Most businesses today are having difficulty implementing AI, whether large or small. According to reports from Gartner, McKinsey and Deloitte, up to 70-80% AI projects globally do not meet expectations or fail in the early stages, and this situation is especially evident in developing markets like Vietnam. The 3 main groups of causes are: Unprepared data platform - Lack of strategy and specialized personnel - Unintegrated systems and processes.
This article will analyze in detail the barriers to applying artificial intelligence, and at the same time present comprehensive solutions that Bizzi provides, helping businesses overcome challenges in the journey of applying AI. Let's find out what problems businesses encounter with AI!
Background: AI – An inevitable but challenging trend
Artificial Intelligence (AI) is gradually becoming a pillar in the digital transformation journey of enterprises, with a series of practical applications in finance, manufacturing, logistics, marketing and operations management. However, according to Gartner, more than 70% AI projects fail because they do not create measurable value (ROI) - mainly due to fragmented, unstandardized and improperly managed data.
In Vietnam, many businesses have begun experimenting with AI in activities such as automating financial reporting, invoice processing, or customer care chatbots, but most do not have a data platform that is “clean” and unified enough for AI to truly be effective.
The question is: What problems do businesses have with AI – and what is the right direction to turn AI into real value, not just a trend?
Difficulties in implementing AI in enterprises
Although AI is an inevitable trend, most businesses still not yet achieved real value from the implementation projects. difficulties in implementing AI It's not the technology itself, but the The governance, data and people platforms are not ready.
| Problem group | Root cause | Consequences |
| 1. Data | Data lacks standardization and is not managed consistently across departments. | AI “learns wrongly”, makes misleading predictions, and cannot create reliable insights |
| 2. Technology infrastructure | ERP/Closed CRM, old technology does not support API integration | AI cannot access or synchronize data, leading to “stillborn projects” |
| 3. Human resources | Lack of AI experts, internal teams not yet capable of exploiting new technology, fear of change | Project is delayed, dependent on external suppliers, cannot be expanded |
| 4. Strategy | No clear AI roadmap, lack of specific KPIs and ROI goals | Management loses confidence, projects are halted or redirected midway |
| 5. Cost | High initial investment, costly model maintenance and update costs | The project lacks sustainability and is difficult to maintain without quick results. |
| 6. Ethics & Legality | Lack of data control policies, risk of security breaches or AI bias | Causing legal risks, loss of reputation and trust from customers and partners |
| 7. Corporate culture | “Silo” thinking – departments do not share data, fear of being monitored | AI does not have enough data to learn, the analysis results are not comprehensive |
| 8. Financial and data management | Lack of FP&A (Financial Planning & Analysis) department and EPM (Enterprise Performance Management) to control data quality | Fragmented data does not create a solid foundation for AI to operate effectively. |
In short, most failures in AI projects do not come from algorithms, but from weak “data governance infrastructure”. That is why EPM becomes a mandatory stepping stone – helping businesses standardize data, unify processes, and create a foundation for AI to learn correctly – predict correctly – and bring real value.
Let's analyze typical reasons why businesses have problems with AI.
Lack of quality data and data governance
AI is only smart when the data is “clean” and “standard” enough. But this is an inherent weakness of most Vietnamese businesses.
Currently, data in businesses is often divided by department, for example: accounting, ERP, marketing or operations departments are all stored and managed separately, without a seamless connection.
Metadata standards are also not unified — the same supplier, but the ERP system sets one code, but the invoice uses another code. As a result, the system cannot “recognize” identical data for AI to learn accurately.
More importantly, many businesses do not have a Data Governance Framework:
- Who is responsible for creating and maintaining the data?
- Who can edit?
- Who ensures that data is audited and updated in a timely manner?
Without clear answers to those questions, data is prone to be biased, duplicated, or no longer useful.
Real-life example: A large FMCG company in Vietnam deployed an AI model to forecast sales, but the results were 25% off due to duplicate and inconsistent retail data between 3 systems (POS, ERP, CRM). The result was that the AI model "learned wrongly", forecasted wrongly, and led to wrong business decisions.
Technology infrastructure is not ready
What problems do businesses encounter with AI in terms of technology infrastructure? AI cannot operate effectively without a strong enough technology platform to “feed” and “connect” data. To avoid difficulties when deploying AI, businesses need two prerequisites:
- Big Data Infrastructure — enables the collection, cleaning, and analysis of massive amounts of data in real time.
