REAL ESTATE NEWS

The most useful application of artificial intelligence in retail real estate may be its ability to connect work that has traditionally been handled in separate steps: reviewing a public solicitation, testing development options, pressure-testing financial assumptions, visualizing a concept and preparing leasing materials.

That was the premise of a presentation by Mateo Romano, co-founder of Adaptig.ai, at ICSC@Western. Romano showed how retail owners and developers can use a combination of free or widely available AI tools to move through that process faster.

For commercial real estate investors, the appeal is not simply faster document summaries or polished renderings. It is the potential to evaluate more alternatives earlier in the process, identify issues in an RFQ, test how shifts in rents, vacancy, costs and cap rates affect a project, and carry the same project information into leasing and marketing.

The approach still depends on people to supply, test and validate the business assumptions. But Romano's presentation made the case that AI can reduce the time it takes to get from raw information to a decision-ready development strategy.

The workflow is already producing measurable results for NewMark Merrill Companies, where Ross Carpenter introduced Romano. In an August 2026 announcement, the retail owner, developer and manager said a 12-week AI Pioneer Program with Adaptig produced 53 employee-built AI solutions and 3,783 hours of annual recovered capacity.

NewMark Merrill said acquisition workflows that had required roughly eight hours of manual document review were reduced to less than an hour. The company also said property management teams used AI to summarize security patrol reports and flag exceptions, while its accounting team used the technology to automate CAM reconciliations.

Start With The Source Material

Romano's demonstration began with a 27-page city request for qualifications, or RFQ—the type of document that can shape a project long before a development team begins detailed underwriting.

Using Google's NotebookLM, he uploaded the RFQ and generated a summary, then used the platform to ask questions about the document. The material could also be converted into other formats, including an audio discussion, presentation, video or infographic.

The objective was not merely to condense the document. It was to make the RFQ's requirements, site restrictions, advantages and stated vision easier for different people on the team to absorb and use.

A senior executive may want a one-page overview. A development team member may find a presentation or an audio discussion more useful. An infographic can provide a visual reference for the project's requirements before the team begins considering what to build.

That early step matters because the quality of the information flowing through the rest of the process depends on the source material and the instructions given to the AI.

Romano stressed that prompts should establish the role the system is being asked to play, the objective, the relevant context and the desired output. A prompt asking AI to act as a real estate development analyst, for example, will produce a different response than one asking it to approach the same material as an architect, designer or leasing executive.

He also showed how teams can prompt AI tools to ask follow-up questions and request feedback. That creates a more iterative process, rather than relying on a single output that may not reflect the team's full objectives or assumptions.

Turn Requirements Into Scenarios

Once the RFQ has been distilled, the next question is more consequential: What can the site support, and which development alternatives are worth pursuing?

Romano demonstrated how the initial analysis could generate three quantified development scenarios based on specified business assumptions. The AI can be prompted to work through potential alternatives as a development analyst, producing options that a team can then evaluate against the project's requirements and investment criteria.

That is where AI can help a deal team explore possibilities more quickly. Rather than replacing traditional analysis, the workflow gives the team a faster starting point for comparing alternatives and deciding which ones deserve deeper attention.

The scenarios can then become inputs for visualization. Romano used ChatGPT to show how development assumptions and the initial project infographic could be translated into conceptual renderings. By changing the prompt, the AI could be directed to approach the task as an architect or designer.

The result is not a final design. It is a set of visual alternatives that can help a team assess how different development concepts may fit the opportunity.

That creates a feedback loop between the financial and physical sides of the project. A team can compare each scenario against explicit criteria, revise the assumptions and continue refining the analysis without rebuilding the process from scratch.

Test The Financial Assumptions

The workflow becomes more relevant to investors when it reaches financial sensitivity analysis.

Romano showed how information from the prior steps could be moved into Lovable to create an interactive dashboard. The dashboard allowed users to adjust variables including costs, rents, vacancy, cap rates and participation, then see how changes in those assumptions affected the project.

For developers and investors, that distinction is critical: an impressive AI output is not the same as a useful decision-making tool. The value is not in accepting an AI-generated development scenario as a conclusion. It is in using the technology to test the assumptions behind it more efficiently.

A team can use the workflow to explore how a different rent assumption, construction cost, vacancy level or cap rate changes the outcome. That can help identify which variables deserve the most scrutiny before a project moves forward.

The underwriting judgment remains with the people making the investment decision. AI can speed the work of organizing inputs, generating alternatives and showing the effects of changing assumptions, but it cannot determine whether those assumptions are realistic or whether a project meets a firm's return requirements.

Carry The Work Into Leasing

Romano's demonstration did not stop with development analysis and financial testing. The same underlying project information can also be used to support leasing.

Using Gamma, a team can generate presentation materials by prompting the AI to act as a retail leasing marketing director. Those materials can be adapted for tenants and brokers and used to create a master brochure, fact sheets and storefront maps.

This lets teams repurpose work already done during the development stage rather than recreating the project narrative for each audience. The same RFQ requirements, development assumptions and project positioning can carry through to materials used in the market.

Romano also demonstrated the use of Grok Imagine to create short videos depicting a proposed shopping center. The videos could show different views or positioning concepts, such as a food-centered gathering place or a grocery-anchored neighborhood center designed to encourage more than a single shopping trip.

The more context a team provides about the property's intended role, Romano said, the more consistently the resulting materials can communicate that vision.

The broader lesson from the presentation was that individual AI tools matter less than the workflow connecting them. Romano's process moved from an RFQ to an infographic that retained the source requirements, then to quantified development scenarios, conceptual renderings, financial sensitivity analysis and tenant-facing materials.

Each step produced information that fed into the next. For retail real estate teams, that may be the most immediate opportunity: not a wholesale replacement of existing deal processes, but a faster and more connected route from early diligence to development decisions and leasing strategy.

Check back with GlobeSt.com for more coverage from the ICSC@Western event last week and click the stories below for what you might have missed.

CRE Leaders Say Sponsors Must Put Women in Revenue-Generating Roles

How Personal Energy Helped Tony Giordano Close $125M in Deals

Longer Approval Timelines Put Retail Development Returns At Risk

Primestor CEO Says CRE Must Move Beyond Talk on Women's Advancement

Retail Developers Rely on AI and Preleasing to Make Projects Pencil

Higher Yields Create New Buying Opportunity in Retail

Developers Target Underused Retail Land for Housing and Mixed-Use Projects

Retail Growth Shifts To Suburbs And Smaller Markets

Source: GlobeSt/ALM