Best Practices & Templates related to Enterprise AI-automation Strategy: Charter, Partner Selection, Risk-returns
Best practices for AI-automation workforce management: Upskilling on non-tech skills like design thinking in AI
Must-do practices & templates for AI-automation data management: Debiasing, relevance, security management
Examples of generic architecture templates for AI-automation solutions design
Enabling key culture-change levers for AI-automation: Communication frameworks
Best practice templates & metrics examples, balanced scorecards, RoI reporting systems
The first S of AI-SWITCH (IP) is Strategy- which is the first place to start, for any enterprise embarking on the AI-automation journey. Strategy management includes: Partner assessment, current-state maturity benchmarking, failure mode analysis, AI-automation policy-making & governance models, prioritization of initiatives/ use-cases, cost estimation & RoI projections, and journey-mapping.
AISWITCH_2021 CALM3 AI LEADERSHIP FOR AGILE AI STRATEGY (pdf)Download
AISWITCH_2021_AI MVS CALM1 TOC CONSTRAINTS (pdf)Download
AISWITCH_2021 CALM4 AI MVS AGILE STRATEGY (pdf)Download
AISWITCH_2021 ENTERPRISE AI STRATEGY EXAMPLES (pdf)Download
AISWITCH_2020_AI STRATEGY COST FOCUS DIFFERENTIATION (pdf)Download
AISWITCH_2020_AI STRATEGY HOW TO USE MCKINSEY 7S (pdf)Download
AISWITCH2020 How to Evaluate AI implementation Partners Objectively with 5 Next Practices (pdf)Download
AISWITCH2020 Ethical AI Manifesto- 12 Principles (pdf)Download
AISWITCH2020 AIOps Maturity Model (pdf)Download
AISWITCH2020 Why AI fails (pdf)Download
The W of AI-SWITCH (IP) is Workforce- the most critical lever in an organization's AI-automation machinery. Workforce management for AI-automation deals with the challenges of AI-automation CoE planning, organizational structure, conflict management, skill scarcity/ skill-gaps, digital reskilling requirements, training & certification plans, new KRAs & KPIs for both tech roles and business analysts/ business user roles. These are the critical steps within AI-automation workforce management. Following best practice research notes briefly describe these processes & applicable frameworks.
The I of AI-SWITCH (IP) is Information (Data)- which is literally the Eye of the AI storm. Data quality vs. quantity is a major issue- be it process related data- discovered and mined by process mining tools like Celonis, or the training, test and validation datasets for different AI usecases, e.g. structured, parametrics or non-parametric data, to unstructured data like text corpus and image databases, to hybrid data. Apart from the well-researched noisy data and sparse data issues, a critical question of data quality is the presence of multiple types of biases in training & test data. There are also the questions of data relevance and less-data scenarios, e.g. post-COVID, many of the underlying assumptions have changed in case of historical databases and data patterns - from economic forecasts to stock markets to healthcare to pharma to manufacturing to supply chain datasets.
The T of AI-SWITCH (IP) is Technology- the critical underlying tech-stacks, from AI-optimized chip to the APIs, that power and run the AI-automation solutions for businesses. Managing the tech-stacks for AI-automation requires a drastically different view about the key assets, given that the assets include datasets, APIs, dedicated sandbox environments for testing, runtime infrastructure which is often a mix of cloud and on-prem. They require highly secure infrastructure which is often hardcoded at the platform level e.g. INTEL SGX. The architectural dependencies between assets are complex and spaghetti-type. Hence, the AI-automation asset catalogues have extensively different structural requirements. So do the process models for AI-automation major incident management, capacity management, SLM, change & config management, availability, reliability, continuity & security management.
The C of AI-SWITCH (IP) is Culture- which is universally accepted across all major surveys of enterprise leaders, as the Most Difficult Problem that's accentuating the AI-automation adoption & scale challenges. Contrary to how most of us feel, more often than not we the practitioners find that technology is the least of the problem, and Culture eats not just strategy but technology, information, workforce reskilling plans- pretty much everything- for breakfast, lunch & dinner. Changing the organizational culture, through processes and interventions, no matter how good the internal or external Change Masters are, is much easier said than done. But communication plays the most important role as a culture change lever, no one can deny that.
The last H of AI-SWITCH (IP) is Human-AI augmentation metrics. The ultimate challenge of AI-automation implementations are to predefine the target outcomes and then measure the post-implementation impact metrics and declare the initiatives either as successes or failures. The most effective human-AI augmentation metrics are always focussed more on strategic outcomes, impact and value, than purely based on operational cost reduction. A balanced approach towards the AI-automation outcomes metrics, makes the claims of Wins credible, data-driven and evidence-proved.
Here is a quick collection of short (2-6 pagers max), evidence-based AI-automation implementation best practices & frameworks (new materials uploaded every week), including actual application examples in specific client scenarios. For detailed consultative project reports & templates: firstname.lastname@example.org