- Integration Layer between AI and core systems such as ERP, CRM, EPM — helps data flow smoothly and synchronously.
However, the reality in Vietnam shows that:
- About 60% small and medium-sized businesses are still saving data manually on Excel or Google Sheet.
- Many ERP systems are outdated and lack open APIs, making it nearly impossible to connect to AI tools.
Real-life example: A logistics company deployed AI to optimize delivery routes, but repeatedly encountered “timeout” errors and lack of GPS data because the old operating system was not compatible with the new AI platform. As a result, the project stalled, the AI model did not achieve the expected efficiency, causing a waste of investment resources.
Lack of internal expertise and capacity in data – AI
One of the biggest challenges businesses face when implementing AI isn't the technology, but the humanSo what problems do businesses have with AI?
In Vietnam, human resources with expertise in data and AI are still extremely scarce — especially in key positions such as Data Scientist, Machine Learning Engineer, or Data AnalystMeanwhile, the internal team (especially the finance and accounting department) is mainly familiar with manual processes, depends on Excel, and has no experience in data modeling or reading and understanding results from AI.
Real life example: A retail group deployed AI to forecast cash flow, but the accounting team still entered data manually and did not standardize the format. The result: the AI model “learned wrongly”, produced erroneous forecasts, and forced the project to stop midway.
Besides, culture of "fear of change" AI implementation can also be difficult within an organization. When employees don’t trust or understand the benefits of AI, projects can easily encounter hidden resistance, leading to delays or failure.
Lack of clear AI strategy and ROI direction
One of the common reasons why AI projects fail is the lack of an overall strategy and clear quantitative goals.
Many businesses today deploy AI as a fad — seeing competitors doing it and following suit, or experimenting in fragmented ways in each department without linking it to the organization’s long-term growth strategy. As a result, AI does not create real value, but just stops at the “try it out for the sake of it” level.
In reality, most businesses do not have an AI Roadmap: they do not define specific KPIs, do not measure ROI (return on investment), and do not have a process for managing results.
Real-life example: A financial company deploys an AI chatbot to support customers, but does not set clear goals on how much % of customer service costs will be reduced or how much % of automatic response rate will be increased. As a result, without data to evaluate effectiveness, the project falls into a state of "clinically dead".
The core problem is not AI technology, but the lack of a mechanism to connect strategy – data – performance. And that is the gap that EPM systems can fill, helping businesses turn strategic goals into plans, forecasts and measurable results.
Barriers to AI Adoption: Tintegration between AI and existing systems
For AI to be effective, the prerequisite is Seamless connectivity with core systems such as ERP, CRM or accounting software. However, the reality in many Vietnamese businesses is completely opposite.
Most of Legacy ERP or CRM lacking open API or SDK, making AI historical data inaccessible — which is an important “raw material” for training predictive models. AI integration therefore becomes a complex problem, requiring high IT costs, long implementation time and risk of operational disruption if not tightly managed.
Real life example: A manufacturing company with five plants had production planning data stored in the SAP system, while accounting data was stored in another internal software. When AI was deployed to forecast raw material demand, the system failed to “connect” these two data sources — resulting in high errors and projects that did not achieve expected results.
This issue shows: AI cannot operate in a fragmented data environment.. Businesses need an intermediary platform like EPM Sactona or Bizzi Helps consolidate financial - operational - business data, creating a standardized data foundation before AI is applied at a strategic level.
High investment costs and disproportionate efficiency
AI is not just a technology problem, but a long-term investment problem – requiring large costs for software, hardware, human resources and training.
For many small and medium-sized enterprises, the difficulty in implementing AI comes from the initial investment of up to billions of VND. In addition, AI is not a “one-time purchase” project – the system needs to be nurtured with continuous training data, leading to increased costs for maintenance and model optimization over time.
Real-life example: A transport company invested more than 1 billion VND to deploy AI to predict vehicle maintenance. However, after only one year, the project had to be suspended because the cost of cleaning and standardizing data took up more than 40% of the total budget, while the results were unclear.
