KIBO AI https://kibo-ai.com Turning complexity into clarity with AI. Mon, 15 Jun 2026 09:47:21 +0000 en-US hourly 1 https://wordpress.org/?v=7.0 https://kibo-ai.com/wp-content/uploads/2025/07/kiboai_favicon-16x16-1.png KIBO AI https://kibo-ai.com 32 32 Machine Learning Implementation for Business https://kibo-ai.com/insights/articles/machine-learning-implementation-for-business/ Mon, 15 Jun 2026 06:36:34 +0000 https://kibo-ai.com/?p=4253

Machine Learning Implementation for Business

Rusi Yang (Consulting Manager)

kibo-ai-from-data-to-decision

How Kibo AI Delivers Machine Learning in Real-World Programs: A Story-Driven Journey from Fundamentals to Forecasting

What Is Machine Learning Implementation in Business?

Most companies have years of sales, operational, and financial data sitting in disconnected systems. Decisions on top of that data are still being made with manual spreadsheets, gut feel, and reports that took three people a week to build. Machine learning implementation is the structured process of turning that raw data into a working prediction system, covering 12 defined steps from problem definition through to deployment and continuous improvement. Organisations that complete all 12 steps correctly report invoice processing up to 90% faster, fraud detection accuracy above 90%, and cash flow forecasts accurate enough to cut late payments in half.

Overview

How Machine Learning Appears in Daily Life

Machine learning runs quietly in the background of most industries today. In business specifically, it forecasts revenue, identifies risks, classifies emails, understands documents, allocates resources, prioritises sales pipelines, and automates processes that used to require hours of manual work. When implemented correctly, it becomes an invisible engine for better decisions.

How Machine Learning Is Implemented in Real Companies

Implementing ML is not an algorithmic exercise. It is a structured business journey. The objective is simple: bridge the gap between “we have data” and “we make smarter decisions every day.” First comes clarity on the business problem. Then comes the search for the right data, followed by exploration, cleaning, and transformation. Only then can models be trained, validated, deployed, and improved. The following story shows exactly how that journey looks in practice.

Alice’s Story: A Human Introduction to ML Before the 12-Step Machine Learning Methodology

Alice had just joined GreenGrocer, a local organic supermarket chain that struggled every week with the same painful problem: shelves overflowing with spoiled berries, while avocados ran out by mid-week.

Everyone had theories. No one had answers. Alice decided this would be her first real project.

1. Understanding the Problem

Before touching a line of code, she sat with store managers, cashiers, and buyers. She asked simple questions:

  • “When do you usually run out?”
  • “Which items cause the most headaches?”
  • “What makes demand unpredictable?”

The answers painted a clear picture: GreenGrocer didn’t need ‘more data’ — they needed clarity. Their real problem wasn’t logistics. It was forecasting.

2. Collecting the Right Data

Alice next asked for all the historical sales information. What she received was… chaos. Spreadsheets from different years, formats, and file types; some came from POS systems, others from suppliers.

She gathered everything: daily sales, stock levels, weather logs, holiday calendars, discount history, pulling them all into one place.

3. Cleaning and Preparing the Data

The deeper she looked, the messier it got. Zeros that were actually missing values. Duplicate entries from system glitches. Strange outliers, like 400 avocados sold in one hour.

Alice cleaned patiently, row by row, fixing dates, removing duplicates, smoothing spikes, and filling gaps. The data finally felt trustworthy.

4. Exploring the Data (EDA)

Now the fun part began.

Plots. Correlations. Seasonality curves. Weekly patterns.

That’s when she saw it:

  • Berries sold more when temperatures were above 22°C
  • Avocados spiked before long weekends
  • Leafy greens dropped sharply when competitors ran promotions

Alice wasn’t just analysing numbers. She was discovering the hidden behaviour of customers.

5. Engineering Useful Features

She realized raw sales numbers weren’t enough. So she enriched the dataset:

  • “Days until holiday”
  • “Weekend indicator”
  • “3-day moving average”
  • “Temperature change”
  • “Discount level flag”

Each new feature helped the model understand demand the way a human would.

6. Splitting the Data for Training and Testing

Before choosing any model, Alice needed to make sure she wasn’t fooling herself. A model can look perfect if it is tested on the same data it learned from. The real challenge is whether it can predict what it has never seen.

So she divided the dataset carefully.

She trained the model on several years of historical data and reserved the most recent months as a “future simulation.” The model would never be allowed to see these final months during training.

This mirrored the real world: the model learns from the past and must forecast the future.

By splitting the data this way, Alice could see whether her system genuinely understood demand patterns or was simply memorising historical values. If it could accurately predict the unseen period: holidays, surprises, slow weeks, sudden spikes, then she knew it was ready for proper modelling.

With the split complete, she finally had a trustworthy setup: one part for learning, one part for testing, and a realistic way to judge if any model could stand up to real-life volatility.

7. Trying Different Models

Alice didn’t know which approach would work best. So she tried several:

  • Linear regression
  • Random Forest
  • XGBoost
  • A few neural networks

Each had strengths; each had blind spots. She compared accuracy, looked at errors, and watched how they handled extreme cases.

8. Training the Best Candidate

She selected the model that performed the most consistently. Not just on good days, but on messy, unpredictable ones. Then she trained it properly:

  • Good parameter tuning
  • Correct validation
  • Preventing overfitting
  • Making sure it learned patterns, not noise

The model began predicting demand with surprisingly sharp intuition.

9. Evaluating Its Real-World Performance

Alice wasn’t satisfied with “high accuracy.” She tested the model on completely unseen weeks:

Hot weeks. Cold weeks. Holiday rushes. Slow seasons.

She measured how far predictions deviated, which products remained tricky, and whether the model could anticipate sudden spikes.

She improved it until it passed every stress test.

10. Deploying the Model

Now came the moment of truth. Alice integrated the model into the purchasing workflow.

A simple dashboard displayed recommended order quantities for each product every morning. Store managers didn’t need to understand machine learning. They just needed reliable numbers.

Marta, the senior buyer, tried it first.

“Only 32 boxes of avocados this week?” Alice nodded. “Trust it.”

