Making smart decisions with RPA and Machine Learning

Machine learning, a subset of artificial intelligence adds intelligence to RPA bots. Machine learning models capture human knowledge as model parameters and then enable the RPA bots to take decisions like humans. It elevates RPA to “intelligent” RPA meaning it can address a larger variety of use cases. The multiple scenarios such as support from experts, handling of unstructured data, reviewing of responses, etc.

As an example, we added a machine learning model to enhance bot capabilities in the service provisioning and order entry space for a tier 1 service provider in the US. The existing process could not be automated entirely with RPA bots. Traditionally, an RPA bot would be paused till an expert is present to provide a decision where human judgement is needed. We created a machine learning model that could learn and eventually remove the need for human judgment-based decision making.

With help from the process expert team and machine learning experts, we collected decision metrics for the previous few months and analyzed them to build a text-based intelligent-decision-engine. As the response data was unstructured, we applied NLP and natural language understanding (NLU) techniques to data analysis and model building. We trained the prediction model with earlier responses dating a back few months, and then connected to the RPA bot. Now, when the RPA bot retrieves response data from the commissioning server, it calls the machine learning model via an API and proceeds in line with the model’s output. We removed the dependency on human judgment-driven decision-making, gaining an initial decision accuracy greater than 93%, something that will constantly improve with retraining.




AI is the new electricity – Andrew Ng’s latest writing on machine learning


Dear friends,

Attached are the first fourteen chapters of my book, Machine Learning Yearning. In these chapters, you will learn how to align on ML strategies in a team setting, as well as how to set up development (dev) sets and test sets. Recommendations for how to set up dev/test sets have been changing as Machine Learning is moving toward bigger datasets, and this explains how you should do it for modern ML projects.  

I hope you find this useful, and will send more chapters next week!



Andrew Ng Machine Learning 2018

Four Phases of Digital Transformation


wf model.JPG

As Intelligent Automation begins to mature, larger
patterns around implementation are starting to
emerge. We’re learning that in order for a digital
transformation to be successful, there’s a need
for certain program management and technical
competencies. These competencies include use case
selection, data privacy controls reducing the total cost
of ownership, and more, as the IA experience widens.
We benchmarked four stages of maturity that are
relevant for most organizations. We have identifed
the programmatic competencies required to support
each stage, and beyond, to help you move seamlessly
through the continuum.
But, due to the nascent nature of Intelligent
Automation, its maturity is not linear. As you examine
the components of each stage, you may fnd that you
have skipped to certain enablers without a clear sense
of what’s next.
Our aim is to provide a 10,000 foot view of these
stages. We hope it will help you assess gaps and spot
potentially missed opportunities along the way

the big conversations in the robotic process automation (RPA) space this year are centred on adoption.

James Dening, Vice President for Europe at Automation Anywhere says the big conversations in the robotic process automation (RPA) space this year are centred on adoption.

His advice to organizations that are planning to scale up their use of RPA is to talk to people who have already deployed RPA at scale and find out how they’re doing it.

What are the big conversations happening within RPA this year?

So, for me, the big conversation this year has been about adoption. There’s a huge chasm, still, between people who know and people who don’t quite know yet. And it’s a bit like, if you remember the book, Crossing the Chasm, which was all about start-up’s.

Building your first five robots isn’t that hard, building your first 500 is much harder, and the companies that have figured that out and are churning out robots at an industrial level, and are getting that return on investment at real scale, they’re the ones who are really succeeding.

What is the advice you would give to people who are looking to scale up?

It’s a really good question; I think there are two ways to look at this, one is where do you get that advice, where do you get that knowledge from? I would say that the first place to look at is go and talk to people who are already doing it. So, these shows are a really good opportunity to find delegates who are rolling out robots at scale, and go and talk to them about how they’re doing that.

The second way to look at it is what are the different things you need to figure out, things like best practice, centres of excellence, governance models, all of those things. The frameworks within which you operate your RPA programme, those are really important. You can get those from customers, you can get those from vendors, if you come and talk to Automation Anywhere, if you go and talk to UI Path, those people will help you. They’ll point you in the right direction.

Who should head up automation, the business or IT?

So either can work, but the thing to always remember is it’s business that benefits from an RPA rollout, so business really should be steering the ship, with support from IT. I think the reality is you need to have both involved, and one of the things that we certainly talk about a lot is the importance of getting those stakeholders, both from the business, from senior management, and from IT, getting those stakeholders involved as early as possible is absolutely essential if you want to succeed.