This reflects a reality: without a standardized data platform and a clear ROI measurement mechanism, the difficulty of implementing AI is that it becomes a “financial burden” instead of a value creation tool. That is also the reason why many CFOs are starting to turn to EPM – a solution that can help standardize data, control costs, and simulate return on investment (ROI) before businesses invest in large-scale AI projects.
Ethical, legal and data security issues
AI opens up huge opportunities for businesses, but comes with barriers to adopting artificial intelligence,
Ethical and legal risks if not strictly controlled. One of the difficulties in implementing AI is AI bias – when the model learns from unbalanced data, leading to biased or unfair results. At the same time, the problem of personal data infringement and violation of customer information security regulations is also increasingly worrying, especially in industries such as finance, insurance, and healthcare.
In Vietnam, the legal framework for using customer data for AI training is still limited. Many businesses do not have a clear process to ensure that data is anonymized, encrypted and used for the right purposes.
Real-life example: An insurance company applied AI to score customer risk, but faced backlash when it was discovered that it was using unanonymized personal data. This incident not only affected the brand reputation but also forced the company to pause the project to review the entire security process.
This issue raises urgent management requirements. data transparency and compliance (Data Governance & Compliance) — elements that a system like EPM can help establish through data authorization, access control, and tracking the flow of data throughout the organization.
Lack of Data-Driven Culture
AI only truly shines when businesses operate on data, not gut feelings. However, this is an inherent weakness of many organizations today.
In reality, employees and managers often make decisions based on personal experience, rather than on real data. In addition, the lack of a transparent data sharing process between departments causes information to be fragmented – data is considered the “private property” of each department. As a result, AI cannot learn from a comprehensive picture, leading to inaccurate forecasting models and loss of application value.
Real-world example: Sales keeps transaction and customer pipeline data separate from finance. When deploying AI to forecast revenue and cash flow, the system only learns from a portion of the data — resulting in inaccurate forecasts and misaligned budget strategies.
This shows that the difficulty in implementing AI is that businesses do not have a “data culture” — where all decisions, processes and measurements are based on shared and verified data. This is also the reason why pioneering businesses choose EPM (Enterprise Performance Management) as the central platform: to help standardize, share and visualize performance data, thereby forming a sustainable culture of data-driven decision-making.
Bizzi – Comprehensive solution for accounting and finance departments ready to apply AI
The difficulties in implementing AI show that the decisive factor is not the algorithm, but data quality, governance and analytical capabilities of the finance and accounting department. This is the “bottleneck” group in most digital transformation projects, where planning, forecasting and actual data are not synchronized or controlled uniformly.
In that context, Bizzi – in the role of Pioneer in financial automation and exclusive distributor of EPM Sactona solutions in Vietnam – providing a solution ecosystem to help accounting and finance departments overcome barriers to applying artificial intelligence:
- Standardize data and performance management (EPM) processes
- Automate planning, forecasting, and analysis (FP&A) tasks
- Create a high-quality data platform that is AI-ready.
1. The role of the accounting and finance department in AI application
The accounting and finance department concentrates on high-value data: plans, costs, revenues, cash flows, budgets. However:
- 70% Unnormalized Financial Data, separated between ERP, accounting, Excel and operations.
- Manual reporting, data is slow and unreliable.
- Lack of data quality control mechanism, leaving AI without “clean material” to learn and predict.
Before implementing AI, businesses need to Building an EPM data management infrastructure: Unify data sources, standardize planning – forecasting – analysis processes, and create a training ground for AI. This thorough preparation will help remove barriers to applying artificial intelligence during deployment.
2. Bizzi solution ecosystem for accounting and finance department
a. Sactona – Enterprise Performance Management (EPM)
Sactona is the solution Enterprise Performance Management (EPM) Developed by Outlook Consulting – Japan's leading EPM consulting and implementation company, with more than 25 years of experience in financial performance management for global corporations.
In Vietnam, Bizzi To be exclusive partner Distribute and support the implementation of Sactona, helping domestic enterprises access Japanese standard EPM technology.
Unlike traditional reporting software, Sactona not only aggregates data, but also transform them into actionable insights, help businesses:
- More accurate forecasts,
- Make decisions faster,
- And most importantly: be ready to integrate AI most effectively.
Sactona is designed for businesses:
- Have revenue of over 200 billion VND/year
- Deployed major ERP systems such as SAP, Oracle, Microsoft Dynamics...
