Guest: Yaakov Zar, CEO, Lev

How Loan Brokers Automate Capital Markets

 

Watch how Yaakov Zar of Lev.com built the AI deal platform loan brokers actually use.

 

CRE brokerage software has historically failed at the point of adoption: brokers configure Salesforce, spend $30,000 doing it, and then nobody enters the data. Lev.com, founded by Yaakov Zar in 2019 and backed by $110 million in venture capital from investors including NFX, JLL Spark, First American Title, and Capital One, offers an AI-native deal management platform purpose-built for financing brokers, investment sales teams, and leasing operations. 

The platform automates CRM updates by processing incoming emails and documents, generates offering memoranda from uploaded deal files, maintains a verified lender database of loan preferences and recent transactions, and exposes the entire data layer via API (a programming interface that lets external software talk directly to Lev's data) and MCP connectors (a standard that lets AI assistants like Claude or ChatGPT plug into Lev without any custom coding) so firms can build custom workflows on top of a production-grade foundation.

 

Key Takeaways

• CRM adoption fails because data entry is separated from doing the work. Lev embeds automation into the deal workflow itself, so the system updates as brokers operate rather than requiring a separate logging step.
• The deal index is the core differentiator. Lev ingests multiple conflicting documents - rent rolls, T12s, marketing materials - and produces a single structured source of truth with source attribution per line item, resolving contradictions rather than passing them downstream.
• Lender outreach has been systematized end-to-end. The platform searches a proprietary lender database, sends personalized email blasts through the broker's own Outlook or Gmail account, and automatically parses responses - including term sheets - into a comparative quote matrix.
• Mid-market CRE deals under $20 million can move from document ingestion to a finished OM in minutes. Larger, more complex deals benefit from the same foundation but require additional input on underwriting narrative.
• The platform is now open infrastructure. Lev's API and MCP connectors allow enterprise clients and early-adopter firms to build organization-specific workflows on top of Lev's deal data layer without rebuilding the underlying security, compliance, or document processing stack.
• Vibe-coding a replacement is harder than it looks. Yaakov Zar argues that enterprise readiness - security compliance, edge case handling, data normalization at scale - is the moat that off-the-shelf AI tools and junior vibe-coding sessions cannot replicate.
• The LEV agent is the product roadmap's centerpiece for 2026. Zar is building toward a CRE-specific AI assistant that has direct access to a firm's deals, relationships, and lender data - what he describes as the difference between Claude out of the box and Claude with your world inside it.

 

With more than 30 years and $1.5 billion in transactional experience across commercial real estate, including clients who collectively manage over $45 billion in assets under management, Adam Gower has watched the industry cycle through every iteration of CRM software from paper files to Salesforce implementations that nobody uses. The question Yaakov Zar is answering - how do you build technology that professionals actually adopt while doing high-stakes work - is the same question every serious operator and capital markets team eventually confronts.

 

Why CRE Brokerages Keep Failing at Software - and What Lev.com Is Fixing

The commercial real estate brokerage industry has not suffered from a lack of software options. It has suffered from a specific, recurring failure mode: firms invest heavily in platforms like Salesforce or HubSpot, spend months on configuration, and then watch adoption collapse because the system requires brokers to update records separately from the work itself. As Zar put it during the demo:

"what sort of compliance or engagement are you getting in those systems? You've spent all this effort on something like Salesforce. Are people actually using it while they're doing the work, which is what you want, meaning when I have a deal, I put the deal in, etc. Or are they going in after the fact and trying to get the information in there so they get paid, which is what we usually see happen at a lot of these organizations."

Lev's architectural response to this is to eliminate the manual update step entirely. When a broker sends or receives an email, the platform processes it. When a document is uploaded to a deal, the AI extracts structured data from it automatically. The broker's pipeline stays current because the platform is embedded in the workflow, not adjacent to it.

This approach is particularly relevant for commercial real estate capital markets technology teams that operate at the mid-market level - organizations large enough to need systematic deal tracking, but not large enough to staff a dedicated Salesforce administrator and compliance team.

 

The Deal Index: How Lev Resolves Conflicting CRE Document Data

One of the more technically distinctive features Zar demonstrated is what Lev calls the Deal Index - a structured, source-attributed database of every material fact about a property or transaction, built automatically from uploaded documents.

The problem it addresses is familiar to anyone who has worked on a deal with multiple data sources. A rent roll, a T12, a prior appraisal, and a broker's marketing summary will frequently disagree on the same figures. NOI is the most common point of conflict. If downstream tools - including large language models - simply pull from the first available source, the outputs are unreliable. The OM goes out with the wrong number. The quote matrix is built on bad inputs.

Lev processes each uploaded document separately, identifies every fact claim, maps conflicts, and presents the broker with a reconciled view that shows which source was used for each line item and why. The system defaults to the most authoritative source based on document type and recency, but the broker retains override control. As Zar described it:

"we spent a lot of time and effort creating normalizations around that. So, when you upload the documents, we take all of that and give you this deal index, which is a breakdown of every sort of fact and figure about the property or the deal that matters. And each one, it's broken down by section."

The practical implication is that every downstream output - the OM, the lender pitch, the quote matrix, any data pulled by external AI tools - draws from the same verified source of truth rather than from whatever the AI happened to retrieve first. For CRE due diligence and data management workflows where errors have material consequences, this is structural, not cosmetic.

 

Lender Search, Automated Outreach, and the Quote Matrix

The lender outreach workflow is where Lev's financing broker heritage is most visible and, for that audience, most immediately actionable. The platform maintains a proprietary database of lenders enriched by multiple methods: purchased third-party data, an internal team doing direct outreach, and automated agents that pull LinkedIn profiles for every lender contact every 30 days to verify current employment and track job changes. Recent news mentions and transaction records are also ingested to update loan preference profiles.