11. Monitoring the System

Over the next weeks, Alice watched the predictions closely. When a heatwave hit unexpectedly, the model slightly underpredicted berry sales. She analysed the cause, added new features, retrained the model, and improved its resilience.

The system grew sharper with each observation.

12. Continuous Feedback and Improvement

Store managers gave feedback:

  • “We always sell more during Friday evening rush”
  • “Tomatoes rise aggressively when local farms run promotions”
  • “Rain kills foot traffic”

Alice incorporated these insights, making the model more aligned with real operations.

GreenGrocer’s waste dropped. Stockouts disappeared. Revenue rose. And the stores finally stopped having “Worry Meetings.”

What Alice built for a local supermarket, organisations across finance, operations, and compliance are building at scale. Industry data shows that companies implementing ML-based invoice processing correctly achieve straight-through processing rates above 88%. Those running ML on fraud detection report accuracy rates above 90%. And businesses using ML for cash flow forecasting have cut late payments in half while saving hundreds of thousands annually in unnecessary financing costs. The methodology behind those results is the same 12 steps Alice followed. The only difference is the size of the problem being solved.

How Kibo AI Implements the Full 12-Step Machine Learning Lifecycle in Real Organisations

12 Steps to Build a Machine Learning Model]

1. Business Understanding

At Kibo AI, every ML engagement begins with alignment. Starting with the wrong problem is the most expensive mistake any organisation can make. This first step resembles choosing a destination before starting a journey. Many teams want to jump straight into the data or model selection, but without a shared objective, they risk delivering something technically impressive that brings no business value.

Business leaders contribute context: financial pressure points, delays, inefficiencies, missed opportunities. Data teams contribute structure: the ability to convert broad frustrations into measurable objectives. Together, they define what success looks like and how it will be measured.

Forecasting is a clear example. In many companies, regional teams submit Excel files filled with assumptions and last-minute adjustments. Finance spends weeks merging versions, correcting errors, and negotiating inconsistencies. When executives ask why forecasts fluctuate, the explanations are uncomfortable and unclear.

Kibo AI facilitates a structured conversation between Finance and data teams to define the real challenge: forecasts are slow, inconsistent, and unreliable. The objective becomes precise: “Predict monthly revenue for each region within ±5% using historical sales and key business drivers.” With this clarity, the entire ML initiative gains direction and purpose.

2. Data Collection

Once the objective is clear, Kibo AI gathers the raw ingredients. ML cannot deliver results if critical data is missing, unreliable, or segmented across silos. This step is investigative: identifying where data lives, who owns it, what format it is stored in, and how complete it is.

In real-world revenue forecasting programs, Kibo AI often sees data scattered across CRM systems, financial tools, marketing platforms, and external economic sources. The goal is to connect everything into a unified dataset.

As the unified dataset comes together, common problems surface: missing months from certain regions, mismatched currency conversions, inconsistent campaign data. With patience and structure, Kibo AI consolidates all of this into a single source of truth that updates automatically. For many organisations, this is the first time all regions finally look at the same numbers.

3. Exploratory Data Analysis (EDA)

This is where the data begins to speak. Kibo AI visualises trends, seasonality, spikes, dips, and correlations, turning raw data into shared understanding.

Clear seasonal patterns appear. December shows strong revenue surges. Summer dips become evident. Marketing spend reveals a two-month lag before results appear. Regional anomalies prompt discussions: Finance reveals that a sudden drop wasn’t a demand issue but a supply chain disruption.

EDA turns charts into understanding. By the end of this phase, everyone sees the business through the lens of the data. A foundation that ensures the model reflects real behaviour rather than assumptions.

4. Data Cleaning and Preprocessing

Raw data is never clean. If left untreated, inconsistencies will damage the model’s accuracy. This stage is where disorder becomes structure.

As the dataset comes together, Kibo AI uncovers predictable issues: gaps in reporting, inconsistent formats, outdated conversions, one-time promotions that distort trends. Each problem is resolved systematically until the dataset becomes trustworthy and aligned.

Once the cleaning is complete, the dataset finally reflects how the business actually behaved: consistent, complete, and ready for modelling.

5. Feature Engineering

This is where raw data turns into intelligence. Kibo AI collaborates closely with Finance to identify what truly drives revenue. Instead of feeding the model raw inputs, the team designs features that mirror how an experienced analyst thinks.

Rolling averages capture trend momentum. Ratios reflect marketing effectiveness. Calendar flags highlight seasonal peaks. Inflation adjustments distinguish real demand from price-driven increases.

Each new feature gives the model a richer, more human-like understanding of the business. This stage is often the difference between a mediocre model and one that truly delivers value.

6. Data Splitting

Before training the model, the data must be divided carefully. Kibo AI ensures that training and testing reflect real-world conditions. The model trains on historical data and must predict the most recent quarter: information it has never seen.

If it can accurately forecast Q4 using only data available before Q4 began, confidence in the model grows. This realistic evaluation prevents the system from memorising history instead of learning genuine patterns.

7. Model Selection

Model selection is about choosing the right tool for the problem. Kibo AI evaluates several models, considering accuracy, interpretability, scalability, and how easily results can be explained to leadership.

ARIMA understands seasonality but misses marketing effects. XGBoost captures complex interactions but needs explainability tools for stakeholder trust. A hybrid approach often combines the strengths of both.

With SHAP, Kibo AI shows Finance exactly which factors drive predictions: marketing spend, inflation, trend changes, building transparency and credibility. The chosen model becomes the core of the forecasting solution.

8. Model Training

The model now begins to learn. It adjusts itself as it observes patterns across years of data. Kibo AI monitors each iteration, ensuring the system learns genuinely useful relationships instead of memorising noise.

The model gradually stabilises, showing consistent performance across unseen validation data. Finance reviews early predictions and begins to understand the model’s logic, building trust ahead of formal evaluation.

9. Model Evaluation

Evaluation is the truth test. After training, the model forecasts the most recent quarter using data it has never encountered. Kibo AI analyses where predictions deviate and works through the reasoning with business stakeholders. In financial operations, this is where the gap between manual forecasting and ML becomes concrete.