- Own an FP&A team of 5 or more people, need to standardize and automate the process of planning - forecasting - performance analysis
Sactona is an EPM platform exclusively distributed by Bizzi in Vietnam, providing:
- Consolidate plan – forecast – actual data from ERP, CRM, manufacturing systems, Excel… into a “single source of truth”.
- Automate FP&A processes: synthesize, compare, update data in real time.
- Setting up Data Governance: decentralization, censorship, data flow monitoring – AI-ready.
- Data Platform for AI: standardized data, metadata, quality measurement, helping AI learn accurately and forecast reliably.
Sactona was born to reconcile two worlds: the familiar flexibility of Excel and the administrative power of the Japanese standard EPM system. Thanks to that, Vietnamese businesses can easily access and deploy quickly, but still achieve high efficiency in planning, forecasting and making financial decisions.
Typical case study
Real-world evidence from leading corporations shows that most businesses that encounter AI problems when implementing Sactona not only shorten forecasting time and reduce operating costs, but also create clear ROI and improve global financial analysis capabilities for the CFO team.
- Panasonic: Reduced financial forecasting time from 15 days to 5.5 days.
- LIXIL: Deployed synchronously in 8 countries in just 2 months, reducing 40% financial planning management costs.
- Monex Group: Speed up financial analysis and decision-making by 3x, thanks to data consolidation and predictive AI on the Sactona platform.
Key solutions that Bizzi provides
In addition to distributing Sactona, Bizzi provides solutions to support accounting and finance departments to operate effectively:
- Automatically enter and compare invoices and documents → reduce manual operations to 70%.
- Synchronize data from multiple sources: ERP, POS, Excel, bank → keeps data clean and consistent.
- Deviation monitoring and alerting tools → detect abnormal data early, support quick decisions.
- Digital and debt optimization solution helps track receivables/payables and control cash flow continuously.
- The platform connects businesses with financial institutions to support working capital and supply chain financing, helping businesses maintain stable cash flow and reduce the risk of production and business disruption.
This solution not only reduces manual errors but also prepares “AI-ready” data for cash flow forecasting, costing, or performance analysis.
c. Smart Reporting & Analytics
Bizzi supports the finance department to create Dynamic, visual reporting, easy to understand for leaders:
- Real-time KPI dashboard, trend and variance analysis.
- “What-If” scenario analysis to forecast costs, cash flow and performance.
- A standardized data platform helps AI learn financial forecasting models, reducing the risk of bias.
d. Digital transformation consulting & AI integration
Bizzi accompanies businesses in:
- Assess current state of FP&A data and processes → Identify gaps before implementing AI.
- EPM Setup & Automation Solutions → standardize data, integrate existing ERP/CRM.
- Deploy predictive AI & performance analytics → data has been standardized and monitored, AI models quickly create real value.
Conclude
The above article has summarized the barriers to applying artificial intelligence in businesses, the reasons why businesses have problems with AI. From fragmented data, unprepared technology infrastructure, to lack of expertise and strategy, failure does not lie in the AI algorithm, but comes from difficulties in implementing AI due to non-standard data platforms, lack of automation in FP&A processes and unprepared teams.
In that context, EPM – like Sactona exclusively distributed by Bizzi in Vietnam – is not just software, but a strategic piece. help businesses:
- Standardize financial and operational data, creating a trusted foundation for AI.
- Automate planning, forecasting, and performance analysis (FP&A), improve decision-making capabilities.
- Seamless integration with ERP, CRM systems, making AI work efficiently and measuring ROI clearly.
- Forming a sustainable data-driven culture, reducing risks and accelerating digital transformation.
In addition to Sactona, Bizzi solution ecosystem – from accounting automation, debt and cash flow management, smart reporting to AI integration consulting – helping businesses overcome difficulties in implementing AI, comprehensive AI implementation from data normalization to smart analytics application, ensuring AI truly brings value.
In short, without a solid data and EPM foundation, AI is just a theory; with Bizzi and Sactona, AI becomes a real value-creating tool, helping businesses make quick, accurate, and sustainable decisions in the digital age.
Registration here to experience the solution to see the real dashboard, discover how Sactona – Modern CFO Assistant operates intelligently and optimizes financial performance! See more solutions from Bizzi in