Adnoc Gas Megaproject Cost Estimator with Google ADK & Gemini 2.5 Flash
Contractors announced $8.2bn in awards for Adnoc Gas projects. Build a multimodal estimation agent using Google ADK and Gemini 2.5 Flash to ingest engineering diagrams, bill-of-quantities (BOQ) spreadsheets, and site reports. Produce cost variance projections with a confidence threshold within 3% of baseline engineering estimates.
What you are building
The core problem, expected build, and operating context for this challenge.
Utilize Google ADK with Gemini 2.5 Flash to process engineering blueprints and tabular BOQs for gas megaproject financial control.
How work is evaluated
Evaluates accuracy of cost estimation and scope mismatch detection from BOQ files and diagrams.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
valid_calculation
Checks sum calculation accuracy across BOQ lines
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
estimation_accuracy
Accuracy relative to benchmark gold cost estimation • target: 0.97 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Configure Google ADK with Gemini 2.5 Flash for multimodal data processing.
Parse engineering schematics (P&ID) alongside pricing spreadsheets.
Detect cost anomalies and scope gaps across multi-billion-dollar project contracts.
Generate structured budget summary reports conforming to corporate governance rules.
Reference links and supporting material
BoQ datasets and schematic drawings based on UAE gas pipeline and processing facility expansion projects.
Challenge setup
Evaluates accuracy of cost estimation and scope mismatch detection from BOQ files and diagrams.
Challenge input
JSON with listed items, quantities, unit prices, and blueprint diagram image reference
Google ADK
Enables Vertex/Gemini native agent building.
Gemini 2.5 Flash
Fast processing of visual diagrams and text BOQs.
Evaluated output
JSON summary with total variance, flagged items, and risk score
- Checks sum calculation accuracy across BOQ lines
- Accuracy relative to benchmark gold cost estimation • target: 0.97 • range: 0-1
- Benchmark: MegaprojectCostBench
- Estimation Accuracy target: 0.97
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
- Google ADK · Required
- Google Gemini · Optional
- Gemini 2.5 Flash · Optional
Evaluation contract
- valid_calculation · Weight 1
- estimation_accuracy · Weight 1
Metrics
- estimation_accuracy · Target 0.97
Benchmarks
- MegaprojectCostBench
- energy
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON with listed items, quantities, unit prices, and blueprint diagram image reference
Output: JSON summary with total variance, flagged items, and risk score
Execution harness: Python
Evaluation tools
- Google ADK · Agent System Framework
- Gemini 2.5 Flash · Multimodal Model
Recipe freeze marker
No recipe freeze time is recorded.