Manual cash flow forecasts built on fragmented spreadsheet data typically miss liquidity risks by weeks. ML-based models that are properly trained and evaluated have demonstrated accuracy rates of 90-95% in comparable programmes. FP&A confirms whether the model reflects business reality. Once it hits the accuracy target, deployment begins.

10. Model Deployment

Deployment is where the model becomes a real business tool. Kibo AI packages the forecasting system as a secure cloud service. New data triggers automatic predictions that flow into dashboards and reports without manual intervention.

Controllers open their usual Power BI workspace and see updated forecasts instantly: no spreadsheets, no versioning issues, no delays. The forecasting process becomes faster, more consistent, and always up to date.

11. Monitoring and Maintenance

Business conditions evolve, and models must evolve with them. Kibo AI monitors the model’s performance continuously. If accuracy drifts, the system alerts the team.

In one region, forecasts begin deviating because a new product launched: something the model couldn’t have learned earlier. Kibo AI retrains the model with updated information, restoring performance immediately.

Monitoring ensures the system stays aligned with reality.

12. Feedback and Continuous Improvement

The final stage turns ML into a living asset. Each month, users provide context: delayed campaigns, regulatory changes, unexpected disruptions. Kibo AI feeds this knowledge back into the model during retraining cycles.

Over time, forecasts become stronger, more stable, and more aligned with actual business behaviour. The model grows with the organisation, not away from it.

Conclusion: How Kibo AI Turns Machine Learning Into Real Business Value

The companies that structure their ML implementation correctly are already seeing where the gains come from. Finance teams processing invoices in hours instead of days. Compliance teams catching duplicate payments and fraud patterns that rule-based systems miss entirely. Treasury functions forecasting cash positions with enough accuracy to cut unnecessary financing costs.

None of this requires a data science department or a multi-year transformation programme. It requires a clear problem, the right data, and a structured path from raw information to a working prediction system. That is what the 12 steps in this article describe. And it is what Kibo AI delivers.

 

Ready to implement machine learning in your organisation? Let’s talk

Contact us

Frequently Asked Questions

How long does a machine learning implementation take in a business?

A typical ML implementation from business understanding to first deployment takes between 8 and 16 weeks, depending on data availability, complexity, and organisational readiness. Monitoring and improvement continue after deployment.

Most organisations already have enough data to begin. What matters more than volume is relevance and quality. Historical sales records, CRM data, financial results, and operational logs are common starting points. A structured data audit is usually the first step.

Traditional analytics describes what happened in the past. Machine learning goes further. It identifies patterns in historical data and uses them to forecast future outcomes or automate decisions. ML adapts as new data comes in, while traditional analytics requires manual updates.

Not necessarily. With the right external partner and a well-defined business problem, even mid-sized organisations can implement and benefit from ML. The key is having a structured methodology and close collaboration between business and data teams throughout the process.

Success is measured against the business objective defined at the start. For example: forecasting revenue within ±5% accuracy, reducing stockouts by 30%, or cutting manual reporting time in half. Technical metrics like RMSE or MAPE support these business KPIs but are not sufficient on their own.

Machine learning delivers value across industries wherever there is historical data and a need for better predictions or decisions. Common use cases include revenue forecasting in Finance, demand planning in Retail and Supply Chain, churn prediction in SaaS, risk scoring in Financial Services, and resource allocation in Operations.

Rusi Yang

Consulting Manager, Kibo AI

Linkedin

Rusi leads AI strategy and digital transformation initiatives across finance, automotive, manufacturing, consumer goods, pharmaceuticals, and entertainment.​

Rusi has shaped enterprise roadmaps and guided cross-functional teams in delivering automation, data science, and AI programs that generate measurable business value.​

With 6+ years of international experience, he partners with executives to align technology with business objectives and drive sustainable, innovation-led change.​

Expert in AI & ML, Data Science, Intelligent Document Processing & Google Cloud.

]]> EU AI Act Compliance Guide for Business Leaders https://kibo-ai.com/insights/articles/eu-ai-act-compliance-guide/ Wed, 20 May 2026 12:13:19 +0000 https://kibo-ai.com/?p=5029

EU AI Act Compliance Guide for Business Leaders

Markus Schaumberger (Managing Director D/A/CH)

The EU AI Act applies a risk-based approach. What’s prohibited, what’s high-risk, and what’s simply a transparency requirement: these distinctions exist.

The EU AI Act Is Already in Force. Most Companies Are Not Ready. Here Is What That Means for You.

Most organisations I speak with have two reactions when the EU AI Act comes up. The first is “We know it is coming. We will deal with it properly before August 2026.” The second, once they realise parts of it are already in force, is something closer to: “So what do we actually do now?”

That second reaction is the more honest one. And it is the one I hear far more often than anyone admits publicly.

The problem is that parts of it are not coming. They arrived on 2 February 2025. The AI literacy obligation under Article 4 is already law. If your employees use ChatGPT, Microsoft Copilot, or any AI tool as part of their work, and you have not put a documented training or guidance programme in place, you are already in breach.

According to the IAPP EU Digital Laws Report 2025, nearly 7 in 10 businesses report difficulty understanding their obligations under the EU AI Act. That number is not a prediction about who might fall behind. It describes where I see most organisations sitting right now: aware that something is required, uncertain about what exactly, and unsure where to begin.

This article covers what the Act actually requires, which deadlines are real, and what getting ready looks like in practice.

Five things you need to know right now

If you only have two minutes, start here. The rest of the article gives you the context behind each point.

1. The AI literacy obligation (Article 4) has been in force since 2 February 2025. It is not upcoming. It applies now.

2. The regulation applies to any organisation using AI in its business processes. You do not need to build AI to be in scope.

3. Penalties reach up to 35 million EUR or 7% of global annual turnover for the most serious violations, and up to 15 million EUR or 3% for most deployer-level non-compliance.

4. Most organisations have four gaps right now: no AI inventory, no defined governance owner, no documentation structure, and no AI literacy programme for staff.

