RFI-AI-Augmented Support.pdf

APPENDIX D – AWARD TABULATION & SELECTION SHEET Page 1 of 8

REQUEST FOR INFORMATION (RFI)

Artificial Intelligence (AI) Augmented Cost Estimating and Estimation Support Capabilities

1. PURPOSE

The South Carolina Department of Transportation (SCDOT) is seeking information from industry regarding the current and emerging use of Artificial Intelligence (AI), Machine Learning (ML), Predictive Analytics, Generative AI, and related technologies to support transportation project cost estimating.

The purpose of this Request for Information (RFI) is to better understand:

• Available technologies and solutions

• Potential use cases within transportation agencies

• Benefits and limitations of AI-assisted estimating

• Implementation approaches

• Data requirements

• Cost considerations

• Risks and challenges

• Lessons learned from other organizations

SCDOT is not seeking proposals, pricing commitments, or contractual offers through this RFI. Responses will be used for market research, strategic planning, and evaluation of potential future initiatives.

2. BACKGROUND

SCDOT develops transportation project cost estimates throughout the project delivery lifecycle, including:

• Planning-level cost projections

• Program development estimates

• Project scoping estimates

• Preliminary design estimates

• Engineer’s estimates prior to advertisement and letting

The Department maintains significant historical information related to transportation project development and construction, including:

• Historical bid tabulations

• Awarded contract data

• Pay item databases

• Quantity histories

• Unit price histories

• Construction performance information

• Project characteristics and classifications

• Geographic and district-specific information

Increasing project complexity, workload demands, market volatility, inflationary pressures, supply chain fluctuations, and workforce transitions have created interest in technologies that may improve the consistency, defensibility, transparency, and efficiency of estimate development.

APPENDIX D – AWARD TABULATION & SELECTION SHEET Page 2 of 8

SCDOT is exploring whether AI-enabled technologies can augment, but not replace, engineering judgment and established estimating practices.

3. PROBLEM STATEMENT

SCDOT seeks information regarding AI-enabled solutions that may assist with:

3.1. Planning-Level and Scoping Estimates

• Development of planning-level estimates from limited project definition

• Development of conceptual cost estimates

• Development of funding forecasts

• Cost range development for programming and budgeting activities

3.2. Preliminary Design Estimates

• Identification of likely pay items

• Identification of potentially omitted pay items

• Estimation of conceptual and preliminary quantities

• Comparison against similar historical projects

• Development of preliminary cost estimates

3.3. Engineer’s Estimate Development and Review

• Recommended unit price development

• Historical bid analysis

• Regional cost adjustments

• Quantity scale effects

• Inflation adjustments

• Commodity and market trend adjustments

• Estimate quality assurance and quality control

3.4. Estimate Validation

• Identification of quantity outliers

• Identification of unit price outliers

• Detection of inconsistent assumptions

• Comparison against historical project trends

• Identification of potential estimating errors

3.5. Forecasting and Continuous Learning

• Post-letting bid analysis

• Forecasting future costs and market conditions

• Continuous improvement based on awarded contract data

• Identification of systematic estimating biases and trends

APPENDIX D – AWARD TABULATION & SELECTION SHEET Page 3 of 8

4. DESIRED FUNCTIONAL CAPABILITIES

SCDOT is interested in understanding whether solutions can provide one or more of the following capabilities.

4.1. AI-Assisted Cost Prediction

• Predict project costs based on project characteristics

• Generate planning-level cost estimates

• Generate conceptual project budgets

• Forecast probable project costs

4.2. AI-Assisted Quantity Estimation

• Generate preliminary quantity estimates

• Estimate quantities using limited design information

• Leverage historical project relationships and benchmarks

4.3. AI-Assisted Pay Item Recommendations

• Recommend appropriate pay items

• Identify missing pay items

• Compare projects to historical project libraries

4.4. AI-Assisted Unit Price Recommendations

• Recommend unit prices using historical data

• Consider regional pricing differences

• Consider quantity-based pricing effects

• Incorporate inflation and market conditions

4.5. Risk and Uncertainty Analysis

• Generate confidence intervals

• Produce low, most likely, and high estimate ranges

• Identify major cost drivers

• Quantify estimate uncertainty

• Support contingency development

• Support risk-based cost estimating

4.6. Estimate Review and Quality Assurance

• Detect anomalies and outliers

• Validate assumptions

• Compare estimates against historical norms

• Support estimate review workflows

APPENDIX D – AWARD TABULATION & SELECTION SHEET Page 4 of 8

5. HUMAN OVERSIGHT AND EXPLAINABILITY

SCDOT is particularly interested in solutions that operate within a human-in-the-loop framework.

Respondents should describe how their solution:

• Supports engineering judgment and decision-making

• Provides explainable recommendations

• Identifies underlying data sources

• Documents assumptions

• Provides confidence scores

• Allows engineers to modify recommendations

• Allows engineers to override recommendations

• Maintains audit trails and decision histories

• Documents rationale for estimate changes

SCDOT is not seeking fully autonomous estimate generation and anticipates that professional engineering review and approval will remain an integral component of the estimating process.

