past couple of years, we have been swarmed by a number of new kinds of devices and software collectively known as digital assistants. These Digital Assistants come in various forms and shapes either embedded within our Smart Phones or as a separate device sitting in our home or as Bots as they became mainstream channels for customer engagement.


These are automated hardware or software machines that are powered by the advances in Artificial Intelligence (AI) technologies. Recent developments in machine learning algorithms, such as deep learning and deep reinforcement learning, have improved the performance of AI tasks such as Automatic Speech Recognition (ASR/Speech To Text), Natural Language Understanding (NLU/NLP), Text to Speech Synthesis (TTS).


These recent advances in AI technologies has resulted in new ways of interacting with systems and machines, using natural language and voice commands leading to interactive experiences. Its not distant future where AI Robots would look and talk like humans understanding the nuances of sarcasm, irony or pun and responding in similar fashion in natural language providing unparalleled consumer experience, in consumer interactions.


Anyone who is a fan of SCI -FI flicks would not have missed the Passengers(2016) starring Jenifer Lawrence and Chris Pratt. One of the scenes resembles quite closely how the future would look like, when Chris Pratt almost assumes that he is talking to a human when he first speaks to droid bartender named Arthur who is the only company he has on the spaceship. We still have a long way to go toward building digital beings that are capable of seamless natural language conversation, like what Arthur does, based on plethora of knowledge he has garnered from all possible sources available at the time.


However, a recent surge in interests and massive investments and pursuing the Conversational AI based interactions industry giants like Apple, Google, IBM and Amazon have invested heavily in developing the NLP engines and tool kits that support, chatbots and conversation interfaces. IBM Project Debater pushes the frontiers of AI to facilitate intelligent debate so we can build well-informed arguments and make better decisions. Apply that to our business meetings, global debates at UN or as simple as having a simple conversation and we have super intelligent assistant questioning the ways of our thinking and actions we take.


https://www.research.ibm.com/artificial-intelligence/project-debater/


These intelligent digital beings have become in some industries like retail, banking, health, public services a main customer contact channel and offers personal, empathic customer experience.


SAP with its 350,000 plus customer base and its new SAP FIORI UX paradigm(and soon to come Fiori 3.0) which has been new face of UX for SAP for a while now, was off course looking for the next shift in the ways of interactions with its digital core. With some of the third-party Bots already making its way to the SAP world and proving to valuable for customer touch points, it was a high time that SAP got their conversational UX put into action with introduction of SAP CoPilot.  First introduced as part of the 1705, was merely another chatbot which was contextually aware, could do object discovery, and could be used across the enterprise as a chat engine with facilities to share business objects and files without living the SAP Fiori environment.


With subsequent releases SAP started hinting towards conversational AI and making SAP CoPilot smarter using a TensorFlow based cloud service for natural language interactions(NLI). The acquisition of Recast.AI was another move to further its plans to expand into several external channels like Slack, Microsoft Teams, messenger and by offering a single platform for integration of Bots on SAP Solutions with external channels and third party bots, through the SAP CoPilot BOT integration platform.



SAP CoPilot BOT Integration Hub


Journey to SAPPHIRE 2018 with the Beta Service :

The Beta version of the SAP CoPilot Cloud Service for NLI was made available to some of the key partners across the globe early Feb 2018, IBM was also invited and was involved from the beta stages. The journey since then has been enthralling to say the least, with several challenges right from getting the setup completed to the point where IBM has its own set of SAP CoPilot Skills which has surpassed the expectation of the SAP Product teams on what could be achieved with limited set of features and tools available during the beta phases.

During the initial stages of our journey, I had to put together a storyboard and mock ups for a use case which would prove beneficial if we planned to deploy it on any customer project. These were around Enterprise Asset Management(EAM) which I had vast experience from past assignments and was thinking how the Intelligent Conversations with SAP CoPIlot could be applied to any of the roles I am conversant of.