What is the golden line you would use to get an executive on board with an RPA solution?

It’s really about fast return on investment; with traditional IT transformation, you might not see a return on investment for nine, 12, 24 months, maybe even longer. With RPA, if you get it right, if you crack the code on building robots, you can see your return on investment in a few months.

So, I think both the speed you get that return on investment, and also, the scale; if you go in to talk to people like AT&T, or ANZ Bank, or UBS, they have hundreds and thousands of robots, the cost savings they’re seeing, and the ability to redeploy staff, this isn’t a few hundred thousand dollars a years, these are millions, or tens of millions of dollars a year are being realised to these companies through the judicious application of RPA, and that interests everybody.

What are the major challenges faced in the automation space right now?

One of the problems that I see at the moment is people are getting distracted by the next generation technologies. So, AI, cognitive, we have our IQ bots, and there is absolutely a place for those, and we have customers who have rolled out a lot of RPA robots, and are now using cognitive for that next generation, to deal with that semi-structured data. But for most customers out there, most companies out there, they still need to focus on getting their RPA deployment right, and not be distracted by the shiny AI, the promise of the future. So, I think that is a problem, that distraction.

Recently Guy Kirkwood, UIPATH, stated that AI is not real; what are your thoughts on this?

So I think there’s a nuanced answer, do self-learning robots exist? At one level, yes, they do, our IQ bot takes semi-structured data, and the more it processes, the more it learns, the more user cases it sees, the better it gets. So, it’s an adaptive algorithm that processes data better and better the more data it’s exposed to.

But I think what Guy is talking about is there are there robots you can just point at the business and say go learn how to do stuff. Now, today, no I think he’s right, that doesn’t exist. Will it happen in the future? I think it will, and I think it will happen quite quickly.

If you look at the way that Automation Anywhere builds robots, we record what people do to give you the bones for a robot, it’s not a huge leap for us to have something that essentially looks over somebody’ shoulder without explicitly being taught, and looks for patterns in what they’re doing, and how they’re processing data. So, yes, I could absolutely see that happening over the next two or three years.

The 4 phases of digital transformation: a roadmap to Intelligent Automation

A nice read from workfusion


Automate Mistakes.jpeg


It’s the end of the year and you’re feeling stuck. You’ve reached the end the road in outsourcing. You’ve been dinged by potholes of legacy systems and your smartest people are too busy struggling under the load of paperwork. You suspect that there’s only one way to get past these roadblocks, and that’s to start a whole new journey. Next stop: Intelligent Automation. The only thing is that you have no idea of what you’ll encounter along the way…

The good news is, there are people who do. WorkFusion’s Client Strategy and Transformation team, which focuses on strategic advice and programmatic enablement for enterprises who are embarking on robotic process automation initiatives, has been down this road and around the block a few times already. They have seen patterns emerge and learned from their experiences. Which is why they wrote The 4 Phases of Digital Transformation: The Intelligent Automation Maturity ModelThis complimentary 10-page eBook by WorkFusion will help you determine the best strategy for your operation by mapping each of the four stages of maturity that are relevant for most organizations.

Early stage

This is when you’re actively pursuing a POC or pilot. To successfully get your program off the ground you need to focus on:

· Use case identification — Discovering the right (or wrong) use case is the first step in understanding how to apply the chosen Intelligent Automation tool.

· Change management #1 — Implementing Intelligent Automation requires additional planning and implementation tools than BAU change management.

· Operations & IT Strategy — Developing an initial infrastructure platform doesn’t require large-scale acquisitions, you just need enough to house the production load for a few use cases.

· Automation delivery methodology #1 — Start by exploring different methodologies and delivery centers to plan for the road ahead.


By now you have at least one use case actively in production. To keep it running smoothly you should apply these tactics:

· Cost benefit analysis (CBA) #1 — Identify the initial costs and benefits of your program by estimating high level costs based on the initial use cases.

· WorkFusion Lean #1 — This program gives you a basic understanding of best practices for reengineering processes prior to automation which helps you optimize basics and assess gaps.

· Book of work planning — To get the Enterprise-grade benefits of IA, you will need to perform large-scaling book of work modeling across multiple business and/or timelines.

· Smart Process Automation delivery methodology #2 — Setting up your factory or augmenting your CoE. This is when your methodology moves from tactical program and project management to decisions regarding delivery and whether to centralize.

Growth stage

This is the slightly challenging phase when you have multiple disjointed use cases in production. Keep the momentum going with:

· WorkFusion Lean #2 ­­– Advance reengineering techniques as part of your IA footprint with WorkFusion Lean’s Practitioner Program.