When a broker is ready to go to market on a deal, Lev runs a search across this database against the deal's parameters and returns a ranked list of lenders with rich profiles - what they have lent on recently, what they are actively seeking, and what their program preferences are. The broker selects from this list, and Lev generates personalized outreach emails that go through the broker's own connected Outlook or Gmail account. From the lender's perspective, the email looks like it was written and sent by the broker directly.

Responses are then parsed automatically. If a lender sends terms in the body of an email, Lev extracts the structured figures. If terms arrive as a term sheet attachment, Lev parses the document. The output is a formatted quote matrix - all received terms side by side, with the best figure on each line highlighted in green - ready to share with the borrower client or export to Excel.

For capital markets teams currently running this process through a combination of mail merge, Excel trackers, and an analyst who stays late to catch email responses, the operational differential is significant. The commercial real estate loan broker technology use case is Lev's most fully developed offering, and the one with the clearest immediate return on the Lev.com subscription.

 

Lev as Open Infrastructure: API, MCP, and the Enterprise Build Layer

The more strategically significant development Zar discussed is the opening of Lev's platform as a buildable foundation. Alongside the core SaaS product, Lev now offers a documented API and MCP connectors that allow users to access deal data and trigger platform functions from external tools - including Claude, ChatGPT, and custom agent workflows.

The practical use cases Zar described ranged from asking the platform about pipeline status from a mobile device to enterprise clients who want to integrate Lev's OM generation into their own proprietary systems without adopting the full platform. A leasing broker with an existing pipeline tool they prefer can send deal data to Lev via API and receive a formatted OM output without switching platforms.

Zar was direct about the competitive argument here. Building the document processing, lender database, deal normalization, and security compliance stack from scratch with AI coding tools is possible. The question is whether the result will hold up under enterprise security review, handle the edge cases that appear at transaction scale, and be maintained by a team with the engineering depth to keep pace with the underlying model improvements. His framing:

"the sort of enterprise readiness of an end-to-end solution that works in every scenario is hard to build and maintain, and whatever you're doing to Vibe Code, we have 20 people who are, dare I say, better vibe coders than you, better builders than you, and who each have 20 years of software engineering experience."

For CRE sponsors and operators evaluating where to build versus buy in their AI-powered CRE operations stack, that argument merits serious consideration. The cost of building is no longer the barrier it was. The cost of maintaining, securing, and iterating at the pace of current AI development is.

 

Frequently Asked Questions

What types of CRE professionals does Lev.com serve?

Lev began as a platform for commercial real estate financing brokers and has since expanded to serve investment sales brokers, leasing brokers, lending institutions with incoming deal flow, and servicing departments within larger lending organizations. The platform's core deal management and document processing infrastructure is relevant to any CRE professional who works with multiple data sources, manages lender or buyer relationships at volume, and produces pitch materials. Enterprise clients who already have a system of record - including Salesforce implementations - can integrate specific Lev capabilities, such as the OM generator or lender search, via API without replacing their existing stack.

How does Lev handle document conflicts and data accuracy in CRE deals?

When multiple documents are uploaded to a deal - rent rolls, T12s, appraisals, marketing materials - Lev processes each separately and maps conflicts at the field level. The Deal Index presents a reconciled view of every material fact, with source attribution for each line item. The system defaults to the most authoritative source based on document type and recency, flags conflicts for broker review, and allows manual override. This matters because AI-generated outputs - including offering memoranda and lender pitches - are only as accurate as the input data. Lev's normalization layer sits between raw documents and any downstream output, including API queries from external agents.

Can CRE firms integrate Lev with their existing tools and workflows?

Yes. Lev offers a documented API and MCP connectors that allow external tools - including Claude, ChatGPT, and custom agent frameworks - to read deal data and trigger platform functions. The platform also connects to Outlook and Gmail for email-based deal creation (forwarding a deal-related email to a Lev address creates a new deal record automatically) and for lender outreach (emails go through the broker's own email account, not Lev's infrastructure). Enterprise clients with existing CRM investments can use Lev as a component - accessing document processing, OM generation, or lender search via API - rather than replacing their current system of record.

What is the pricing model for Lev.com?

During the recorded conversation, Yaakov Zar referenced a starting subscription of as little as $100 per month per team. Lev serves mid-market brokerages as its primary customer base, with teams ranging from a few principals to 30-40 people, and has also developed enterprise arrangements for larger organizations seeking custom or component-based integrations. Given that the platform eliminates analyst hours spent on email parsing, CRM updates, and lender outreach coordination, the cost-to-value calculation for active financing teams is favorable. Prospective customers can initiate a trial through lev.com, with historical data migration included in the onboarding process.

 

Next Steps

If your firm's deal tracking still runs on spreadsheets or a Salesforce instance nobody updates, the operational gap described in this episode is costing you analyst time and deal visibility. Start with the fundamentals: AI tools for commercial real estate professionals for an overview of where AI is having the most immediate impact on CRE operations.

For sponsors and operators who want to build AI-powered systems across the full deal lifecycle - from investor acquisition to deal management and capital deployment - the GowerCrowd AI Accelerator Program is the structured path: AI Accelerator Program for CRE sponsors.

 

Adam Gower, Ph.D. is the founder of GowerCrowd and one of the most experienced practitioners in commercial real estate capital formation. With more than 30 years and $1.5 billion in transactional experience - including serving as President of a Universal Studios development division overseeing $400 million in projects across Asia Pacific - Adam now helps CRE sponsors build AI-powered systems for investor acquisition, deal management, and capital deployment. His clients collectively manage over $45 billion in assets under management. gowercrowd.com