AI Disclosure

1. What the AI does and what the engine does

PlanGent combines two parts:

• The AI agent reads your documents, understands the scope, asks you questions, proposes zoning, maps BOQ items to activities and explains the results. It uses a large language model from the provider listed on our Subprocessors page.
• The scheduling engine builds the network from a fixed activity template, calculates the critical path and float, runs the DCMA 14-point assessment and writes the P6 file. These calculations are deterministic: the same inputs give the same result.

2. Known limits

• The AI can misread a drawing, a table or a contract clause, especially in scanned or low-quality documents.
• Quantities split between activities come from the drawings or from your answers. If neither settles a split, PlanGent asks you rather than guessing; if it ever states a split without a basis, treat it as a mistake.
• A passing DCMA result checks the schedule’s structure, not whether durations or logic suit your site.
• PlanGent does not open your file in Primavera P6. You verify it in your own P6.

3. You stay in control

Decisions that belong to the planner, such as soft logic, construction order, splits that the drawings do not settle, calendars and float acceptance, are always asked, with a recommendation and its reason. You can change any answer and rebuild.

4. Learning

If you allow it, PlanGent records your corrections and preferences, such as your preferred splits and sequences, to improve what it recommends to you. Across users it keeps only anonymized patterns. You can switch learning off at any time.

5. Report a wrong answer

Tell PlanGent in the chat (“that is wrong because…”) and it will correct the work, or write to us. Corrections help it improve.