Bryckel AI Review: CRE Lease Agents Inside Claude and ChatGPT

Bryckel AI is a commercial real estate document intelligence company that has made an unusual bet: it no longer asks customers to log into its software. Instead of running lease work inside a standalone platform, Bryckel rebuilt its product around the Model Context Protocol (MCP), the open standard that lets outside tools plug directly into AI assistants. Its CRE agents now run inside Claude and ChatGPT, while the leases themselves stay in the Microsoft 365 or Google folders where most owners, operators and brokers already keep them. An analyst asks for an abstract, a clause report or a stacking plan in the assistant they already use, and Bryckel’s agents do the specialist work underneath.

Founded in 2023 by CEO Sienam Ahuja and based in Mountain View, California, Bryckel is still a seed-stage company, with roughly $120,000 to $125,000 in disclosed funding according to CB Insights and Stanford’s CodeX TechIndex. That makes it one of the smallest vendors we score. It also makes the strategy easier to understand: a team this size cannot win by building a heavier platform than Prophia or MRI, so it is betting that the platform layer itself is losing value. For a CRE principal, the practical question is whether Bryckel’s agents deliver enough specialized, repeatable workflow to justify a contract, or whether they amount to well-written prompts on top of models the firm already licenses. Our updated view is that the answer has moved in Bryckel’s favor since our first review, although the long-term moat question remains open.

Updated October 14, 2026. We first reviewed Bryckel in August 2026 at 66/100. This version reflects the company’s MCP rebuild and a recorded product demo, featured by the CRE AI newsletter PLOT, in which a single Bryckel agent carried one set of retail leases through four linked workflows. Two dimensions moved: Integration and Workflow Fit rose from 8 to 9, and Innovation and Roadmap rose from 7 to 8. The composite moves from 66 to 68. Pricing, support and market reputation did not change, because nothing public about them has.

What Bryckel AI does and how it works

Bryckel’s product is a library of CRE-trained agents exposed to AI assistants through an MCP connector. Once a firm’s IT team approves the connector, an analyst working in Claude or ChatGPT can call on those agents the same way they would ask the assistant any other question. The agents read lease documents from the firm’s existing property folders, apply Bryckel’s predefined workflows and CRE rules, and return their outputs to those same folders. The general-purpose model supplies the conversation. Bryckel supplies the process, the extraction rules and the domain judgment about what matters in a commercial lease.

That division of labor is the heart of the product. A general assistant asked to abstract a lease will improvise a format, skip clauses it does not recognize as important and vary its output from one lease to the next. Bryckel’s agents follow a defined abstraction structure that captures clauses, rent schedules, critical dates and use restrictions, and each extracted term carries a reference back to its place in the source document so a reviewer can confirm it quickly. The agents also flag potential inconsistencies in the file, the kind of conflict between a base lease and a later amendment that a tired reviewer can easily miss on the fortieth document of the day.

The practical change from Bryckel’s original model is where the work happens. The company began as conventional proptech: customers moved documents into its platform, ran workflows there and exported the results to colleagues. Bryckel says the technology worked but adoption lagged, because documents were already organized by property in OneDrive or SharePoint, financial data lived in Yardi or MRI, and every new platform asked busy employees to change habits. The company calls this platform friction, and the MCP rebuild is its answer. Rather than persuading employees to adopt another destination, Bryckel now rides on the AI assistant they are already being asked to use. More detail is available at bryckel.ai.

The retail leasing workflow, end to end

The clearest demonstration of Bryckel’s capability is a retail leasing sequence the company showed in its recorded demo. It begins with lease abstraction across a shopping center’s tenant roster. Those abstracts feed a portfolio use report showing which tenants hold which permitted uses, exclusives and restrictions. From there, the agent builds a stacking plan of the center and runs a gap analysis against the existing tenant mix to identify categories that are missing or underweighted. The final step screens prospective tenants for the vacant space, testing each candidate’s merchandising fit against both the gap analysis and the restrictions already written into existing leases, and checking the prospect against the surrounding trade area.