At the same time The Weather Company(TWC) an IBM Business was looking for channels to expand into the SAP world. It was a Eureka moment as in an instance it dawned upon me that there were synergies if we combine these two (EAM  and TWC data) with SAP CoPilot. Soon I had put together an end to end conversational flow with a Planner / Scheduler role in mind to help this role tackle some of the day to day challenges around scheduling with weather insights. Since this was not going to be just another UX design, had to think through the natural language interaction the user would have as a conversation going through the day to day activities the role performs.

Also, it was crucial to have the design which the UI developers could relate to, while developing the SMART Widgets® keeping the manual interactions to minimum. Also had to consider feasibility in the then limited Beta service features and tools. As expected, I was met with great deal of resistance when I first proposed the entire scenario consisting of 5 Core skills interacting with the SAP S/4 HANA system. I was shot point blank in face with all sorts of technical challenges by our team and SAP teams clearly indicating that what I was proposing was not possible with the Skill Builder and the features available in the beta version and wouldn’t be the case until a few future releases.

But I persisted, my thinking was along the conversational experience the users would lack, if we didn’t design the skills as a natural conversation, the interaction would be too “robotic” with query and response and not naturally flowing . We pushed on to see what could be achieved without changing the design and soon we hit a first block as we couldn’t deliver any Quick Views with Icons and there were limited options with the Quick Views. So we decided to build the custom ones which was fraught with surprises and challenges as it worked at times on desktop and not on the mobile and vice versa, we call them Smart Widgets®.

Also to make the whole challenge interesting, we had submitted the idea of Conversational UX with SAP CoPilot for SAPPHIRE NOW 2018 which was selected hands down as is a key topic of interest for SAP Customers to be presented at IBM booth. Lack of documentation(Beta Version) and experience on building skills or conversations with tools available was a steep learning curve for the whole team. We were racing against time and with some SAP technical help, and unwavering persistence by our teams we had a our first break through, on the day prior to the eve of SAPPHIRE NOW 2018 !

To our surprise the SAP CoPilot Service also got a voice to text button enabled for Voice Commands the very eve, and the SAP CoPilot iOS app could launch our first set of skills on iphone. Our anticipation on some announcements of SAP CoPilot in SAPPHIRE NOW 2018 were answered but with a bit of anti-climax that the voice commands didn’t work well in the SAPPHIRE settings, which raised concerns for what we were planning to demo using voice commands and we kept our fingers crossed. Over the next couple of days we polished the skills and the entire EAM use case scenario was demoed using voice commands at the IBM booth live !

The word spread like wild fire and soon we were at the SAP booth with SAP CoPilot product teams being amazed at what IBM had achieved in such a short time with limited beta features and what was thought to be not possible until a few more releases in future.

The end product was so impressive that SAP CoPilot Product Management invited IBM to co-present at some of their sessions in SAPPHIRE 2018 for clients to understand how value is delivered by using SAP CoPilot combined by automating processes with voice.

IBM TWC Skills for SAP CoPilot.

After SAPPHIRE NOW we, went back to drawing board, this time with all the technical issues sorted we wanted to take the whole conversational experience to next level by adding ML, more RPA's and Weather Data based few more skills and make it into a deployable package which could be used out of the box for clients who are interested in trying these out on their landscape.

The following IBM TWC skills for SAP CoPilot are available as a deployable package for SAP S/4 HANA landscape. We are working with SAP to get these skills ready for the skill store on the apps store and get the deployment automated. These are designed around the role of Planner Scheduler or Customer Service Agent who deal with planning the field work force activities on day to day basis in an EAM environment. These skills use combination of RPA + Machine Learning(ML) + Weather data and base conversational AI constructs available within the CoPilot Framework.


IBM TWC Skills for SAP CoPilot for EAM (Role : Planner / Scheduler)


These core skills can be used in a sequence as explained below, and the ancillary ones will provide some key utilities that augment the core ones.

