×
![Modal Image]()
We’re excited to congratulate the Cadastral team on its acquisition by Legora.
Earlier this year, Cadastral announced a $9.5 million raise to keep building its vertical AI platform for commercial real estate. It has now joined Legora, bringing that technology into a broader AI-native legal intelligence platform.
For Amplified, the acquisition is particularly meaningful because we were there at the beginning.
Cadastral came to us with deep commercial real estate expertise and a clear product vision. Rather than build an engineering team from scratch, it made Amplified its first product engineering partner. We built the MVP.
As Cadastral raised capital and began building its own engineering team, the relationship evolved. Amplified stayed embedded in a hybrid capacity, working across document intelligence, AI workflows, cloud integrations, permissions, product experience, billing, and operational reliability.
What began as building the first version became a longer partnership: first getting the product into users’ hands, then solving the harder problems that appear as an AI product grows.
Commercial real estate runs on documents: leases, loan agreements, insurance certificates, due diligence files, property records, emails, spreadsheets, and years of historical files. The information inside is fragmented across systems, buried in long contracts, poorly scanned, table-heavy, or written in legal language that needs context to interpret correctly.
Cadastral was not trying to build another chatbot over PDFs. The opportunity was to turn those documents into structured, reviewable, searchable data (dates, parties, clauses, rents, obligations, rights, covenants, compliance requirements) that users could compare and act on across a portfolio.
Turning that information into a reliable product required much more than an AI model. The system around it also had to handle ingestion, OCR, extraction, retrieval, source references, human review, permissions, integrations, billing, and the product experience tying it all together.
Document processing sounds straightforward until real customer data arrives. Files run to hundreds of pages, OCR quality varies, layouts are inconsistent, and jobs fail halfway through. A single unusually large document can consume enough resources to affect everything queued behind it.
Amplified engineers built durable processing that resumes after interruptions, retries safely, preserves intermediate results, recovers from partial failures, and keeps any one file from overwhelming the system. Expensive operations such as OCR, page processing, chunking, and AI calls were handled with particular care. A 200-page lease should be a workload the product expects, not an exception engineers rescue manually.
Processing was only the beginning. Users needed to review extracted data, edit fields, rerun extraction, preserve revisions, and see exactly where each piece of information came from. Amplified contributed structured abstractions, reviewable fields, reusable document views, and surfaces where AI output becomes editable, sortable, filterable, source-linked, and reusable.
Lease abstraction was a natural starting point, but commercial real estate does not stop at leases. Teams also work with credit agreements, certificates of insurance, vendor contracts, loan documents, due diligence packets, deal files, property records, and custom document collections.
As the product expanded, it had to support each new document family without rebuilding the experience. Amplified built reusable document and collection patterns (generalized views, structured abstractions, file-listing and update workflows) so collection-based experiences could support multiple document types on one architecture.
The same applied once documents became structured data. Users needed to compare information across properties and agreements instead of returning to individual PDFs. Amplified contributed table and saved-view workflows: editable columns, column-level filtering, pinned columns, reusable configurations, and persisted views.
In legal and financial workflows, an answer that sounds right is not enough.
People need to know where it came from, so citations and source references became part of the product architecture, not a secondary interface detail.
Amplified worked on source-grounded answer flows, document references, and viewing experiences that take users from an AI answer back to the underlying evidence. A citation had to open the correct source, stay attached to the answer, behave predictably during streamed responses, and keep working as users changed context or came back later.
The AI assistant itself had similar complexity beneath a simple interface.
Behind a simple flow (select documents, ask, read, open the source), the system prepares context from files, folders, buildings, and earlier messages; manages long-running model calls; streams responses; tracks usage; persists the answer; and preserves source references.
Users should not have to understand any of that. They should know what the product is using, what it found, where the answer came from, and what they can do next.
Citations solve only part of the trust problem. The other part is access.
Commercial real estate documents often hold sensitive legal, financial, operational, and compliance information. Sharing an AI conversation cannot automatically grant access to every document behind the answer.
Amplified contributed to permission-aware workflows, role-based access, guarded actions, sharing controls, member management, and interface states that made those rules more predictable.
It gets harder when an answer draws on multiple documents, folders, organizations, workspace roles, or external storage permissions: the system must still verify that the recipient may access the underlying sources.
Enterprise customers already have document systems: Google Drive, SharePoint, Box, Dropbox, OneDrive, and other cloud environments. A useful product works with that reality instead of asking customers to recreate their libraries elsewhere.
Integration means much more than copying files. Folders are renamed, documents move or disappear, permissions change, sync gets interrupted, providers impose rate limits, and large folder structures keep changing while the product tries to stay aligned with them.
Amplified contributed to synchronization and integration work around change detection, selective sync, permission-aware syncing, recovery from interrupted operations, and keeping the platform’s internal document model aligned with external systems.
Email and attachments are another source of important commercial real estate information, so the product also needed to bring them into the same AI workspace with reliable authentication, ingestion, and access patterns.
Document management, AI conversations, collections, notifications, settings, billing, permissions, structured tables, and file workflows all had to feel like parts of the same system.
Amplified contributed to product experience and UI modernization through reusable components, design-system work, page improvements, navigation cleanup, document previews, uploads, file movement, recursive search, attachment flows, and shared state-management patterns.
That work matters because even strong AI is hard to use if people cannot find the right document, see what is selected, preview the source, move between surfaces, or return to earlier work without losing context.
Some of the hardest engineering work is almost invisible to the user. Billing is one example.
AI-native products need to measure actions and usage across subscriptions, credits, invoices, overages, and usage-based pricing. At the same time, technical systems generate retries, duplicate events, delayed events, and events that arrive out of order.
Amplified contributed to usage tracking, AI-action metering, billing and credit workflows, and synchronization between the product and billing systems. A key pattern was idempotency: processing the same event twice should never create duplicate charges, missing records, or inconsistent balances.
Operational reliability created similar challenges. Document processing, OCR, model calls, cloud sync, permission checks, and billing all have different performance profiles, and if they are not isolated and monitored, a spike in one part of the platform can affect everything else.
Amplified worked on background processing, workload separation, monitoring, tracing, testing environments, and release reliability, helping the system identify expensive operations and recover from failures more predictably.
The engineering relationship changed as the company changed. At the beginning, Cadastral brought the domain expertise and product vision, and Amplified provided the team that turned it into the MVP.
After Cadastral raised capital, it began building its own engineering organization. Amplified stayed involved.
The model became hybrid: Cadastral’s internal team and Amplified engineers working across the same product, with our people continuing to own areas, ship features, harden workflows, and solve production problems as requirements, customers, and AI capabilities changed.
That progression is familiar to us: being useful when a company needs to build its first product, then adapting as it develops its own engineering organization, rather than becoming a layer it has to manage around.
For us, that kind of partnership is not about creating dependency. It is about providing the engineering team a company needs at each stage of its growth.
Cadastral went from an ambitious product idea to a company that raised $9.5 million and was acquired by Legora. That achievement belongs to the Cadastral team and the company they built.
For Amplified, it is rewarding to look back at where the relationship began: with industry experts who knew the problem they wanted to solve and trusted us to build the first version with them.
We built the MVP. We stayed involved as the company grew. And along the way, our engineers helped build many of the systems that turned AI capability into a product customers could use, verify, secure, integrate, and rely on.
We’re proud to have played a part in that journey.
