After discussing stage-based work, planning, problem solving, control, organizational agility, and metrics, the next logical topic is the role of data. Generative artificial intelligence can already support project managers in analysis, information summarization, report preparation, risk assessment, solution formulation, and stakeholder communication. But its real value depends on something much more fundamental than the tool itself: the quality, structure, and governance of the data the organization works with.

For project managers, this is a particularly important shift. Until recently, data was often seen as a reporting necessity: schedules, budgets, tasks, minutes, risks, contracts, correspondence. In an environment shaped by generative artificial intelligence, the same data becomes an asset that can support decision-making. If it is incomplete, scattered, or outdated, AI will produce confident but inaccurate conclusions. If it is well organized, verifiable, and contextualized, it can accelerate work and improve the quality of management.

Why data is becoming central to project management

Projects have always depended on information: what needs to be done, by whom, by when, with what budget, under what constraints, and with what expected result. The difference today is that information no longer serves only tracking. It can be used for summarization, comparison, forecasting, identifying inconsistencies, and preparing management options.

This creates a new responsibility for the project manager. They no longer need only to “collect reports,” but to understand what type of data is needed, how reliable it is, and how it can be used safely. In this sense, data management is not only a task for the IT department. It is part of the project management discipline.

The practical question is no longer only “do we have information,” but:

is it accurate
is it current
is it complete
is it connected to the right context
is it accessible to the people who need to use it
is it safe to use with an artificial intelligence tool

The different types of data in the project environment

To use generative artificial intelligence responsibly, we first need to understand the data landscape around the project. It usually includes several categories.

Planning data

This includes schedules, scope, objectives, stages, tasks, dependencies, roles, responsibilities, and expected outcomes. It answers the question of what needs to happen and in what sequence.

If this data is unclear, AI can help with formulation, but it cannot replace the management decision itself. For example, if the project objective is described too generally, the tool may suggest well-written wording, but that will not solve the lack of specificity.

Execution data

This includes statuses, completed tasks, blockers, changes, decisions, meeting minutes, and operational reports. It shows what is happening in real time.

This data is critical for progress analysis. If statuses are updated only formally or with delay, any summary will be inaccurate. Generative AI can summarize the available information, but it cannot compensate for systematically poor reporting.

Risk data

Risk registers, issues, dependencies, assumptions, and decisions are especially valuable. They make it possible to see not only what is planned, but also where the project may deviate.

Here, AI can be useful for grouping risks, identifying recurring themes, and formulating response options. However, responsibility for assessing probability, impact, and management decisions remains human.

Financial and contractual data

Budgets, costs, contracts, annexes, offers, payments, and delivery terms form a sensitive category. They often contain confidential information and must be used with particular care.

In a project environment, it is important to define clearly what financial and contractual data may be used with artificial intelligence tools, under what conditions, and with what restrictions.

Communication data

Emails, meetings, messages, customer feedback, and internal discussions often contain the richest context. This is where ambiguities, tensions, changes in expectations, and decisions that are not always reflected in official documents become visible.

However, this information is also risky: it may contain personal data, trade secrets, unverified claims, or sensitive internal positions. Its use therefore needs to be carefully governed.

Data quality: the foundation of reliable artificial intelligence

Generative artificial intelligence often sounds convincing even when it is wrong. This makes data quality critical. If the input information is weak, the output may be formally well written, but dangerous from a management perspective.

The most important characteristics of quality data are:

accuracy: the information reflects the real situation
currency: the data is not outdated
completeness: there are no critical gaps
consistency: terms and structures are used in the same way
traceability: it is clear where the information comes from
context: the data is connected to the project, decision, or problem for which it is being used

For example, if tasks in a project are described as “done,” “almost done,” “in progress,” “waiting,” and “no problem,” without common rules, AI will struggle to produce a reliable analysis. The data may look rich, but it will be weak for management purposes.

Context is as important as the data itself

In projects, the same data can mean different things depending on context. A three-day delay may be insignificant for an internal administrative task, but critical for a customer delivery or for an activity on which other teams depend.

That is why the project manager needs to provide not only data, but context:

what the project objective is
which constraints are critical
which decisions have already been made
which risks have been accepted
which dependencies matter most
what is considered success

Without context, AI may suggest a generic solution. With context, it can support a more useful management analysis.

Data as a source of questions, not only answers

One of the most useful roles of generative AI in project management is not simply to provide answers, but to formulate better questions.

For example:

Which dependencies are not clearly described?
Where is there a contradiction between the schedule and available capacity?
Which risks recur across multiple projects?
Which decisions were made informally but not documented?
Where is there a mismatch between customer expectations and the internal plan?
Which tasks appear too large or unclear?

This is especially valuable because the problem in a project is often not a lack of information, but a lack of proper interpretation. AI can help identify patterns, but the final decision remains a management responsibility.

Private, internal, and public data: not everything should be used in the same way

One of the most important topics is data classification. Not all data is suitable for use in the same way. Businesses should distinguish at least three levels.

Public data

This is data that can be used freely: public descriptions, general market information, published conditions, regulatory texts, public profiles, and materials.

