In 2012, the Knight Capital Group lost $460 million in 45 minutes because the systems were not setup for the risks the intelligent solution exposed them to¹ – this was down to a lack of proper governance mechanisms across the bot lifecycle. Today AI & Automation solutions are at the forefront of digital transformation and have become an integral part of complex IT landscapes. These solutions leverage a myriad of tools / technologies, from script-based automation to cognitive automation, which have resulted in a bigger challenge – how does one track, operate, and control such varied solutions and ensure that they are governed well – from inception to operations. Take another case - a large banking customer had more than 80 tools deployed for automation across business units, creating a double whammy for operations – to spend time and energy on managing the risks & operations of such AI & Automation solutions, which were originally put in place to simplify their work and make them more efficient and effective.

What are the current ‘governance’ challenges?

While governance processes have been around for several years now, with the advent of these new Intelligent Automation solutions, current governance methods cannot scale up and will require an overhaul. Below are a few of the key challenges:

• Non-centralized, myriad of automation / AI solutions – Different AI & Automation solutions in any organizations, for e.g.– shell scripts, python programs, RPA programs, RBA solutions etc., with different maturity, little or no-standardizations make it very difficult to manage, measure and report in a centralized manner.
• AI Solutions - atypical IT solutions – Managing AI solutions is very different compared to typical IT solutions for e.g. – risks, explain ability, constant accuracy / precision / checks, A/B testing etc. – and requires a completely different level of governance.
• Business value riddle – Given the complexity and mix of solutions, and no clear traceability / tracking, it is difficult to clearly measure the cost & ROI of these solutions.
• Reactive reporting – Most of the current reporting solutions still show data reactively. The need of the hour is to have solutions that can be more proactive, prescriptive and predictive.
• Regulatory & legal compliance – With the increasing scrutiny and stringency of the legal and regulatory compliance w.r.t. privacy, data regulations, customer interactions, etc., it is becoming a challenge to manage the AI & Automation landscape.

Overall, these challenges are derailing the whole purpose of such intelligent bot solutions – since the entire effort that is saved due to such solutions, will be spent in managing these intelligent automation solutions – which defeats the whole purpose.

Unified governance – A Uberized solution for AI & Automation landscape

The new age Unified Governance Platform should not only be able to support the existing legacy automations but should also extend to newer tool-based (or custom) intelligent automation solutions. It should provide a single pane of glass–to various stakeholders to manage their part in the overall lifecycle of such solutions.

The key aspects / features that the solution should cater to are:

• Bot lifecycle management – The solution should provide a capability to manage the entire bot lifecycle management - from ideation to execution, to deployment and to post activities. It needs to track features, cost details, efforts, timeliness, versions, issues / enhancements and deployments – centrally.
• Bot ops & control – This is one of the most important capabilities required from the solution. It should be able to audit, manage and control all the bots running in the landscape. It should provide capabilities to manage the various kinds of digital workers, their health statuses, debugging and remediation remotely, auditability, bot execution management and also scale up /scale down the number of digital workers based on the needs – thereby providing  complete centralized control of the entire automation ecosystem.
• Automation maturity assessment – As any organization-wide automation program progresses, it is important to understand its maturity across the organization i.e. combining the data from various bots and determining the acceptance and depth of automation as its progress across  the organization.
• ML ops – The AI programs / solutions require a very different operability as compared to typical IT operations. The governance solution should be able to manage the training, testing, deployment, monitoring, re-training, measuring (precision / accuracy), scaling up / scaling down etc. across each of  the typical AI Models deployed in production. In addition, the solution should be able to ensure that the AI solutions deployed are auditable, explainable, transparent, and ethical. This is extremely important given the risks and new regulatory and legal compliances that have been set in motion for these solutions – and will help the CIO / CRO / CISO organizations to ensure that the AI & Automation solution is in line with these principles and regulations.
• Business value articulation – The governance solution should provide a clear glide path of the benefits derived from any automation use cases (cost / effort / time), with clear business value articulation, which is objective and data based, rather than qualitative and subjective. This will help the enterprise prioritize their automation projects appropriately.
• Ecosystem ready – The governance solution should have various OOTB adaptors with an ability to integrate with different types of AI & Automation solutions – which use different technologies, are a mix of bespoke and product based, with different maturities and lack of standardization.
• Reporting / dashboarding – The governance solution should provide a variety of access-controlled dashboards / reports, be simple to use, and provide an enriching  experience to the end user – by providing  whatever information the stakeholders want, whenever they want it, with little configuration.
• Prescriptive / Predictive analytics – The governance solution should have the capability to derive insights and provide proactive, prescriptive and predictive analytics w.r.t. various digital workers. These insights can help in the planning, operations and maintenance of various bots or digital workers proactively and preventively – there by ensuring that the AI & Automation landscape is able to run efficiently, and is robust and reliable.

Conclusion

There is a pressing need for a centralized, unified governance solution that can manage, control and operate such diverse solutions. This helps in eliminating risks and ensuring compliance without impacting the flexibility or adding any load onto the existing operations teams. This new-age governance solution has to be multidimensional and provide an easy, effortless, enriching and engaging experience to all end users and stakeholders.

Reference

¹Gartner IT Symposium Expo – Nov 2019

Nikhil Mehta

Lead Architect, Wipro HOLMES™, Wipro Limited.

Nikhil is the Product Owner of Wipro HOLMES™ AIOps in the “HOLMES for IT” product line. With 16+ years of industry experience, he has played a pivotal role in setting up and developing AI product lines on the Wipro HOLMES™ AI platform. He has successfully delivered AI-enabled IT transformations that were focused on responding to IT operations, improving performance and supporting growth. Nikhil has played various roles across multiple business functions like Product Engineering, Product Management, and Service Delivery; with an expertise in defining, building and deploying solutions across multiple domains such as Logistics, Telecommunications, Supply Chain and Retail.

Vaibhav Bakshi

Consultant-AI and Automation, Wipro HOLMES^TM, Wipro Limited.

Vaibhav Bakshi leads the New Solution Definition function for Wipro HOLMES™. With over two decades of experience in the industry across functions such as Consulting, Program Management, Engineering, Solution Architecture and Service Delivery, Vaibhav specializes in delivering business outcomes using AI & Automation. He has delivered these solutions across multiple manufacturing and technology customers across the globe.