Software projects rarely go exactly according to plan.
Requirements change. Technical challenges appear. Priorities shift. Teams discover new information during development. Even with experienced project managers and developers, predicting exactly when a project will be completed can be difficult.
Traditional project estimates are still useful for planning, budgeting and setting expectations. However, a single number can sometimes create a false sense of certainty when the underlying information is limited.
Artificial intelligence is introducing another way to approach software project estimation.
Rather than simply asking AI, “When will this project be finished?”, businesses can use AI to analyse project data, explore estimation techniques, test different scenarios and understand the probability of different outcomes.
One example explored by Waracle combined ChatGPT with Monte Carlo simulation to investigate how AI could support a more data-driven approach to software project estimation.
Why Is Software Project Estimation So Difficult?
Estimating individual development tasks may appear straightforward. A developer might estimate that a particular feature will take five days, for example.
The challenge is that a complete project involves hundreds of variables.
Tasks can depend on one another. Requirements can change. Technical issues can take longer than expected. Teams can become unavailable, integrations can fail and new information can change the original scope.
This means that adding individual task estimates together does not necessarily produce a reliable final deadline.
Instead of asking:
“Exactly how many days will this project take?”
A more useful question can be:
“What are the possible outcomes, and how likely is each one?”
This shift from a single prediction towards a range of probabilities can give project teams a more realistic understanding of uncertainty.
What Is Monte Carlo Simulation?
Monte Carlo simulation is a statistical technique used to explore uncertainty by running many possible scenarios.
For software project estimation, a team could provide information such as minimum, most likely and maximum durations for different tasks.
The simulation then uses different combinations of those possibilities to generate a distribution of potential project outcomes.
Instead of producing one fixed deadline, it can show a range.
For example, a project might have a 50% probability of being completed within one timeframe, while a later date could provide a much higher probability.
The purpose is not to predict the future perfectly.
It is to provide a clearer picture of the uncertainty surrounding the project.
This can make conversations around deadlines, resources and expectations more transparent.
How Can ChatGPT Support Project Estimation?
The interesting part of using AI for project estimation is not simply asking ChatGPT to generate a deadline.
AI can act as a tool for exploring the problem itself.
It can help project teams:
- Understand estimation techniques
- Identify the data required
- Create example scenarios
- Generate analysis
- Develop Python code
- Test calculations
- Explain statistical concepts
- Interpret results
- Create visualisations
For professionals who understand the business or project problem but do not have advanced statistical or programming expertise, this can make complex analysis more accessible.
AI can help bridge the gap between having an idea and testing that idea with data.
AI-Generated Code Still Needs Human Oversight
There is an important lesson here.
Generating code does not automatically mean generating a reliable solution.
Earlier AI systems could produce technically plausible code that still contained errors or failed to achieve the original objective. Fixing one problem could sometimes introduce another.
This highlights a fundamental principle of AI-assisted technical work:
AI can accelerate the process, but people still need to understand and validate the result.
Project professionals need to check whether the assumptions are appropriate, whether the calculations make sense and whether the output actually answers the original business question.
The technology can assist with the work, but responsibility for interpreting the result remains with the people using it.
From Generating Code to Analysing Data
More capable AI environments can take this process further.
Instead of simply producing Python code for someone else to execute, an AI environment can work with data, run calculations, identify errors, refine the analysis and produce visualisations.
This creates a more interactive workflow.
A project team could provide historical project information and use AI to explore different scenarios, calculate potential outcomes and visualise the distribution of results.
For non-technical stakeholders, this can be particularly useful.
A probability distribution or project timeline visualisation can often communicate uncertainty more effectively than a spreadsheet containing dozens of numbers.
AI as a Project Analysis Partner
This represents a broader change in how businesses can use artificial intelligence.
Traditional software generally performs a predefined task.
AI can help people investigate a problem.
It can help identify what information is missing, suggest analytical approaches, explain unfamiliar concepts and explore different scenarios.