5. The window to August 2026 is shorter than most leadership calendars currently reflect.

Not sure where your organisation stands? We walk business leaders through the practical implications in a free 45-minute session every other Thursday.
Free EU-AI Act Webinar

In this article

What the EU AI Act Actually Is

The EU AI Act (Regulation (EU) 2024/1689) is the European Union’s primary legal framework for artificial intelligence. It uses a risk-based approach: the higher the potential harm of an AI system, the stricter the requirements.

Here is the part that surprises most organisations. This regulation does not primarily target the companies building AI. It targets deployers: the organisations using AI in their business processes.

That is your HR team running CV screening software. Your customer service department using a chatbot. Your finance function relying on AI-assisted decision support. Your procurement team using a SaaS platform with embedded AI features they may not even know are there.

If your organisation uses AI in any of these areas, you are in scope:
  • Human resources and recruitment (CV screening, performance evaluation, workforce planning)
  • Customer interactions (chatbots, virtual assistants, automated eligibility checks)
  • Decision support (credit, risk, compliance, procurement)
  • Critical operations (logistics, safety, infrastructure)
  • Content generation (copilots, document assistants, synthetic outputs)

The Dates That Actually Matter

The timeline is not one deadline. It is a series of obligations that started at different points. Several have already begun.

Last updated: May 2026, reflecting the Digital Omnibus provisional agreement of 7 May 2026.

1 August 2024: The Act entered into force across all 27 EU Member States.

2 February 2025: Banned AI practices and the AI literacy obligation under Article 4 became applicable. Both are legally enforceable right now.

2 August 2026: Transparency rules under Article 50 apply from this date. Chatbots and AI tools that interact with people must clearly identify themselves as AI. National enforcement begins on this date.

2 December 2026: Providers of generative AI systems already on the market before August 2026 must have their content marking solutions in place. A new prohibition on AI systems generating non-consensual sexual imagery or child sexual abuse material also takes effect.

2 December 2027: The strictest obligations apply for high-risk AI systems. This covers AI used in hiring, education, credit scoring, law enforcement, and critical infrastructure. The original deadline was August 2026 and was extended following the Digital Omnibus agreement of 7 May 2026.

2 August 2028: Rules for high-risk AI built into regulated products such as medical devices, machinery, and toys apply from this date.

On 7 May 2026, the EU Council and European Parliament reached a provisional political agreement on the Digital Omnibus on AI. The agreement still requires formal adoption before it becomes law, but formal endorsement is expected before August 2026. For planning purposes, December 2027 is now the operative baseline for high-risk obligations.

Two things the Omnibus did not change. Article 5 (prohibited practices) remains fully in force. Article 4 (AI literacy) also remains in force from 2 February 2025. The Omnibus adjusts how that obligation is framed, moving it closer to an obligation of means rather than strict result, with the Commission and Member States providing supporting guidance. Organisations deploying high-risk AI systems still carry a direct training obligation for staff responsible for human oversight. When enforcement begins in August 2026, regulators will look for evidence of a proportionate, documented approach regardless of risk tier.

The most common misconception I encounter is this: organisations believe they have a full two-year grace period until August 2026. The February 2025 obligations are not a warm-up round. They are legally enforceable right now.

The Four Risk Categories Every Deployer Must Understand

A deployer is any organisation that uses an AI system built by someone else. If your teams use Microsoft Copilot, ChatGPT, or any AI-powered tool as part of their work, your organisation is a deployer under the EU AI Act.

The Act classifies AI systems into four categories. Your obligations depend on which category your specific use cases fall into, which is why mapping them is the essential first step.

Prohibited AI covers practices that are banned outright and cannot be used under any circumstances. These include social scoring systems, real-time biometric surveillance in publicly accessible spaces for law enforcement purposes, AI systems that manipulate people through techniques they are not aware of, and AI that targets the psychological vulnerabilities of specific groups. A further prohibition on AI systems generating non-consensual sexual imagery or child sexual abuse material is expected to be formally adopted before August 2026, with a compliance deadline of 2 December 2026. The existing prohibitions have been enforceable since February 2025.

High-risk AI carries the strictest obligations. This category covers AI used in hiring decisions, CV screening, performance evaluation, credit scoring, access to essential services, law enforcement, and critical infrastructure. If your organisation uses AI in any of these areas, the compliance requirements are significant: human oversight must be in place, fundamental rights impact assessments must be conducted, and system performance must be monitored throughout use. The planning date for these obligations is 2 December 2027, following the Digital Omnibus provisional agreement.

Transparency-required AI covers systems that interact with people directly or generate synthetic content. From 2 August 2026, any chatbot or AI copilot must clearly tell users they are interacting with AI. If your organisation produces deepfakes or AI-generated content, that content must be labelled as artificially created. Most commercial GenAI tools fall into this category.

Minimal-risk AI covers the largest share of AI use cases and carries lighter obligations. One requirement still applies here regardless: the AI literacy obligation under Article 4, which has been in force since February 2025.

The first step is knowing which of your AI use cases fall into which category. A study by appliedAI covering 106 enterprise AI systems found that 40% could not be clearly classified. Most organisations have not done this mapping yet.

What the Penalties Look Like

The EU AI Act uses a three-tier penalty structure under Article 99.

Tier 1: Fines of up to 35 million EUR or 7% of global annual turnover and applies to violations of the prohibited AI practices under Article 5. These prohibitions have been enforceable since February 2025.

Tier 2: Fines of up to 15 million EUR or 3% of global annual turnover and applies to non-compliance with deployer obligations under Article 26, transparency requirements under Article 50, and a range of other provisions. This is the tier most private-sector organisations need to focus on.

Tier 3: Fines of up to 7.5 million EUR or 1% of global annual turnover and applies to supplying incorrect or misleading information to competent authorities.

For larger enterprises, the higher of the two amounts applies. For SMEs and startups, the lower applies. For context: GDPR’s maximum sits at 4% of global annual turnover. The AI Act’s top tier is 7%. National authorities begin formal enforcement from 2 August 2026.

In my experience, the penalty numbers land differently on different executives. The CFO does the maths on 3% of turnover. The General Counsel thinks about documentation gaps. The CEO thinks about how a regulator inquiry looks in the press. They are all right to be concerned. The question is not whether the exposure is real. It is whether you have started addressing it.

