Data readiness is your first and foremost step in starting with AI. We all know that data is beating heart of AI and the quality and integrity of your data directly impacts the effectiveness and reliability of your AI outcomes. As I represent a global software company, I regularly work with clients from around the world of which the majority are getting their data house in order. And if they’re lucky they have already begun this effort and need to pivot specifically to AI readiness. This is where I want to begin this blog, by answering how clients are pivoting their data strategy and where we see AI readiness going in 2025.
The importance of AI readiness and AI-ready data for your strategy
AI readiness is really about getting your data house in order and preparing your organization for successful and safe use of AI. It’s setting your AI strategy, your ambitions and educating your workforce on what AI is and how to work with it.
The risks of not achieving AI-readiness
It’s been stated that 74% of companies struggle to achieve and scale value from AI and that at least 30% of GenAI projects will be abandoned after proof of concept this year. Given these intimidating stats, it’s fair to say that there are some significant risks to your organization if you’re not AI-ready.
- AI development never making it to production: Not having available and trusted data for your new AI development.
- Unreliable results from AI and analytics: Your AI models and analytics may produce unreliable results due to data inconsistencies, drift and duplication.
- Lack of context: Business context around the data is just as import as the data for LLM’s and AI Models
- Increased risk and exposure: You could suffer from a lack of AI data governance with incomplete lineage tracking and unregulated AI model training.
- Delayed AI development: Your data preparation and validation processes could be slowed by manual data curation and quality control.
- Regulatory uncertainty: Your compliance teams might struggle to meet emerging AI governance mandates.
- Limited data literacy: Your business users may lack the tools to understand and trust AI-driven insights.
Understanding AI-readiness and the role of AI-ready data
We are at the next stage (next meets now), bigger than all those that went before us with AI readiness, explainable AI and responsible AI. Call it what you want, but we need everyone playing this game and fighting the good fight now of data transparency, observing data, collaborating and aligning data behind AI in an ethical sense. I say this from a risk-based approach but it’s also the simple fact that if you want your AI to get smarter and better, you need your data to continuously improve as well.
What is AI-ready data?
AI-ready data is data that is structured, cleaned, governed and contextualized to ensure it is reliable, accurate and actionable for your AI applications. It supports the effective and ethical deployment of AI by being complete, consistent and aligned with the specific requirements of your AI use cases. AI-ready data is characterized by robust data governance, quality management and continuous monitoring to maintain its integrity and relevance over time.
So, how are clients achieving AI readiness today?
Step one is always building and vetting your AI strategy with guidance provided by an AI Center of Excellence (AI CoE) or governance council. Collaborate on your AI mission and answer questions like:
- What’s our appetite for risk?
- Are our corporate values and ethics widely known and crystal clear?
- Will the AI we are developing be internal and external facing?
- Are there any boundaries or can we consider Agentive AI?
- How will we monitor and validate the source data?
- Who is responsible for AI gone wrong?
- How will we enable and educate our company on AI and data literacy?
- What data solutions can we leverage to automate and mature our data?
The AI strategy is a living body that will be owned, curated and delivered by a C-level executive. We typically see the office of the CDO, CIO and in some cases, a new role has arisen… the CDAIO.
AI governance
At a foundational level data governance offers a way to find known data that is described, curated and contextualized. Instead of pointing LLM’s at any data and all data we want our LLM’s to be using well defined, smart data for smart answers. Data modeling technology begins with creating a business information model (BIM). This naturally describes your data from a business perspective and begins to tag the logical view with business descriptors and how to use the data. This is often used as an accelerator when creating your business glossary where business stewards begin to marry up the business terms, rules and policies to the data.
In a recent Gartner ThinkCast podcast, “Pace yourself in the AI Race”, they discuss how the rooms in your data house can be quite messy. Given that the AI can find anything, even the dirty socks under your bed, do you really want it to find your dirty socks? Umm, no. But now you can see that your data governance needs to be just as fast and ready for AI. This is where AI data governance and the analytical marketplace enters the picture.
So, how will you find the dirty socks under the bed before AI does? AI governance will need to outpace AI and data governance is not known for being uber fast and effective. To me, this is a new strand of governance just for the analytical space.
AI-ready data: Key considerations for AI readiness
Here are some aspects and capabilities to consider to set a new course on a very old problem of keeping a handle on your data:
AI education and enablement
Unite the business versus technical resources and mandate that the business understand what AI is with some form of enablement and the same with the technical resources. They must know what the core KPI’s of your business are and what the organization mission is across all lines of business. There are many new AI concepts and considerations. Teach them what an LLM is, core data quality principals, how to write an effective prompt etc. Think of it like this way… smart people make smart data which returns smart answers from AI.
You might want to consider an AI Center of Excellence to give your programs structure and roles within the structure to keep AI on track with your intent. Your new AI strategy and purpose will be funneled through the teams on the ground. Consider subject matter experts from multiple functional areas so that you can make decisions quickly and effectively as data typically knows no borders and the same data can be used in different ways with different meanings. The goal is for meaningful results and outcomes you can act on.
Channel data and create collaboration techniques
Other capabilities used to effectively speed up AI development and results is an analytical marketplace. A shopping cart for data products, AI models and components (e.g., vectors, API’s, agents, prompts). Track their performance, score and certify the AI models, and curate the information.
Use the efficiency of AI to make data recommendations and suggestions. Actionable metadata is proactive in nature and personalized. Let AI help you clean up the room and warn you that there may be dirty socks under the bed in a room that could be exposed to guests. If we don’t have time to clean the entire house, we should start in the guest room perhaps.
Ambition and risk
You already set your AI ambitions within your data and AI strategy. Whether it is a light, loose or big bang approach with great power comes great responsibility. Let’s turn the dial to the latest installment of the EU’s AI act where regulators can ban the use of AI where there is “unacceptable risk” or harm. In Article 5, they list out just what is considered high risk:
- AI used for social scoring (e.g., building risk profiles based on a person’s behavior)
- AI that manipulates a person’s decisions subliminally or deceptively
- AI that exploits vulnerabilities like age, disability or socioeconomic status
- AI that attempts to predict people committing crimes based on their appearance
- AI that uses biometrics to infer a person’s characteristics, like their sexual orientation
- AI that collects “real time” biometric data in public places for the purposes of law enforcement
- AI that tries to infer people’s emotions at work or school
- AI that creates — or expands — facial recognition databases by scraping images online or from security cameras
“Companies that are found to be using any of the above AI applications in the EU will be subject to fines, regardless of where they are headquartered. Companies could be on the hook for up to €35 million (~$36 million), or 7% of their annual revenue from the prior fiscal year, whichever is greater.”
Conclusion
AI moves fast, near real time. AI gets smarter over time and your data will too when you have some automation in place with a combination of humans and machines. As we move further into 2025, AI readiness will continue to be a critical focus for organizations aiming to leverage AI effectively and responsibly. Ensuring AI-ready data through robust governance, continuous education and strategic collaboration will be paramount. This doesn’t need to be a laborious governance process; in fact we recommend a sprint with a product-based approach using “Govern as you Go” to become data ready at this point. Companies must prioritize data transparency, ethical AI practices and proactive risk management to stay ahead in the rapidly evolving AI landscape. By fostering a culture of data literacy and aligning AI initiatives with corporate values, organizations can unlock the full potential of AI, driving innovation and achieving sustainable success.