Core Skills:

Combination of the following core skills is projected to save at least 20-30 mins per job for the planner, if the same tasks were to be carried out manually using SAP transactions like weather data look ups and the decision points that the user has to go through to get the jobs scheduled/ rescheduled based on weather changing dates and times, ensuring technician availability and the tools assignments for getting the job completed in first go. Eliminates the need for the user to be trained and experience in these tasks, as simple voice commands can get the same tasks completed within minutes. Following are brief descriptions of what these skills deliver along with what they use under the hood.

  • Agent Assist: Uses RPA + SAP S/4 HANA


Prioritise your jobs based on SAP CoPilot suggestions. Instead of sifting through the heaps of pending jobs, let copilot assist you to decide the most critical and pending jobs that require your attention, based on the completion dates, criticality or longest aging in the queue.

  • Reschedule with Weather: Uses RPA + Weather Data + SAP S/4 HANA process data


Reschedule your critical jobs that are clashing with inclement weather and increase your first-time right rate for Job completion. Use TWC and the  SAP S/4 HANA back end to get the right priority jobs scheduled in the most suitable weather conditions. Be it the windflow, humidity, precipitation levels, snowfall or temperature requirements. There are several business scenarios where these parameters are crucial and the skills could be used.

  • Engineer Assignment: Uses RPA + SAP S/4 Config


With these set of commands CoPilot will assign available engineers with the right skills sets, priority of the jobs and Start and Finish dates on the critical jobs.

  • Tools Assignment: Uses RPA + Watson Machine Learning + SAP S/4 Config


The skills helps locate jobs in the past on the equipment / FL, and finds tools missing from the standard PRT that will help complete the job right first time. Definitely suggested if much of the knowledge on work orders resides with the field engineer and never gets updated back to your System of Records, a common problem with EAM in all industries. Suggests tools based on frequency and usage within operations that were needed to be performed on similar or same equipment’s in the past.

  • Reschedule with Weather for Special Assets Uses Watson Machine Learning + RPA + Weather Data + SAP S/4 HANA Data.


Any special assets that require key weather insights for getting work done. E.g. work done on assets next to High Tension poles, work done in a low-lying area easily flooded with slightest of rains. The assets around these high-risk areas need special handling by drawing a geofence for acceptable thresholds of weather elements within a perimeter to ensure Environmental Health and Safety of the technician(refer the Geofencing based ML section). This skill uses Watson Machine Learning to learn from the acceptable thresholds specified for an equipment type, weather conditions, location and any other special conditions to schedule jobs at appropriate times.

  • Scheduling Bot: Uses RPA + Watson ML + Weather Data + SAP S/4 Hana data


This skill is a combination of core skills to schedule, assign engineers and tools and release orders for the field work force to action. With a single command the scheduler can get a day’s worth job completed in couple of minutes. The Scheduling Bot helps the scheduler target all the High and Very High Priority jobs to schedule / reschedule based on weather conditions.

Supporting / Ancillary skills:

  • Meet and Greet: Uses RPA + S4 HANA data


Enable your business users with latest and most accurate weather data 24hrs a day with simple voice commands on SAP CoPilot. The user also gets statuses on connected systems, API's to the SAP S/4 HANA, and gives a simple answer to confirm that they are up and running or have some problems. Under the hood calls the The Weather Company weather data to give the most accurate predictions on weather across the globe and interfaces with the connected systems. This gives user a first hand information before the planner starts work.

  • Text Translator: SAP Leonardo Machine Learning


Want to interpret a comment entered in different language in your work orders, job completion forms or invoices.  Or want to talk in foreign language with any of your colleagues from other countries and make them feel home, just ask copilot with simple voice command get translations in native character set as you speak. Or use this skill to translate text from any language to English. It also offers a nice widget to detect any source language to user chosen language with native character set.

  • Scheduling performance / KPI Skills: uses RPA + SAP HANA CDS Views


Get real-time graphical view of your scheduling performance across a plant for your scheduling teams within a quarter. Also gives graphical view of the number of reschedules that have taken place within a quarter giving a gauge of first time right schedules based on weather. Uses SAP HANA CDS views for the latest live information in neat interactive graphs.