Why Most Organisations Are Not Ready

In working with enterprises across the DACH region, the same four gaps appear consistently. They are not unique to any industry or company size.

The invisible AI problem. Nobody has a complete picture of where AI is actually being used inside their organisation. It is not just the tools IT approved. It is the SaaS platforms with AI features quietly embedded into products your teams have used for years, the vendor systems processing customer data, and the tools individual employees started using on their own last quarter.
One of our insurance clients began a compliance check expecting a handful of relevant systems. The inventory came back with over thirty. None of them were surprises to the individual teams using them. All of them were surprises to legal, risk, and compliance.

The ownership gap. The EU AI Act touches governance, procurement, HR, IT, security, risk, compliance, and every business function using AI tools. When a question like “are we compliant with the transparency requirements for our customer chatbot?” arrives on someone’s desk, there is often no single person or function who owns the answer. Everyone has a partial view. Nobody has the complete one.

The documentation deficit. High-risk AI use cases require evidence: logs, risk assessments, human oversight records, incident processes. Even organisations that have thought seriously about compliance often have no structured approach to creating or retaining this. The IAPP puts it plainly from their readiness work with deployers: the gap is rarely legal theory. It is missing operational proof. If regulators ask, the answer cannot be “we handled it appropriately.” The answer needs to be a file.

The literacy gap. Article 4 has been applicable since 2 February 2025. The majority of organisations have not yet run any structured AI literacy programme for their employees. Every employee using an AI tool as part of their job is part of the compliance picture. For most companies, that is the majority of their workforce.

If any of these four gaps sound familiar, our free EU AI Act webinar is the right next step. In 45 minutes, we walk through how to run the AI inventory that most organisations have been putting off for months, and what the first 30 days of a compliance programme actually looks like in practice. Next session: every other Thursday.

What Getting Ready Actually Looks Like

Compliance under the EU AI Act is an operational programme, not a legal opinion you commission and file. It requires execution across multiple functions simultaneously, and it takes time to build properly.



When I work through this with leadership teams, I always start in the same place. Not governance frameworks or legal analysis. Five concrete operational areas that everything else builds on.

 

1.Start with the inventory. Not a rough list your IT team pulls together in an afternoon. A structured mapping of every AI system, tool, and vendor in active use, including the AI embedded in SaaS platforms your teams have been using for two years without anyone formally logging it.



2. Classify each use case against the EU AI Act’s risk categories.



3. Assign ownership: one person or function with clear authority for AI governance decisions, not a committee where everyone has a view and nobody has accountability.



4. Build human oversight mechanisms for any AI system supporting important decisions.



5. Create documentation that will hold up if a regulator asks. The logs, records, and processes that demonstrate compliance when it counts.



None of this requires slowing down AI adoption. The organisations best positioned by August 2026 are those building governance alongside their AI work, not as a separate track that starts later. GDPR taught that lesson clearly: late adopters faced last-minute panic and paid for it.

What We See in Practice

One pattern we see consistently: organisations that treat compliance as a separate legal workstream end up with two parallel efforts that never properly connect. The ones that build governance into their operating model from the start move faster.

One insurance client approached us after their legal team flagged the EU AI Act as a material risk. Their assumption was that exposure would be limited to a handful of AI tools they knew about. When we conducted the use case inventory, the picture looked different.

AI and automation were already embedded across claims processing, customer service routing, underwriting support, and several internal workflows. None of it was rogue. All of it had been introduced thoughtfully by individual teams solving real problems. But none of it had been mapped against the Act’s requirements, classified by risk level, or documented with appropriate governance evidence.

The compliance check produced a risk heatmap, an initial use case classification, and a prioritised 30/60/90-day action plan. The conversation shifted from “we need to think about this” to “here is what we are doing and who owns it.”

The EU AI Act does not require perfection. It requires evidence of a considered, proportionate, and documented approach. That is achievable. But the window to August 2026 is shorter than most executive calendars currently reflect.

Three Questions for Your Next Leadership Meeting

Bring these three questions to your next leadership discussion. The answers will tell you where you actually stand.

  1. Can we produce a complete list of every AI system, tool, and vendor in use across the business today?
    Not a rough estimate. A complete list. If the answer is no, or “probably not,” you have an AI inventory gap at the foundation of every other compliance requirement. You cannot classify risk you have not found.

  2. Do we have a defined owner for AI governance and compliance decisions?
    One person or function with clear authority and accountability. If ownership is distributed across legal, IT, risk, and business with no central coordination, every decision will take longer than it needs to and evidence will end up in the wrong places.

  3. Have our employees received any structured guidance on responsible AI use since 2 February 2025?
    Article 4 has been in force since that date. If the answer is no, that is not a future risk. It is a compliance gap that exists right now.

Where to Go From Here

If your organisation is early in this process, the most useful next step is a clear picture of what the Act means for your specific situation, mapped to your actual AI use cases, business functions, and risk profile.

We run a free 45-minute session every other Thursday for business leaders working through exactly this. No technical or legal background required. The session covers the obligations that apply to most deployers, the gaps we see consistently, and how to build a compliance approach that does not get in the way of the AI work your teams are already doing.
Join the next EU AI Act webinar
Ready to move? Two options.

We train your teams, including a dedicated EU AI Act programme, so compliance is handled from within. Or we handle it for you entirely as an external partner: use case inventory, risk classification, governance setup. You focus on the business. We produce the deliverables.
See the Kibo AI Trainings
Ask for individual help

Frequently Asked Questions

Does the EU AI Act apply to my organisation if we only use third-party AI tools?

Yes. The Act applies to deployers, meaning organisations that use AI systems in their business processes, regardless of whether those systems were built internally or licensed from a vendor. If your teams use Microsoft Copilot, ChatGPT, Salesforce Einstein, or any AI-enabled SaaS tool in ways that affect employees, customers, or business decisions, you have deployer obligations under the Act.

