How can you advise clients on the use of artificial intelligence in their business operations?

In my experience, AI is a huge body of work to embark on. Not only do you need to invest in the necessary infrastructure and tooling, you also need to invest in creating suitable data lakes and business process modelling. Understanding what implications AI have on the business and its workforce is of primary concern. Blindly embarking on AI because "AI" can lead to bad business decisions that may cripple a business' profit or balance sheet. Equally, simply adding a badly trained chatbot for first-line customer support can turn away customers. Yes you have AI, you adhere to the hype, but does it really serve your business?

Explaining to the business what AI is/can do but also what it is NOT/can't do is integral to assessing readiness/ level setting with regards to expectations

Achieving AI readiness in a company goes beyond mere financial considerations; it demands a commitment seldom spoken about. This includes cultivating a suitable company culture, nurturing a skilled talent pool, developing robust data capabilities, and investing in the necessary infrastructure. A company should educate executives on the topic prior to embarking on an AI project or program of works.

It’s best to think of AI and its impact in layers. AI will revolutionize your business in so many ways. Start with productivity Look for ways in which AI can make your employees more productive this is usually in the form of chat bots, virtual assistants, meeting management and more. Next look at how AI can impact your sales/marketing/ efficiency. From there it can become more complex as you look at ways integrate Ai into areas like Operations Supply chain Software development Human resources and more In parallel build an AI governance model that includes a strong AI ethical standard. This ensures people will feel more comfortable and will buy in to the program. Education and enablement of your employees is also critical.

Start by evaluating the client’s current operations to identify areas where AI can be beneficial. This involves understanding their business model, workflow, data management practices, and any existing technology infrastructure. Look for processes that are repetitive, data-intensive, or prone to human error, as these are typically prime candidates for AI enhancement. For instance, if a client has a large customer service operation, implementing AI through chatbots or automated response systems could significantly improve efficiency. Understanding the current state of the client's operations helps pinpoint where AI can add value and crafting a tailored AI strategy that aligns with their specific business needs and goals.

The first thing to identify in the initial assessment is the company's data strategy. AI initiatives rely on the quality and relevancy of the data. A data strategy needs to focus on collecting, managing and governing data in a way that ensures its accuracy and relevancy. It also entails how current personnel is using data in their daily workflows. If no data strategy is employed, the first step in the roadmap should be data literacy.

Start by thoroughly understanding the client's business objectives. Identify key areas where AI can enhance operations, such as automating repetitive tasks, improving decision-making, or enhancing customer experiences. Demystify AI by providing a basic understanding of its concepts, benefits, and potential applications. Highlight real-world examples in similar industries to showcase the practicality and positive impact of AI solutions. Conduct a thorough assessment of the client's current operations. Identify pain points, bottlenecks, and areas where AI can streamline processes. This analysis provides a foundation for tailoring AI solutions to specific needs.

Most organisations (and to be honest most people) have *no idea* what AI actually is, or what it means (or could mean) to them or their companies. So unless you have a company that has identified a very particular problem and is looking at a set of appropriate AI/ML solutions, then you've probably got a cultural/transformational challenge, not an AI one. So you need to focus on the current organisational culture (and skills), identify the gaps that need to be filled to adopt and iterate a successful AI solution, and set out a pathway to a future organisational culture (and skills) that builds organisational capacity to enable successful AI adoption. Otherwise they'll have tools they don't understand - which means tools they won't use.

This is fundamental. AI is a layer upon which data and automations can be expanded. Unfortunately many non profits still do not have good systems or data so they need to set that up first.

After an initial assessment, work closely with your client to define the scope and roadmap for AI implementation. Clearly outline the impact-driving milestones and key performance indicators (KPIs). Communication and alignment on the scope are critical for setting realistic expectations and ensuring a smooth transition to AI-powered operations.

Based on the initial assessment, work with the client to define the specific areas where AI will be integrated. This might include processes like customer service, data analysis, inventory management, or predictive maintenance. Once the areas are identified, develop a roadmap that outlines the steps for implementation. This plan should detail the timeline, the technologies to be used, the required investment, and any changes needed in the existing infrastructure. For instance, if AI is to be used for enhancing customer service, the roadmap might include steps like selecting the appropriate chatbot technology, integrating it with the existing customer management system, training staff, and a phased rollout plan.

