AI‑powered analytics platform built for commercial real‑estate investors, announced on May 12 2026 that it achieved top‑tier results on four independent artificial‑intelligence benchmarks—DRACO, SpreadsheetBench Verified, BullshitBench, and GAIA—outperforming rivals from Google, OpenAI, Meta and other leading vendors.
Leni’s latest benchmark scores signal a shift from hype‑driven model talk to performance‑focused AI engineering. The company, founded in 2022 and backed by $8.5 million in venture funding, positions its platform as a secure, accuracy‑driven alternative to generic large‑language‑model (LLM) services that dominate the cloud AI market.
What the technology does
At its core, Leni combines an “agentic” AI engine with a Universal Data Model (UDM) that standardizes multifamily‑real‑estate data across PDFs, spreadsheets, and core property‑management systems. The UDM, developed with input from MIT alumni, Greystar, EY and Geoffrey Hinton’s Vector Institute, creates a common schema that eliminates the data silos that have long plagued the sector. By abstracting the data layer, Leni can ingest raw source files and output decision‑ready analytics—valuation models, risk assessments, and lease‑optimization recommendations—without requiring customers to build and maintain their own AI infrastructure.
Why the announcement matters
The four benchmarks where Leni excelled each test a different facet of real‑world AI utility:
- DRACO (Perplexity AI & Harvard) – measures depth of research. Leni’s 71.6 % score placed it ahead of Google’s and OpenAI’s deep‑research offerings.
- SpreadsheetBench Verified – grades performance on hundreds of spreadsheet tasks; Leni ranked in the global top two, correctly completing 365 of 400 tasks.
- BullshitBench v2 – evaluates an AI’s ability to reject nonsensical premises; Leni caught 98 % of fabricated statements, the highest among 142 public models.
- GAIA (Meta & HuggingFace) – tests multi‑step task execution without early‑stage errors; Leni posted a 77.0 % validation score, beating Genspark, Manus and OpenAI Deep Research.
These results matter because they address a documented gap between AI promise and operational reliability. An EY survey released in October 2025 found that 99 % of enterprises experienced AI‑related financial losses, averaging $4.4 million per company. In commercial real‑estate, where deal margins can be razor‑thin, the cost of inaccurate analytics is amplified.
Industry impact
Leni’s performance challenges the prevailing narrative that generic LLMs are sufficient for enterprise workloads. While Google Cloud, Microsoft Azure, and Amazon Bedrock continue to push “foundation models” as one‑size‑fits‑all solutions, Leni demonstrates that domain‑specific data models and purpose‑built pipelines can deliver higher fidelity with lower risk. For enterprise marketing teams, the implication is clear: AI‑driven content and campaign analytics that rely on generic models may need a supplemental layer of industry‑specific validation to avoid costly missteps.
Competitive comparison
Compared with OpenAI’s ChatGPT Enterprise, which offers robust security but still depends on a broad‑scope LLM, Leni’s architecture separates the “harness” (the orchestration and data integration layer) from the underlying model. This modularity lets customers swap out the inference engine while retaining a consistent data backbone—a flexibility not offered by most cloud AI platforms. Moreover, Leni’s focus on “trusted, verifiable output” aligns with emerging regulatory expectations around AI transparency, a concern that Gartner predicts will drive a 30 % increase in AI governance spending by 2027.
Voices from the field
Scott Jones, VP of IT at Ram Realty Advisors, described the platform as “faster and easier,” noting that teams can now trust data that flows directly from source systems into analytical dashboards. Leni’s Head of Industry Strategy, Marcio Sahade, emphasized that the benchmarks measure the very gap that stalls AI adoption: the ability to produce finished, reliable work rather than plausible‑sounding text.
Future outlook
With a portfolio of assets under management exceeding $40 billion, Leni is poised to expand beyond multifamily real‑estate into broader CRE segments such as office and industrial properties. The company’s next roadmap includes tighter integration with leading AI cloud ecosystems—potentially offering hybrid deployment options on Azure and Google Cloud—while preserving its proprietary UDM.
Market Landscape
The AI infrastructure market is currently dominated by a handful of cloud providers, yet analysts from Forrester project that niche AI platforms with vertical specialization will capture 12 % of the enterprise AI spend by 2028. In commercial real‑estate, the JLL 2025 Global Real‑Estate Technology Survey reported that 92 % of firms have piloted AI, but only 5 % have met all strategic objectives. Leni’s benchmark success directly addresses the performance shortfall that has hampered broader adoption.
Top Insights
- Leni’s benchmark scores place it ahead of major cloud AI offerings in research depth, spreadsheet accuracy, nonsense detection, and multi‑step task execution.
- The platform’s Universal Data Model standardizes real‑estate data, eliminating silos and enabling secure, model‑agnostic automation.
- Enterprise teams can reduce AI‑related financial risk, which EY estimates averages $4.4 million per company annually.
- Leni’s modular architecture offers flexibility to integrate with existing AI cloud services while maintaining domain‑specific reliability.
- Adoption of specialized AI platforms like Leni is expected to grow as governance and trust become central to enterprise AI strategies.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