That makes AI potentially useful as a project analysis partner.
However, this does not mean that AI replaces project managers, developers or consultants.
The human team still needs to define the business objective, understand the context, challenge assumptions and make the final decisions.
AI can make analysis faster and more accessible, but the quality of the outcome still depends on the quality of the questions and information provided.
Better Estimates Start With Better Data
AI cannot create reliable project intelligence from nothing.
If the underlying project information is poor, the resulting analysis will also have limitations.
This makes historical project data increasingly valuable.
Businesses that consistently record:
- Task durations
- Project timelines
- Dependencies
- Resource requirements
- Changes in scope
- Delivery delays
- Historical project outcomes
can eventually build a much stronger foundation for data-driven estimation.
AI can then help analyse that information and identify patterns that may otherwise be difficult to see.
The lesson is simple:
Better data creates better opportunities for better analysis.
Project Estimation Should Support Decisions
A statistical model cannot capture every factor affecting a software project.
Business priorities can change. Customers can request new features. Market conditions can shift. Resources can become available or unavailable.
These factors can influence project timelines regardless of what a mathematical model suggests.
For this reason, AI-generated probabilities should support project conversations rather than replace them.
Instead of telling stakeholders:
“The project will finish on this date.”
Teams can have a more informed conversation:
“Based on the available data, these are the potential outcomes and their associated probabilities.”
That difference can lead to more transparent planning and more realistic expectations.
Making Complex Analysis More Accessible
Perhaps one of the biggest opportunities for AI-assisted project estimation is accessibility.
Previously, advanced statistical modelling might have required specialist knowledge in programming, data science or statistics.
Today, a project professional can use generative AI to learn about a technique, create a starting point, develop code and explore the results.
That does not remove the need for expertise.
It lowers the technical barrier to experimentation.
For consultants, project managers and business leaders, this can make it easier to investigate questions that might previously have remained unexplored because the required technical resources were not readily available.
Visualising Uncertainty
Another important advantage is the ability to turn complex calculations into understandable visual information.
Project stakeholders do not always need another spreadsheet.
They may need to see the range of possible outcomes.
Charts and probability distributions can help communicate why a particular deadline carries more or less risk.
Instead of presenting a single completion date, a project team can show different scenarios and discuss what level of certainty is required before making a commitment.
This can make project planning more transparent and collaborative.
What Does AI Mean for Project Managers?
AI-assisted estimation does not make project management less important.
It changes where some of the effort can be focused.
AI can assist with calculations, data analysis, scenario modelling and technical experimentation.
Project professionals still need to:
Understand the business context.
Challenge assumptions.
Communicate uncertainty.
Manage stakeholders.
Make decisions.
The future is therefore less about AI versus project professionals and more about how AI can give those professionals better information to work with.
The Future of AI-Powered Software Estimation
As AI tools become more capable of analysing data, writing and executing code, testing scenarios and explaining complex results, software teams will have more opportunities to use advanced estimation techniques.
Historical project data could become easier to analyse.
Scenario planning could become faster.
Potential risks could be explored before major commitments are made.
But businesses should avoid treating AI-generated estimates as guarantees.
Software development will always contain uncertainty because technology, people, requirements and business priorities can change.
AI cannot eliminate uncertainty.
What it can do is help teams understand that uncertainty more clearly.
The real opportunity is not to find a perfect project deadline.
It is to make project planning more data-driven, transparent and informed.
AI is changing the way businesses approach complex problems, and software project estimation is a good example.
Rather than relying solely on a fixed number, teams can combine historical data, statistical techniques such as Monte Carlo simulation and AI-assisted analysis to explore a wider range of possibilities.
The result is not certainty.
It is better information.
And when project teams have better information, they can have better conversations about risk, resources, deadlines and delivery.
AI may not tell you exactly when your project will finish. But it can help you understand what could happen before you commit to the plan.