The Act entered into force on 1 August 2024. The first obligations, covering banned AI practices and the AI literacy requirement under Article 4, became applicable on 2 February 2025. Transparency rules (Article 50) for new AI systems apply from 2 August 2026, which is also when national enforcement begins. Following the Digital Omnibus provisional agreement of 7 May 2026, high-risk AI obligations under Annex III now apply from 2 December 2027, and high-risk AI embedded in regulated products from 2 August 2028, pending formal adoption. Article 5 is not affected by the Omnibus. Article 4 remains in force from 2 February 2025, with the Omnibus adjusting its character toward an obligation of means, supported by Commission and Member State guidance.

Article 4 requires every organisation deploying AI to take measures, to the best of their extent, to ensure that relevant staff have a sufficient level of AI literacy. This applies to any employee using AI tools as part of their work. It became applicable on 2 February 2025 and covers all AI systems regardless of risk category. The Digital Omnibus provisional agreement adjusts the character of this obligation — moving it closer to an obligation of means than strict result — and tasks the Commission and Member States with providing supporting guidance and resources. Organisations deploying high-risk AI systems retain a direct training obligation for staff responsible for human oversight. For all deployers, the practical expectation remains: a documented and proportionate approach to AI literacy, ready to demonstrate if asked.

The EU AI Act uses a three-tier structure under Article 99. Violations of prohibited AI practices carry fines of up to 35 million EUR or 7% of global annual turnover, whichever is greater. Non-compliance with deployer obligations and transparency requirements carries fines of up to 15 million EUR or 3% of global annual turnover, whichever is greater. Providing incorrect information to authorities carries fines of up to 7.5 million EUR or 1% of global annual turnover, whichever is greater. Reduced caps apply for SMEs and startups, where the lower of the two amounts applies.

Start with an AI inventory: a structured mapping of which AI systems, tools, and vendors are in use across your organisation. Without that foundation, risk classification and governance cannot be built accurately. From there, a risk classification exercise maps your use cases to the relevant EU AI Act categories and identifies where the most material obligations sit. A structured diagnostic typically completes this in 2-3 weeks and produces a prioritised action plan.

No. The AI literacy obligation under Article 4 applies regardless of risk category and has been in force since February 2025. Transparency requirements apply to chatbots, copilots, and synthetic content tools from August 2026. The stricter documentation and oversight obligations apply to high-risk systems, but even organisations with only minimal-risk AI use are within scope of the regulation.

Based on conversations with business leaders across the DACH region, the most common one is this: we have time. First it was “we have until August 2026.” Now, with the Digital Omnibus provisional agreement shifting high-risk obligations to December 2027, I am already hearing “well, we have even more time now.” Both miss the point. The AI literacy obligation under Article 4 has been in force since February 2025, and the prohibitions on certain AI practices have been enforceable since then too. The Omnibus did not remove Article 4 — it adjusted how the obligation is framed and who supports implementation. Organisations that have not addressed it were already behind before the Omnibus. The delay on high-risk obligations does not change that.

Markus Schaumberger brings 25+ years of experience in enterprise consulting and has built high-performing teams while leading major digital transformation programs.

Markus Schaumberger

Managing Director D/A/CH, Kibo AI

Linkedin

Markus Schaumberger is Managing Director D/A/CH at Kibo AI, based in Bavaria. He leads enterprise client relationships across Germany, Austria and Switzerland, with a focus on AI strategy, governance, and implementation for large organisations. Before joining Kibo AI, he spent 25+ years in enterprise consulting, leading major digital transformation programmes across the DACH region.

The team behind Kibo AI has delivered 4,000+ AI and automation projects for 2,000+ clients across EMEA, the US, and APAC, including work alongside McKinsey on AI design programmes for major European insurers. Kibo provides practical governance and readiness support and does not provide formal legal advice.

EU AI Act specialist, advising deployer organisations on governance strategy and practical compliance

Sources

]]> The Kibo Forecast, 2026 https://kibo-ai.com/insights/articles/the-kibo-forecast-2026/ Wed, 28 Jan 2026 23:22:58 +0000 https://kibo-ai.com/?p=3714

What will 2026 mean for enterprise AI?

2026 is the year AI transformation strategies move from planning into production. After an intense phase of experimentation, the focus is shifting to governance, skills development, and real outcomes. The conversations that used to start with “What is AI?” are now starting with “How do we co-create with it?”

2025 was a big year in the world of artificial intelligence – from agentic adoption to ambitious AI-generated commercials, to Time Magazine naming the Architects of AI its coveted “Person of the Year” status. and this year we’re sure to keep on building. With more and more businesses taking the leap into AI-supported workflows, we expect to see a massive surge in ambition – and a bold and exciting new reality to match. 

So what does the Kibo AI team think 2026 has in store? Our head of engineering Ignasi Peiris has some ideas! 

From strategy to reality

2026 will be the year where companies will finally kick off, and turn their AI Transformation Strategies into a reality. And with new methodologies, they’ll be able to merge their ambition with real and measurable outcomes.

After an intense phase of experimentation, businesses are now preparing to embed AI into their workflows, creating turbo-boosted processes that cross all previous limitations. This means they can focus their efforts on driving real impact, and finding unique, AI-powered ways to manage their strategic goals.

Leaving hype behind

Enterprises are essentially shifting focus from hype to practical value – focusing on training, governance, ROI evaluation, and strategic deployment rather than ideas and drafts.

We have seen that the AI race is way faster than we predicted. New models and products are released every day, at unprecedented speed. And with each new update, what used to be “too hard to tackle” has evolved into “what if we do it this way?” This has created sandboxes for teams with bold and creative mindsets to turn ambitious concepts into reality. This transition includes not just generative capabilities but also agent-oriented AI systems that can orchestrate workflows, plan multi-step actions, and operate semi-autonomously, quickly changing how organizations scale work and make decisions.

The New Normal.

In 2026, we can stop asking “What is AI?” or “Why should everyone use it?” This is now the “new normal”. Our conversations will shift to co-creation, working side by side with clients and customers to imagine how AI can be used in the most creative, useful and efficient ways – while also focusing on turning those agreed outputs into a reality.

We’ll jointly turn question marks into exclamations, and what used to be the end of a project will instead open up new possibilities, with a tremendous amount of potential applications. AI will help companies explore a different range of limits – limits that will depend only on the creativity behind each product.