6. DATA ENVIRONMENT

SCDOT maintains data that may include:

• Historical bid tabulations

• Awarded contract information

• Pay item databases

• Quantity histories

• Unit price histories

• Project characteristics

• Route classifications

• Geographic information

• District information

• Project categories

• Construction performance information

Respondents should describe:

• Minimum data requirements

• Recommended historical data volumes

• Data formatting requirements

• Data cleansing requirements

• Data quality expectations

• Training data requirements

• Ongoing maintenance requirements

• Data governance considerations

APPENDIX D – AWARD TABULATION & SELECTION SHEET Page 5 of 8

7. INTEGRATION CONSIDERATIONS

Respondents should describe integration capabilities with:

• AASHTOWare Project

• AASHTOWare Estimation

• Enterprise data warehouses

• SQL Server databases

• GIS platforms

• Business intelligence platforms

• REST APIs

• Microsoft Azure environments

Please describe:

• Integration architecture

• Data exchange methods

• API capabilities

• Import/export capabilities

• Cloud deployment options

• On-premises deployment options

• Hybrid deployment options

8. INFORMATION REQUESTED

8.1. Organizational Experience

Please describe:

• Your organization

• Relevant transportation industry experience

• State DOT experience

• Public-sector implementations

• Relevant client references

• Solution Overview

Please describe:

• Solution capabilities

• Core technologies utilized

• AI, machine learning, predictive analytics, or statistical methodologies employed

• Current maturity level of the solution

• Transportation Use Cases

Please identify:

• Transportation estimating use cases currently supported

• Most successful implementations

• Use cases that provided the greatest value

• Use cases where expected benefits were not realized

APPENDIX D – AWARD TABULATION & SELECTION SHEET Page 6 of 8

8.2. Accuracy and Performance

Please provide information regarding:

• Demonstrated accuracy improvements

• Methods used to measure accuracy

• Confidence interval generation

• Forecasting capabilities

• Performance metrics and benchmarks

8.3. Challenges and Risks

SCDOT is particularly interested in understanding implementation and operational challenges.

Please describe:

8.3.1. Organizational Challenges

• Change management considerations

• User adoption challenges

• Training requirements

• Staffing impacts

8.3.2. Data Challenges

• Data quality concerns

• Data availability concerns

• Historical data limitations

• Data governance considerations

8.3.3. Technical Challenges

• System integration complexity

• Infrastructure requirements

• Scalability concerns

• Model maintenance requirements

8.3.4. AI-Specific Risks

• Hallucinations

• Bias

• Explainability limitations

• Model drift

• Overreliance on automated recommendations

• Regulatory or legal considerations

8.3.5. Lessons Learned

• Common implementation pitfalls

• Critical success factors

• Recommended pilot approaches

• Lessons learned from prior deployments

APPENDIX D – AWARD TABULATION & SELECTION SHEET Page 7 of 8

8.4. Governance and Security

Please describe:

• Human review processes

• Audit capabilities

• AI governance controls

• Data governance controls

• Cybersecurity architecture

Security certifications and compliance standards

Please identify alignment with standards such as:

• NIST Cybersecurity Framework

• NIST AI Risk Management Framework

• SOC 2

• ISO 27001

• FedRAMP (if applicable)

8.5. Implementation Approach

Please provide information regarding:

• Typical implementation timelines

• Recommended pilot approaches

• Staffing requirements

• Training requirements

• Time to achieve measurable value

• Recommended deployment strategies

9. COST INFORMATION REQUESTED

SCDOT is not requesting formal pricing proposals; however, respondents are requested to provide budgetary planning information and estimated cost ranges.

Please provide estimated ranges for:

9.1. Proof of Concept

• Typical scope

• Duration

• Estimated cost range

9.2. Pilot Deployment

• Typical scope

• Duration

• Estimated cost range

9.3. Enterprise Implementation

• Software licensing

• Professional services

• Data preparation

APPENDIX D – AWARD TABULATION & SELECTION SHEET Page 8 of 8

• System integration

• Training

• Change management

9.4. Ongoing Operations

• Annual licensing

• Hosting costs

• Support and maintenance

• Model maintenance

• Additional services

Please identify primary cost drivers and factors that influence implementation costs.

10. STRATEGIC QUESTIONS

SCDOT seeks industry input regarding the following:

• What transportation estimating use cases provide the greatest return on investment?

• What organizational capabilities should be established before implementing AI-assisted estimating?

• What level of data maturity is required for success?

• What implementation approach would you recommend for a state DOT?

• What common mistakes should agencies avoid?

• What measurable benefits have been achieved by transportation agencies?

• What emerging capabilities should agencies monitor over the next three to five years?

• Which capabilities are currently immature or unlikely to provide near-term value?

11. DESIRED OUTCOMES

SCDOT seeks to understand whether AI-assisted solutions can:

• Improve confidence in planning-level estimates

• Improve confidence in preliminary design estimates

• Improve consistency across districts and programs

• Reduce estimate preparation effort

• Reduce estimate review effort

• Improve estimate transparency and defensibility

• Improve forecasting of project funding needs

• Identify systematic underestimation or overestimation trends

• Improve estimate quality and accuracy

• Support risk-informed decision-making

12. DISCLAIMER

This RFI is issued solely for information gathering and planning purposes. Information obtained through this process may be used to inform future pilot projects, proof-of-concept activities, procurement strategies, or technology evaluations.

SCDOT reserves the right to pursue, modify, or discontinue any future activities related to this subject matter.