Each step is useful on its own, but the chain is what earned the rescore. A leasing team that once needed a lease administrator, a leasing analyst and a broker’s research desk to move from “what do our leases allow” to “who should we call about this vacancy” can now run that sequence in a single conversation, with every output traceable to the documents underneath. Lease abstraction has become table stakes across this category. Carrying the same document intelligence through a multi-step leasing decision is not, and it is the strongest evidence that Bryckel is building workflow rather than just extraction.

Beyond leasing: the rest of the agent library

Bryckel’s broader library extends into estoppel review, lease risk review, portfolio onboarding, CAM and real estate tax reconciliations, market surveys and LOI analysis. The company packages much of this into two named pipelines for enterprise firms. Abstract 360 brings lease abstraction and portfolio intelligence in house and is positioned as a replacement for outsourced abstraction vendors and standalone abstraction software. Clause Protect is a CRE lease review engine positioned as a substitute for outside counsel. Bryckel also says it can deploy on a customer’s own AI model, so documents never leave the firm’s environment, which matters to owners whose counsel will not allow leases to be processed on a vendor’s servers.

We would read the outside counsel positioning with care. A clause review engine can catch deviations from a firm’s standard positions quickly and consistently, which is genuinely valuable, but it carries no professional liability and should not be the last reader of a lease with real money at stake. The more defensible framing is that Clause Protect reduces how much routine review time a firm buys from counsel and concentrates that time on the clauses that actually need a lawyer.

9AI Framework: the score, dimension by dimension

CRE Relevance: 8/10

Bryckel is built for commercial real estate from end to end. Its agents understand concepts that generic document tools stumble over, such as base year expense stops, triple net structures, ROFO and ROFR rights and CAM reconciliation, and its workflow library maps to real CRE job functions: lease administration, leasing, asset management and acquisitions. The retail leasing chain in particular reflects how centers are actually re-tenanted. We held the score at 8 rather than 9 because Bryckel brings workflow expertise rather than proprietary CRE data. It knows how to read a lease, but it does not know what comparable space is leasing for. In practice: an asset manager does not need to explain what a percentage rent breakpoint is before asking about one.

Data Quality and Sources: 7/10

Bryckel does not sell market data. Its outputs are only as good as the leases, amendments and estoppels a firm feeds it, so poor scans, missing amendments and incomplete data rooms produce incomplete abstracts. The source references attached to extracted terms are a real strength, because they make every number auditable. The trade area check in the leasing workflow draws on outside information, but Bryckel has not published which sources it uses, so we score data quality on the document layer alone. In practice: clean, complete lease files produce reliable abstracts, while a disorganized data room produces gaps that can look deceptively confident.

Ease of Adoption: 8/10

For the end user, adoption is about as easy as CRE software gets, because there is no new interface to learn: the interface is Claude or ChatGPT. The cost has moved upstream. Enterprise connectors still require security review, permission mapping and governance sign-off, and Bryckel has acknowledged that IT approval at large brokerages is already slowing deals with firms that want the product. For a lean owner or operator with one decision maker, deployment can take days. For a national brokerage, the approval queue may be the longest part of the project. That trade is why the score holds at 8. In practice: your analysts will not need training, but your IT team will need a security review on its calendar.

Output Accuracy: 7/10

Bryckel runs on frontier language models, so it inherits both their strength in reading long legal documents and their occasional tendency to misread a figure or miss a cross reference. Two features keep the score solid: the defined abstraction structure, which removes much of the variance of free-form prompting, and the source references, which let a reviewer confirm any term in seconds. Chained workflows raise the stakes, however, because an error in an early abstract can carry into the stacking plan, the gap analysis and the prospect list. Firms should verify abstracts before relying on anything built from them. In practice: expect a strong first draft that a lease administrator can check far faster than they could have built it.