· Power User Program #1 — This is the right time to start thinking about how you might support a citizen developer model.

· Scaling IT — Provisioning the appropriate infrastructure can begin as early as mid-stage, but we often find that it’s most useful to expand in the growth stage.

· Change Management #2 — Change management plans that cover the stages of change are typically ready for execution in this stage.

· Power User Program #2 — Before you distribute this technology as an EUC (end user computing tool) it is essential to “work out the kinks”.

Mature stage

Phew! You’re out of the weeds and have a centralized program that’s ready to scale. This is how you do it:

· Portfolio management best practices — Advance from KPIs to business intelligence. Get ahead of dashboard standardization to be ready to connect to broader data visuals and reporting.

· Modernize IT — As technology changes, so do the options that are available to support it. Now you might be ready to move to a cloud platform, or even Intelligent Automation as a service.

· Ad hoc thought leadership — A necessary final step in becoming an IA BAU enterprise is to ideas reach out into a more mature network to leverage the best practices of others.

Along with mapping out these stages in detailThe 4 Phases of Digital Transformation: The Intelligent Automation Maturity Model will also will help you determine the best strategy for your operation, make critical decisions based on best practices, and catalyze change by identifying key signals within your business.

To learn more about the CST team and how to expand your RPA program to take advantage of all our Intelligent Automation techniques, visit



How to get superior employee experience: AI, robotic process automation can help reach desired goal

RPA Common Cases HR.JPG HR function traditionally has been geared to create and implement HR policies aimed at smooth functioning of the business with a long-term perspective. As a result, all its activities around HR administration, training, recruitment and other employee life cycle are mostly designed for steady-state business. With the onset of the digital era and the business dynamics resulting in the need for agility and quick adaptation, HR function needs to redefine its functioning and orientation to business needs. The fundamental premise around customers, customer needs, business model, competition landscape and offerings is undergoing changes in the context of digital transformation and hence HR function too needs to become agile and supportive of the dynamic needs of the business. Building a culture of flexibility and shift in the focus towards delivering solutions has become important in order to be counted as reliable partners to the business.

Just as businesses are recognising that customer experience is central to their success, delivering delightful employee experience has to become the core of HR function. As a partner to business, HR function has to play an important role in propelling the organisation to digital workplace. Use of appropriate digital technology required for its own function should be thought through carefully including the use of digital tools such as AI and robotic process automation to deliver superior employee experience. Learning support has to go beyond the current approach towards learning or content management system.
Artificial boundaries prescribed for learning should cease to exist and static content for learning should be replaced with dynamic and personalised content, based on customised learning paths, encouraging continuous learning by providing access to innumerable learning resource inventories that could be curated on the digital platform from multiple avenues.
With talent continuing to be the key differentiator for businesses, the tasks related to acquiring, grooming and retaining quality talent for growth and sustenance of the business would have to be managed with extreme care as resources would expect personalisation and customised employee journeys during their association with the organisation.
HR heads have to therefore be willing to revisit all dimensions of the traditional HR model and using design thinking approach, visualise the new-age employee journey. What this may lead to is the necessity to relook at systems that may be working in silos to facilitate seamless functioning of individuals and functions.
IT systems in HR have traditionally mirrored the processes followed prior to automation. Most benefits of automation for HR have been on account of reducing labourious work and time in administrative processes such as payroll, compensation and benefits and attendance management. HR function should embrace digital tools empowering HR professionals and the employees and help design customised services as per the preferred mode of access with respect to place and time of their choice.
Organisations can respond to queries of employees with speed and accuracy with the help of AI bots, have the opportunity to proactively address employee concerns and or initiate timely steps regarding motivational issues or likely attrition in critical roles. Employee life cycle and their journeys throughout different phases in the organisation need to be considered as a continuum and an integrated approach is required for serving and supporting employees such that the success of the individual and the business are synchronised. Developing analytical capabilities within HR function and developing useful insights juxtaposed with business parameters would enable HR function to actively contribute to the strategic trajectories of the business.

The insights drawn from analytics could go a long way in deciding effective spend on employee engagement as well as in building linkages with learning and development for maximising performance management.  HR function also has an important role to play in fostering innovation in the organisation. Building communities of practice within the organisation to nurture best practices, new ideas and innovation and connect them with the expertise available in academic institutions and other organisations in the ecosystem would be essential for sustaining its competitive advantage in the marketplace.

The writer is CEO, Global Talent Track, a  corporate training solutions company