Geofencing based decision making with Machine Learning: Watson Machine Learning uses the Asset Type, Location and The Weather Company Weather data to determine the safe perimeter for the assets around this High Risk Asset to determine the acceptable thresholds of weather conditions, like snow fall, precipitations, wind flow, temperature and so on. Anything within the perimeter will have different thresholds for e.g. the overhead High-Tension wires / poles or transformers would never have maintenance work being carried out if the precipitation is above 20 % but the same can be acceptable for assets outside the geofenced perimeter. Same goes for a low-lying area where underground pipes, pumps are prone to water clogging due to floods in heavy rains. So, the area will not serviceable when the precipitation reached 40 % or above. Once the scheduled order is technically complete Watson ML API is initiated to feed the data into the ML to retrain it and adjust accuracy and threshold levels for future predictions / decisions.

Watson Machine Learning applied for Critical / High Risk asset maintenance


Back to Future .....

The latest announcement by Bjorn Goerke in the SAP TechEd Barcelona key note on using Intelligent Robotic Process Automation Tool along with SAP CoPilot and hints of ML being used are reaffirming that IBM had the right strategy with the new technology at hand and are rapidly moving in the right direction by innovating new ways interaction with your digital core for any industry clients to truly achieve an intelligent enterprise.

The value of SAP CoPilot with Conversational AI is not using the conversations to get standard responses for queries, but augmenting the decision making for key roles within the organisation by providing intelligent insights combining its context awareness with RPA, ML and External data from The Weather Company to create competitive advantage. It drives employee engagement, productivity, efficiencies in completing the daily tasks and freeing up the hours spent in doing the same approvals, orders or requisitions to actually improving and optimising the business process, bringing in real business transformation and there by a competitive advantage.

SAP positions the SAP CoPilot offering as Intelligent Enterprise’s Digital assistant for employees and the SAP Conversational AI(Recast.AI) for consumer facing channels. Pushing the envelope further not only one will be able to interact with SAP S/4 HANA systems with voice commands through SAP CoPilot but also it will respond back with its own voice. What if I say this would be possible in your language of choice with German, French, Korean already on the roadmap for the Natural Language Interactions (NLI).

With 1811 release for SAP CoPilot will have its own voice and personality available for Enterprise channels. Whether this functionality will be extended to the consumer facing channels is what SAP is still working on and only time will tell. From my perspective it makes sense, to make SAP CoPilot a wholistic offering to cover all facets of Consumer or Employee interaction with enterprises digital core on multiple channels. Security and seamless integration on these consumer channels is of key importance and SAP is making every attempt to secure the channels first before offering the parity on innovations on all channels.

SAP CoPilot for Enterprise Channels and Conversational AI for Consumer Channels.


IBM have co-presented with SAP in the CGE203: SAP CoPilot Overview and other sessions on the EAM Package for IBM TWC Skills, several times between both SAP TechEd Las Vegas and SAP TechEd Barcelona. SAP TechEd Las Vegas has this topic covered in good amount of details with demos on what is possible today and the future outlook along with short demo of IBM TWC Skills included in there. IBM was the only partner who had the Conversational UX paradigm mastered and has key capabilities to apply emerging technologies to business challenges create differentiating value for their clients in evolving markets.

https://events.sap.com/teched/en/session/41239

SAP CoPilot has just started the journey on the conversational interactions, there is more in wish list of features to make it a true competitor in the Conversational AI + Digital Assistants space not only for the B2B, B2E but also B2C scenarios. SAP alone can’t create all the skills and would require partners like IBM who understand the design thinking required in this field of AI and has deep rooted experience in designing conversational experiences to fill the void once SAP customers understand the potential value of combining the RPA + ML + Conversational AI and that is not too far from now. The future is exciting and at the same time intriguing of what you could make of conversational AI / UX / UI for SAP and non SAP applications with SAP CoPilot.

Let me know what do you think about the SAP CoPilot and Conversational AI features, what could be improved or changed and what would differentiate it from the Googles and Alexas of the world.
1 Comment
Labels in this area