Integration and Workflow Fit: 9/10 (up from 8)

This is where Bryckel improved most. Its agents now run through MCP inside both Claude and ChatGPT, with inputs and outputs tied to the property folders a firm already maintains in Microsoft 365 or Google Workspace. Nothing has to be migrated, and deliverables land where colleagues, brokers and attorneys already look for them. Bryckel is also listed in Rent Manager’s integration marketplace. The remaining gap is the accounting system of record: financial data still lives in Yardi, MRI and similar platforms, and buyers should confirm exactly how abstracted data reaches those systems rather than assuming it flows automatically. In practice: the work happens inside tools your team already opens every day.

Pricing Transparency: 4/10

Bryckel still does not publish pricing. Third-party directories list it as a subscription starting around $1,000, but buyers should expect a custom quote shaped by document volume, the workflows they need and the deployment model. A free trial has been offered so teams can test abstraction on their own leases. The MCP architecture also changes what a buyer is paying for. Because the customer’s own Claude or ChatGPT license supplies the language model, the firm is effectively paying for intelligence twice: once for the assistant and once for Bryckel’s workflows on top of it. That is a reasonable arrangement, but it should be reflected in the quote. In practice: ask for pricing per lease abstracted and per workflow, then compare it with what you pay your current abstraction vendor.

Support and Reliability: 5/10

Bryckel remains a seed-stage company with a small team, which means enterprise buyers are relying on founder responsiveness rather than a customer success organization. The MCP model helps on one front, because the conversational layer runs on Anthropic, OpenAI and Microsoft infrastructure rather than a startup’s servers. It adds a dependency on another, because changes to how those platforms handle connectors can affect Bryckel’s product without notice. The company has publicly claimed more than 200 enterprise users with no churn, a figure we cannot independently verify. In practice: put response times in writing, and keep your source documents and abstracts in your own folders, which this architecture makes easy.

Innovation and Roadmap: 8/10 (up from 7)

Bryckel has moved from single-task abstraction to agents that carry the same document intelligence through multi-step workflows, and it has rebuilt its delivery model around a standard most of this category has not yet adopted. The retail leasing chain is a real step beyond what we saw in August. The roadmap risk is capital: a company with roughly $120,000 in disclosed funding has to execute on a thin budget against vendors that have raised far more. In practice: Bryckel is shipping faster than its funding would suggest, and it is building with the platform shift rather than against it.

Market Reputation: 5/10

Recognition is growing. Bryckel has been featured in CRE AI newsletters and demo series, has pitched at industry AI events, and its founder has been quoted in trade coverage on why proptech gets dropped from tech stacks. It has not yet announced institutional customers by name, and its disclosed funding remains among the smallest in our index. In practice: choosing Bryckel still makes your firm an early adopter, with the negotiating leverage and the vendor risk that come with that.

Who should use Bryckel AI

Bryckel fits best at owner-operators, retail landlords and lean asset management teams that already work in Claude, ChatGPT or Microsoft 365 and want CRE-specific workflows without adopting another platform. Retail owners with active re-tenanting needs are the strongest match, because the chain from abstraction to tenant prospecting maps directly onto their work. Acquisition teams facing a short diligence window on a lease-heavy asset will get real value from bulk abstraction with source references, as will firms that currently pay an outsourced vendor per lease and want that work back in house. Lease administrators handling estoppels, CAM reconciliations and critical date tracking across a growing portfolio round out the natural buyer set.

Who should look elsewhere

Global REITs and large brokerages with strict vendor requirements may find Bryckel too young, and should expect its connector to spend time in an IT approval queue. Firms that need a full lease administration or accounting platform should look at products built for that job, because Bryckel is an intelligence layer, not a system of record. Organizations whose policies prohibit processing leases through general-purpose AI assistants will find the MCP model a poor fit unless Bryckel deploys on a model they control. Firms looking for external market data, such as rent comparables or tenant credit, will need a separate source.

Pricing, cost structure and ROI

Bryckel’s architecture produces an unusual cost structure, and buyers should understand it before negotiating. A traditional proptech platform pays for its own compute and hosting and passes those costs through its subscription. Bryckel’s agents run on the customer’s own AI model and inside the customer’s document environment, so the heaviest variable cost of an AI product, model usage, is largely carried by a license the customer already pays for. That should allow Bryckel to run lean, and it gives buyers a fair basis to push back on any quote that appears to include compute.