In short…

2026 will be the year of smart design, strategy and planning, investment in governance and skills development, and a focus on outcomes over outputs.

And what do you think? Is 2026 set to be the Year of AI? How are you planning to integrate it into your life and work? 

Whatever your plans for the year ahead, Kibo AI is here to help! Reach out to the team below for insights, advice, and offers! 

Contact us

Questions About AI in 2026

What will 2026 mean for enterprise AI adoption?

2026 is the year where AI transformation strategies move from planning documents into working systems. After an intense phase of experimentation, businesses are embedding AI into daily workflows with a focus on measurable outcomes. The question is no longer whether to adopt AI, but how to govern, train for, and scale it across the organization.

It means shifting budget and attention from pilots and demos toward training, governance, ROI evaluation, and strategic deployment. Enterprises in 2026 are not experimenting with AI, they are running it. New models and products are released at a pace that has made previously complex challenges approachable, and organizations with structured processes are pulling ahead.

Agent-oriented AI systems go beyond generating text or images. They orchestrate workflows, plan and execute multi-step actions, and operate semi-autonomously within defined boundaries. This allows organizations to scale work and make decisions differently, since an agent can take a business objective, break it into steps, and execute them across different systems without constant human input.

]]> From cost center to value engine –  how AI is redefining the IT spend https://kibo-ai.com/insights/articles/from-cost-center-to-value-engine-how-ai-is-redefining-the-it-spend/ Wed, 21 Jan 2026 21:29:44 +0000 https://kibo-ai.com/?p=3677

How does AI turn IT from a cost center into a value engine?

Corporate IT budgets have long been treated as operational costs with little clear validation. AI changes this by enabling IT teams to shift from managing infrastructure toward proving measurable business value. Through agentic workflows that self-optimize and automate multi-step tasks, the same IT staff can create significantly more output without adding headcount.

The future of industry is in smarter IT expenditures. So how can businesses shift their current thinking, and get ahead of the competition?

The cost of ambiguity

While corporate IT budgets have grown over the years, they reflect an operational cost which, in comparison to capital expenditures, is less clearly understood, and for which the specific value is hidden. In other words, seen as a mere necessity, IT is a cost center with little obvious validation.

But with artificial intelligence on the rise, IT spending is being reframed – from cost to value. And with this comes an essential question for every business leader: “What am I getting for this investment?”

The time has come for company IT strategies to be reoriented – away from simply managing infrastructure, and towards proving real business value through AI-enabled impact.

The inefficiency gap

For most businesses, IT is not a core function. Neither customer-facing nor revenue generating, IT is a support function like HR and Operations, enabling the core purpose of the business – insurance, production, healthcare – from the background.

But even with the rise of scalability products such as playbooks, knowledge management systems and managed service providers, IT is still a traditional, human-centric and inefficient function, too reliant on legacy processes. Data is siloed; workflows are manual and slow; and capital is often tied up with the least measurable return. Robotic Process Automation (RPA) and process mining can create some efficiency, but AI – especially agentic workflows – can use the same IT staff to vastly increase the value created.

AI: the catalyst for value acquisition

The leap to agentic workflows allows companies to adjust rigid business workflows into processes that self-optimize – where AI agents execute multi-step tasks, learn from outcomes, and seek the most efficient path. One example might be replacing manual ticket-handling with an AI-driven service desk agent – streamlining the entire workflow, and turning this and other such laborious tasks into smooth, automated processes. 

AI enables the strategic principle of achieving “more with the same or more with less” across all IT and business functions. Allowing IT staff to reach more outcomes quicker both improves internal customer satisfaction, and expedites the business’ core functions.

Where to apply AI can be difficult, but identifying inefficiencies is a good starting point. Generally, most businesses know where the bottlenecks are in their IT operations – having built up over the years through reorganizations, acquisitions, regulatory components, etc. AI gives these businesses the opportunity to re-imagine their workflows, and allow agents to identify better ways to perform.

The value equation: metrics for modern IT

The modern mandate for IT is to ensure accountability and responsibility for all the money spent on measurable ROI. This covers three areas:

  • Efficiency – operating with a reduced and measurable spend
  • Revenue generation – creating new opportunities for the business core function
  • Risk reduction – eliminating both internal and external risks that can disrupt the business  

 

Practically this could mean anything from IT automation, with predictive maintenance and usage monitoring; operations, and processing efficiency gains through hyper-automation; the supply chain, integrating predictive logistics and inventory optimization; to HR, with AI-powered talent acquisition and personalized employee support, among many other potential applications.

Overall, global IT spend continues to rise as companies invest more in differentiating their businesses. The true winners will be those that shift from volume of spend to quality of ROI.

The future of IT, and serving the business

The age of technology for technology’s sake – or just keeping up with the competition via new software and infrastructure – is over. The race to the cloud for companies that did not properly evaluate their ROI has been a business awakening. AI has the opportunity to change to a value-based evaluation of worth. Most companies are still in early prototype stages and are struggling to identify ROI amongst fancy interfaces and novel ways of interacting with data. Those that use this as an opportunity to re-align to serve their core business function will be the future leaders in their industry. Maintaining a focus on metrics allows for intelligent investment that returns greater benefits. And through the lens of AI-driven value and ROI, organizations can move beyond wasteful expenditure and turn their technology budget into their most powerful competitive advantage.

How we can help

At Kibo AI, we provide essential advisory services for keeping your business on top – from how to spend wisely, to which AI tools work best for which task. Get in touch with our experts, and find out what we can do for you.

Questions About AI and IT Spending

How does AI change the role of IT in a company?

IT is a support function, neither customer-facing nor revenue generating. AI changes this by enabling IT teams to produce more with the same headcount through agentic workflows that self-optimize and execute multi-step tasks. The result is a shift from managing infrastructure to proving real business value.

The modern mandate for IT covers three areas: efficiency, which means operating with a reduced and measurable spend; revenue generation, which means creating new opportunities for the core business function; and risk reduction, which means eliminating both internal and external risks that can disrupt the business.

