How Fifth Dimension Cuts Investment Memo Drafting from Weeks to 30 Minutes
Fifth Dimension, a global AI platform for commercial real estate asset managers and owner-operators, built a multi-model workflow on Google Cloud using Gemini for large-scale document ingestion and Claude for high-precision reasoning. The platform compressed investment memo drafting from days or weeks to just 30 minutes and achieved 99.9% reliability for multi-hour workflows, driving deals with top-10 U.S. asset managers.
Impact
Days or weeks → 30 minutes
Investment memo drafting time
99.9%
Platform reliability for long-running workflows
300,000+ lines
Monthly code generated via Claude Code
98%
Claude Code suggestion acceptance rate
6x
Projected inference spend growth in 2026
Challenge
Commercial real estate deal workflows require processing hundreds or thousands of documents per transaction, with teams spending days or weeks on investment memo drafting and data extraction tasks that demanded a reliable, multi-model AI architecture capable of running for hours without interruption.
Solution
Fifth Dimension built a multi-model orchestration platform on Vertex AI, using Gemini’s one-million-token context window for large-scale document ingestion and Claude Sonnet/Opus for high-precision data extraction and financial reasoning, compressing complex CRE workflows from weeks to minutes.
Tools & Technologies
What Leaders Say
“The commercial real estate market stands to realize a lot of benefits from generative AI given how document-intensive it is. With Google Cloud we can build tailored solutions that unlock more value for the industry.”
“Vertex AI supports long-running workflows that can sometimes last between 1 and 3 hours and have different models used and swapped out at various phases. This is one of the many ways Vertex AI enables our ‘best model for the job’ mentality.”
“While some employees might have been apprehensive at first, Claude Code proved to be incredibly accessible for our non-IT teammates. We’ve upskilled them, but it really didn’t take a lot of effort to help all of our employees leverage Claude Code to get the latest and greatest insights from sources like BigQuery.”
Sign up to read complete case studies, access detailed metrics, and unlock all use cases.
Full Story
Commercial real estate sits at the intersection of finance and asset management, where deals are defined by documents. A single acquisition can involve hundreds or thousands of contracts, leases, rent rolls, financial models, and legal filings. Before decisions get made, teams spend weeks reading, extracting, and synthesizing this information. Fifth Dimension was founded to eliminate that bottleneck using large language models and entered the market with a platform purpose-built for CRE workflows — acquisition intelligence, asset optimization, reporting, and communication.
The core technical challenge was reliability at scale. Commercial real estate workflows are not quick queries; they are long-running processes that can last between one and three hours, involving massive document datasets and requiring models to switch between tasks throughout the run. Fifth Dimension needed a foundation that could handle this without degrading, and it needed access to the best available models on the market without managing multiple infrastructure boundaries. The team chose Google Cloud and Vertex AI as its single secure environment.
Fifth Dimension built a multi-model orchestration strategy on Vertex AI that assigns tasks to the model best suited for each job. Gemini Flash and Pro models handle the large-context intake work — processing datasets that may span thousands of documents — using their one-million-token context window. Claude Sonnet and Opus models take over for high-precision tasks like calculating rental income adjustments, extracting structured data from complex documents, and generating Excel-based financial models. Claude Haiku handles cost-effective code execution on the fly. Gemini starts the pipeline with large-scale document grouping; Claude performs the subsequent high-performance extraction and reasoning. The switch happens automatically within the workflow. Running everything through Vertex AI means client data never leaves a single secure boundary.
The business impact is concrete. Investment memo drafting, which previously required days or weeks of analyst time, now takes 30 minutes. The platform has signed major deals with top-10 U.S. asset managers and grown to a 50/50 split between U.S. and international clients. Internally, every engineer at Fifth Dimension uses Claude Code for development, generating more than 300,000 lines of code per month with a 98% suggestion acceptance rate. Non-engineering staff, including account managers, use Claude Code connected to BigQuery to run real-time natural language queries against production data.
Fifth Dimension projects a 6x increase in inference spend in 2026 as its customer base scales. The company’s next phase involves internationalization and helping clients embed their own institutional knowledge directly into the platform — building what it calls each customer’s “secret sauce” into an AI-native operating system for real estate.