It is the lowest-risk category, but also the most general. It is suitable for initial analyses, drafts, ideas, and comparisons.

Internal data

This includes process data, schedules, tasks, resources, internal reports, and operational decisions. It is highly valuable for analysis, but should not be shared without control.

Here it is necessary to define who has the right to use it, in what environment, and with what tool.

Sensitive and confidential data

This includes personal data, contracts, prices, trade secrets, strategic decisions, financial data, and information about customers and partners. Its use must be restricted and clearly regulated.

The practical rule is simple: if you would not freely send a piece of information outside the organization, you should not enter it without careful consideration into a public or uncontrolled AI tool.

Access management: who sees what and why

Generative AI can accelerate work, but it can also make weaknesses in access management more visible. If organizational data is shared chaotically, tools may unintentionally expose information to people who should not see it.

That is why access management should be part of the preparation:

clear roles and permissions
access based on work needs, not convenience
review of shared folders and documents
rules for using personal, customer, and contractual data
separation of data that can be used for analysis from data that should not be entered into an AI environment

For project managers, this is especially important because they often work with information from multiple functions and stakeholders.

From scattered documents to an organized knowledge base

Many organizations have a large volume of information, but limited real management usability. The reason is that information is scattered: emails, spreadsheets, folders, chats, minutes, personal notes, and different versions of documents.

To be useful in an environment with generative AI, this information needs to become a better organized knowledge base.

Practical steps:

Define the main document categories: plans, decisions, risks, contracts, minutes, reports.
Standardize naming and structure.
Maintain a single source of truth for key data.
Archive outdated versions instead of allowing them to mix with current ones.
Create brief rules on what is documented, where, and by whom.

This is not just an administrative task. It is preparation for more reliable use of artificial intelligence.

When to use generative AI in a project

Not every task is suitable for artificial intelligence. The best results come from tasks that require information processing, summarization, structuring, or preparing options.

Suitable applications:

summarizing meetings and decisions
structuring risks
preparing communication options
comparing requirements
formulating acceptance criteria
analyzing recurring problems
preparing draft reports

Higher-risk applications:

final financial decisions
legal interpretations without expert review
assessment of people
work with sensitive data without rules
automated high-risk decision-making

The practical principle is: use AI as an assistant for preparation and analysis, but keep human responsibility for decisions.

A minimum framework for working with data and GenAI in a project environment

The organization does not need to begin with a heavy policy. It is more useful to start with a short and applicable framework.

It may include:

Data classification
Which data is public, internal, confidential, and strictly restricted.
Rules of use
What may and may not be entered into artificial intelligence tools.
Output verification
Who checks the output, how facts are confirmed, and how assumptions are marked.
Source of truth
Where the current plans, decisions, risks, schedules, and minutes are stored.
Human responsibility
Which decisions cannot be delegated to a tool and who remains accountable.
Traceability
How to document that a given analysis was supported by artificial intelligence, when this is important.

This framework does not slow down work. On the contrary, it reduces risk and makes the use of new tools safer.

How to start: a practical 30-day approach

For small and medium-sized businesses, the best start is limited and measurable.

Week 1: data map

Identify what data you use in projects:

plans
schedules
tasks
risks
minutes
contracts
reports
customer communication

Mark where it is located, who maintains it, and which data is highest-risk.

Week 2: classification and access

Divide the data into public, internal, and sensitive. Review who has access to key folders and documents. Define what should not be entered into uncontrolled tools.

Week 3: pilot use

Choose one safe use case: for example, summarizing project minutes, structuring risks, or preparing a draft report. Measure whether this saves time and improves quality.

Week 4: rules and improvement

Based on the pilot, create brief rules: which tasks are suitable, how outputs are checked, what data is not used, and who remains responsible.

Common mistakes when using GenAI in project work

The most common mistakes are:

using sensitive data without clear rules
rusting output without verification
lacking a single source of truth
using outdated documents
mixing facts, assumptions, and opinions
expecting the tool to solve a management problem the organization has not clarified
lacking training for the people who will use the tool

It is worth remembering: generative artificial intelligence amplifies what already exists. If data and processes are chaotic, it may amplify the chaos. If they are well organized, it can amplify productivity.

Conclusion

Data is the foundation on which generative AI can be useful in project management. For project managers, this means a new practical competence: understanding what data the organization has, what quality it has, what context it carries, who has access to it, and how it can be used safely. Only then can AI become a reliable assistant, rather than a source of confident but incorrect conclusions.

The Ruse Chamber of Commerce and Industry publishes materials like this to support businesses in the region with practical guidance on managing digital transformation, adopting new technologies, and increasing organizational readiness to work with artificial intelligence.

If you would like to discuss how your organization can prepare its data for safer and more useful use of generative artificial intelligence, contact me at sminchev@rcci.bg or +359 895 890 123.

Note: This publication was prepared with the assistance of generative artificial intelligence, which supported the structuring and formulation of the content. The final text reflects the author’s expert contribution, which ensures its accuracy and practical relevance.

The Data Landscape in the Age of Generative AI: How Project Managers Can Prepare the Organization for More Reliable Decisions

Why data is becoming central to project management