An agentic workflow uses AI agents to execute multi-step tasks, learn from outcomes, and seek the most efficient path through a process. A practical example is replacing manual IT ticket-handling with an AI-driven service desk agent, running the same workflow with the same IT staff but at significantly greater scale.

Identifying existing inefficiencies is the most reliable starting point. Most businesses already know where their IT operations slow down, whether from legacy processes, manual workflows, or siloed data built up through years of reorganizations, acquisitions, and regulatory changes. These bottlenecks are the highest-value targets for AI.

]]> Do you need more AI – or just more clarity? https://kibo-ai.com/insights/articles/do-you-need-more-ai-or-just-more-clarity/ Thu, 30 Oct 2025 21:56:59 +0000 https://kibo.opengroup.hu/?p=1

Do companies need more AI, or more clarity?

The exciting potential of AI can come with a loss of focus. Before implementing any solution, business leaders need to define what intelligence, automation, strategy, and value really mean. In reality, AI is only sometimes the answer. The first question is not “can we build this?” but “why do we need it?”

 

We all know the benefits of Artificial Intelligence. But what if we really just need CLEAR thinking? 

The exciting potential of Artificial Intelligence is clear: more automation, smoother workflows, and greater value for businesses and consumers alike. But these endless possibilities can come with a loss of focus, and the threat of oversaturation.  And with many of today’s AI initiatives, there is a visible lack of clarity – and a drift into illusion.

What’s the problem? 

You may have already seen it for yourself: your LinkedIn feed is stuffed with posts about nothing but AI – generated by AI. Engineers are comparing vendors; consultants posting frameworks; executives chasing newer and shinier platforms. The noise is endless, and deafening – from Think about your data! to Take this course! or Try that tool! 

But here’s the truth: AI is transforming not just what we do, but how we think. It’s transforming how we reason and make decisions and, ultimately, how we serve our customers. 

And what’s the solution? 

Socrates said it best:

The beginning of wisdom is the definition of terms.

 

 

Before we chase technology as an end in itself, we need to define what intelligence, automation, strategy and value really are. Neither oversimplifying nor drowning ourselves in jargon, we must navigate between both of these extremes – resisting the hype, and arriving at a realistic, actionable plan. 

That’s what clarity means to us. So now, how do we bring it to your business?

CLEAR thinking, with Kibo AI. 

Firstly, it’s essential to establish context and a clear direction for your business with benchmarking, tool evaluations and competitor analyses. But understanding what Artificial Intelligence can do today is only one part of the process – defining what it can do tomorrow is just as important. 

As business leaders, you must be prepared to define your own innovation game. That means taking a breath before you implement – and articulating precisely what AI might mean for your business model, your operating structures, and what you can offer your customers. 

Because when you align technological potential with strategic intent, that’s where clarity emerges.

So how can you get started in driving more successful outcomes? Here’s five pointers for powerful and CLEAR thinking: 

Context
What problems are we solving? And what is success worth? 

Our most important question is not “can we build this?” but ‘’why do we need it?” It’s critical to look at the most efficient ways to solve the problem – in reality, AI is only sometimes the answer. 

Linkage 

How is it tied to our business reality? 

In an environment of infinite data and endless right answers, clarity begins with a measurable definition of success. Is your goal to unlock creativity, improve efficiency, or enhance control?

Execution 

Do we have the right resources to deliver and maintain? 

Execution is the bridge between vision and impact. Success depends less on algorithmic sophistication than on quality delivery, process design, and quick decision-making.

Accountability 

Who decides what success looks like, and who owns delivery? 

Clearly defined authority helps keep your initiatives on the right track. When ownership is explicit, decisions are clear, and performance is tracked against agreed outcomes, businesses run a lot more smoothly.

 

Relevance


How will this stay valuable as technology evolves?
Relevance means building solutions that adapt to change – technologically and strategically. In a fast-moving world, future-proofing your operations is essential to staying on top. 

How to clarify your daily business.

Clarity in business starts with speaking the same language. While executives are speaking about governance, strategy and the workforce, engineers are talking models and algorithms. And from that gap, the need for clarity emerges. 

At Kibo AI, we provide an essential translation layer, bridging that gap between the technical and the strategic – building a common language, and unifying frameworks and decision patterns. 

Clarity also redefines speed. Real impact comes from proving that AI works in context.  AI initiatives could fall into the trap of aimless experimentation, or into over-engineered frameworks that never get off the ground. True agility lies in learning fast with a clear purpose. That’s why we structure 1-2 month Discovery and Delivery sprints, to test feasibility and measure value. 

In brief?

When your teams get clarity, you reach alignment – and when you reach alignment, the transformation begins. At Kibo, our mission is to bring that clarity back to the table.

Questions About AI Strategy and Clarity

What does clarity mean in the context of AI adoption?

Clarity means neither oversimplifying nor drowning in jargon, but navigating between both extremes to arrive at a realistic, actionable plan. It starts with defining what intelligence, automation, strategy, and value really mean for your business. When technological potential aligns with strategic intent, that is where clarity emerges.

CLEAR is a five-part structure for AI decision-making. Context asks what problems you are solving and what success is worth. Linkage asks how it connects to your business reality. Execution asks whether you have the right resources to deliver. Accountability asks who owns delivery. Relevance asks how the solution stays valuable as technology evolves.

Kibo AI provides a translation layer, bridging the gap between technical and strategic thinking and building a common language across engineering and leadership. Real impact comes from proving that AI works in context. That is why engagements are structured in 1-2 month Discovery and Delivery sprints, to test feasibility and measure value.

Marta Osadchuk advises C-suite executives on large-scale AI and automation programs across industries and has led global consulting engagements across Europe and the US.

Marta Osadchuk

Managing Director and Head of Consulting at Kibo AI

Linkedin

Marta advises CxOs on large-scale AI and automation transformations. ​

Marta has led global programs across financial services, insurance, and multinational industries, helping enterprises redesign operating models, drive productivity, and embed responsible innovation. ​

​
With a track record of steering high-impact projects across Europe and the US, Marta acts as a trusted partner to executives navigating the complexities of innovation driven change.

AI & Automation Transformation specialist, partnering with CxOs to redesign operating models and drive responsible innovation at enterprise scale.

]]>