The pricing anchor shifts as a result. When an analyst can watch Claude or ChatGPT do the reading, it becomes hard to justify a price based on software seats. Bryckel’s own positioning points to the right comparison: what a firm currently spends on outsourced abstraction and on routine outside counsel review. That is the figure to hold any quote against.

The labor math is straightforward. Manual abstraction of a long commercial lease commonly takes two to four hours. If Bryckel reduces that to about 30 minutes of verification, a firm handling 200 leases a year saves roughly 300 to 700 analyst hours. At $75 an hour, that is about $22,500 to $52,500 in direct labor, before counting downstream work such as stacking plans and gap analyses that the agents also produce. Firms that outsource abstraction should run the same comparison against their per-lease vendor fees. Budget time for human verification, and do not count the savings until a pilot on your own leases confirms them.

The bigger question: does CRE software still need its own interface?

Bryckel’s pivot is a bet on a broader thesis. As general-purpose assistants become the place where knowledge work happens, specialist software stops being a destination and becomes a capability layer underneath. If that is right, the adoption problem that has sunk many proptech products, getting busy professionals to open one more application, largely disappears. The burden moves to security review and governance, which IT departments are already organizing around AI assistants anyway.

There is a harder implication that every vendor in this category will eventually face. When the assistant is the interface, workflow logic becomes the easiest part of the stack to copy. A firm with a capable operations team can now encode its own abstraction format, clause rules and reporting templates as instructions for Claude or ChatGPT and connect them to its own document folders. What stays difficult to replicate is proprietary data, ownership of the system of record, and institutional trust in the form of security certifications, audit trails and a track record. Bryckel today competes mainly on workflow design and CRE judgment. That is enough to win lean buyers now, but its long-term position depends on building something a well-staffed customer cannot assemble alone, such as anonymized benchmarks drawn from the leases it processes or a review engine with a documented accuracy record.

The lesson applies well beyond Bryckel. When evaluating any CRE AI tool, ask which parts of the product your own team could rebuild with an assistant and a well-written set of instructions, and which parts depend on data or accountability the vendor uniquely holds. The second category is what deserves your budget.

Competitive landscape

Bryckel competes in one of the most crowded corners of CRE AI. Prophia (BestCRE score: 94) remains the category leader, with a mature platform, deeper structured data and an enterprise track record Bryckel cannot yet match. Findable (87) is strong on document organization and compliance, particularly in European markets. RETS AI (86) and Wilson AI (82) offer specialized CRE abstraction with larger support teams and clearer pricing. MRI Software AI (76) brings native intelligence inside an established property management ecosystem, which matters to firms already running on MRI. Generalist contract tools such as Evisort (70) offer strong extraction without CRE-specific tuning.

Bryckel’s differentiation is architectural. Prophia and MRI ask users to work inside their platforms, while Bryckel goes to where users already are. Its most underrated competitor may be the buyer’s own AI assistant: as Claude and ChatGPT improve at reading long documents, firms will increasingly compare Bryckel not with Prophia but with what their own team could build. Bryckel’s answer is that predefined workflows, CRE rules and chained agents produce consistent, auditable outputs that ad hoc prompting does not, and for most firms that argument holds today.

The bottom line

Bryckel AI is a stronger product than it was two months ago, and a more interesting company. Its MCP rebuild removes the biggest reason proptech tools go unused, and its chained leasing workflow shows the agents doing real CRE work rather than extraction alone. For retail owners, lean asset managers and firms paying per lease for outsourced abstraction, it deserves a pilot on their own documents. Institutional buyers that need vendor scale, published pricing and certified security should still start with Prophia or another established platform.

The more lasting takeaway concerns the category itself. If Bryckel is right that CRE software is becoming a layer beneath the AI assistant, the vendors that endure will be the ones that own something the assistant cannot supply on its own. Bryckel has the workflows. The next year will show whether it can build the rest.

Frequently asked questions