Using AI for Project Estimation Without Historical Data.

Artificial Intelligence Complexity

Srikana Ray

Community Champion

Project Professional, PMP

Has anyone successfully used AI tools to estimate and plan a complex software development project without historical data?

I am particularly interested to learn more about how you approached effort estimation, forecasting timelines and handling uncertainty and risks.

What worked well, what did not work and how reliable were the AI-generated outputs in practice? Please share your insights.

Posted: Apr 8, 2026 7:44 PM

Sort By:

< 1 2 >

Srikana Ray

Community Champion

Project Professional, PMP

Apr 16, 2026 10:32 AM

Replying to Robert London

...

GenAI can help you draft your estimates and schedules using a variety of data. I do a lot of ERPs, and estimates from vendors and posted project schedules are available on the Internet. In addition, since no two projects or schedules are the same, using GenAI to draft and then help you build out your task-level estimates can be effective. History data is just a starting point

Thank you for sharing your insights.

Posted: Apr 16, 2026 8:03 PM

Abolfazl Yousefi Darestani Manager, Quality and Continuous Improvement| Hörmann-TNR Industrial Doors Newmarket, Ontario, Canada

AI can be a useful tool, however, the results may not be reliable specifically of you do not have any historical data. It also depends on the project.

Posted: Apr 17, 2026 7:24 AM

Stefano Gitto Software Architect | AI Project Planning & Scheduling Roma, Italy

In my experience, AI can be valuable for project estimation even when historical data is unavailable, but I wouldn't treat its estimates as definitive commitments.
I believe the most effective approach combines three elements:

AI-assisted planning: Use AI to break down the project into a WBS, identify dependencies, and suggest initial task durations.
Deterministic scheduling: Use a proper scheduling engine to calculate dates, dependencies, calendars, and the critical path rather than relying on AI-generated dates alone.
Quantitative risk analysis: Apply three-point duration estimates and Monte Carlo simulations to understand completion probabilities instead of relying on a single finish date.

The key is keeping human validation in the process. AI should support the project manager's judgment, not replace it.
I'm particularly interested in how other PMs validate AI-generated duration estimates when historical project data is limited.

Posted: Oct 11, 2026 6:17 PM

< 1 2 >

Please login or join to reply

Using AI for Project Estimation Without Historical Data.

Sponsors

Newsletters