If the company has an established and effective data strategy, the first step towards establishing the scope of the project is identifying which business goals align with common AI use cases. The value generated needs to exceed the implementation cost, as not all use cases will deliver impact and deliver ROI. The ROI will come in one of three major categories: productivity gain, minimized cost, or automation. The category will help provide focus to the project and a clear roadmap can be now created based on quantifiable business impact.

At the moment, I would recommend the client to adopt a tech agnostic approach to solutions to the extend possible. We would want to dodge a situation where the clients depends almost exclusively on a platform. For that, multiple frameworks and tools need to be assessed (LLM orquestration tools like Langchain/LlamaIndex, vector databases like Weaviate/Pinecone, etc.) taking into account the creation of a sustainable production environment. Also, deciding whether open source or closed LLM models will be used and for what use cases is an important part in this step.

After defining the scope of AI implementation, assist your client in selecting the most appropriate AI technology and vendor. This involves analyzing various AI solutions to find one that best fits the client’s specific needs, budget, and infrastructure. Consider factors like the scalability of the technology, ease of integration, support and training services offered by the vendor, and the vendor’s track record and reputation. Facilitating the selection of the right technology and a reliable partner is crucial for successfully implementing AI in the client’s operations, ensuring that they reap maximum benefits from this integration.

Choosing the right technology and the right partner is crucial for any organization that wants to adopt AI. Be sure you consider the specific domain covered by a technology and partner and leverage it for your business. Popularity does not determine usability; the most popular model might not be best suited for your organization's needs. Check the cost of maintenance and deployment of the AI model for your business in order to maximize profit and reduce cost.

They say prompting is an iterative process, but AI implementation is too. We continually need to monitor this technology, the impacts and the effects and tweak our designs to maximise the output.

Not disimilar to implementing a complex RPA, effectively implementing and monitoring an AI solution is a comprehensive process that encompasses technical deployment, team training, continuous performance evaluation, and ongoing adaptation and optimization. This holistic approach ensures that the AI solution remains effective, secure, and aligned with the evolving needs of the business.

Assist your client in the rollout of the AI solution, ensuring it integrates smoothly with existing systems and processes. Once the AI system is operational, establish monitoring protocols to track its performance and impact. This should include key performance indicators (KPIs) relevant to the AI's objectives, such as improved customer response times, increased sales, or reduced operational costs. Review these metrics with the client regularly to see if the AI solution is effective and identify areas for optimization. Monitoring also helps detect any issues early on, allowing for timely adjustments or interventions.

Success hinges on the effect of the AI on your client's business operations. Assess the Return on AI Investment (ROAI) as defined at the outset of your AI project. Additionally, gauge user success by examining their perception and adoption of the system. A comprehensive evaluation and resolution of any problems, coupled with a readiness to make essential modifications, can revitalise a faltering AI implementation, ensuring it better meets the client's requirements and anticipates their future needs.

After the AI system is implemented, set up a schedule for periodic reviews with your client to evaluate the solution's performance. Use these reviews to analyze the data collected from the monitoring phase, comparing it against the pre-defined KPIs. Identify areas where the AI system excels and, more importantly, where it falls short. Based on these insights, work with your client and the AI vendor to make necessary adjustments or upgrades. Regularly revisiting and refining the AI solution ensures it remains effective, efficient, and aligned with the evolving needs and goals of the client’s business.

Involve people! The reason why most digitalisation efforts fail is not due to the lack of technical skills in the organisation, but rather due to bad stakeholder management. To succeed in truly integrating AI in business operations, mix strategy with more innovative and agile methods. Examples of strategic initiatives: Assess the AI maturity of the company. Set up a roadmap. Establish clear guidelines for data integrity and use of AI tools. Communicate it with the relevant stakeholders. Examples of innovative and agile methods: Create a competition or workshop where employees can contribute with their ideas on how to use AI for work. Start with smaller, internal (non-customer facing) projects and harvest insights about the organisation.

Always keep the human in the loop. Just letting AI run wild is not sustainable or viable. Ensure humans review any AI produced outputs before using it.

You need to assess the culture of the company all the way to the front line. Artificial Intelligence is the boogie man to many people and could reduce productivity in the work force (in the near term) if your change management process does not factor in the emotional response your workforce may have to this “strategy”. This mixed with the current climate of companies that continue to have mass layoffs in the name of short term growth, you may not get the adoption or lift from AI you anticipate with a naive launch.