Platuni is a platform which manages property and supports residential lease operations. Its Lease AI Agent brings document processing, compliance validation, risk scoring, conversational assistance, and lifecycle automation into one lease management workflow, using Amazon Bedrock, Claude Sonnet 4.6, Amazon Textract, Amazon S3, PostgreSQL, FastAPI and Platuni API to manage leases from ingestion through renewal or termination.
Problem: Fragmented workflows made lease management harder to scale
Platuni's residential lease management involves more than storing documents. Teams need to collect lease data, verify required clauses, identify compliance issues, track signatures and deposits, manage renewals, respond to tenant questions, and maintain an audit trail throughout the lease lifecycle.
Platuni needed these functions to work from the same lease record whether the lease was uploaded as a PDF or filled out using a form. The separate processing paths would lead to inconsistent data, duplicate logic, and increased manual review in downstream lease management workflows. Platform also needed structured compliance decisions and lifecycle automation. Lease data had to undergo field validation, risk assessment, recommendations, signing, deposits, insurance, maintenance, renewals, termination, and other operational processes, while human review was available for uncertain or high-risk cases.
Solution: AI-powered lease management on AWS
GoML developed the Platuni Lease AI Agent as an integrated lease management system connecting document extraction, lease creation, compliance validation, risk scoring, AI recommendations, lease Q&A, English translation, lifecycle automation and reporting.
The implementation aligns most closely with GoML’s AI Matic platform with additional document intelligence and compliance capabilities. Amazon Bedrock provides the AI layer, Amazon Textract processes scanned lease documents, Amazon S3 stores lease files and templates, PostgreSQL provides shared persistence, and the Platuni API connects lease data with tenant, property, landlord, and downstream action workflows.
Lease management document extraction
GoML built two lease-entry ways for Platuni that range to the same underlying data model. Platuni can accept an uploaded lease in PDF format or build a lease directly from structured form data.
• Extracts lease data from native and scanned PDFs.
• Uses Amazon Textract for scanned document OCR.
• Uses Amazon Bedrock and Claude for structured field gathering.
• Stores uploaded lease documents to Amazon S3.
• Routes PDF and form-based leases into the same lease record structure.
Lease management data standardization
GoML built a common field structure for Platuni so that extraction, compliance, lifecycle automation along with lease creation operate on consistent data. The canonical schema contains about 70 lease-field definitions shared across core modules.
• Standardizes lease fields across PDF and form-based ingestion.
• Normalizes dates, currency, document types, and conditional fields.
• Captures confidence scores alongside extracted values.
• Preserves previous versions when a lease record changes.
• Uses one canonical field definition across extraction and validation.
Lease management risk scoring
The risk scoring layer translates compliance findings into a single score of 0-100 for quicker review of leases. It weighs important clauses, jurisdiction requirements, policy conflicts, monetary terms and non-standard lease risks.
• Creates1 a 0–100 lease risk score.
• Classifies leases as compliant, review recommended, or non-compliant.
• Weights compliance issues by category and severity.
• Applies stronger penalties when problems repeat.
• Forces non-compliant status for critical mandatory or jurisdiction blockers.
Lease management compliance recommendations
GoML added an AI recommendation layer after compliance validation. Instead of only identifying a lease issue, the system explains what should change and provides contextual recommendations for review.
• Converts compliance flags into suggested fixes.
• Generates recommended clause changes through Amazon Bedrock.
• Adds reasoning and affected fields to each recommendation.
• Stores each recommendation separately for review.
• Supports individual approval or rejection by an administrator.
Lease management Copilot
GoML developed Lease Copilot to help users ask natural-language questions about an individual lease. The assistant receives relevant fields and clauses and passes on targeted context to Amazon Bedrock.
• Answers questions using data from the selected lease.
• Uses keyword, synonym, and fuzzy matching for retrieval.
• Falls back to LLM-based field routing when retrieval confidence is low.
• Returns up to five relevant fields for a query.
• Retains the last five conversation turns for continuity.
Lease management plain-English translation
GoML added a translation workflow that rewrites individual legal clauses into simpler language. The function preserves the original meaning while reducing legal terminology for users without a legal background.
• Rewrites complex lease clauses in plain English.
• Preserves the original clause meaning.
• Avoids adding interpretations or new conditions.
• Keeps translation calls independent from lease records.
• Logs translation activity for audit purposes.
Lease management lifecycle automation
GoML built a lifecycle engine that continuously evaluates lease state and determines the next required action. Around 21 independent workflows cover the operational stages of residential lease management.
• Automates onboarding and signature workflows.
• Tracks deposits, insurance, and utilities.
• Supports maintenance and roommate-change requests.
• Handles renewals, rent increases, and termination workflows.
• Automates move-in and move-out processes.
• Records workflow states, events, and executed actions for auditability.
Lease management human review
GoML designed the lifecycle engine to stop automation when lease information is not reliable enough for downstream action. This keeps low-confidence or blocking compliance issues from progressing automatically.
• Normalizes extraction confidence into a 0–100 scale.
• Flags populated fields below 60% confidence for review.
• Stops lifecycle execution when a blocker is detected.
• Routes uncertain cases to human review.
• Continues automated workflows after validation requirements are cleared.
Lease management reporting
GoML built reporting around the latest compliance results so users can review lease status without interpreting raw system output. Reports combine compliance data with an AI-generated plain-English summary.
• Generates PDF and JSON compliance reports.
• Shows overall compliance status and total flags.
• Lists affected fields, severity levels, and issues.
• Generates plain-English summaries through Amazon Bedrock.
• Keeps reporting available even when AI summary generation fails.
AI Matic delivery
GoML used the Agentic AI blueprint approach as the delivery foundation for Platuni’s lease management system, combining document intelligence, deterministic compliance rules, AI recommendations, conversational lease assistance, and automated lifecycle workflows.
The AWS stack includes Amazon Bedrock and Claude Sonnet 4.6 for AI processing, Amazon Textract for scanned-document extraction, Amazon S3 for lease storage, FastAPI for APIs, PostgreSQL for shared data, and the Platuni API for property data and lifecycle action dispatch.
AI Matic delivery metrics
- Average TTFV: 20 days
Impact
• Around 21 lease lifecycle workflows supported through one automation engine.
• Around 70 lease fields standardized across extraction, building, and compliance.
• 5 compliance validation passes applied to each eligible lease.
• 0–100 risk scoring for faster lease triage.
• 30-minute lease cache and one-hour Copilot session duration for repeated interactions.
About
Before Gen AI and after Gen AI
“With Platuni’s lease management system, GoML connected document intelligence, compliance, risk scoring, conversational AI, and lifecycle automation into one architecture that keeps every lease workflow working from the same data.”
Prashanna Rao, Head of Engineering, GoML
Key takeaways for AI lease management
Common challenges
• Lease data arrives through different document and form formats.
• Compliance issues often depend on relationships between multiple fields.
• Low-confidence extracted data requires human validation before automation.
• Lease operations span signatures, deposits, insurance, maintenance, renewals, and termination.
• Users need simple ways to understand complex lease clauses and compliance issues.
Practical guidance for AI lease management
• Use one canonical lease schema across every processing path.
• Combine deterministic compliance rules with AI where language reasoning adds value.
• Attach confidence scores to extracted lease fields.
• Stop automation when critical data needs human review.
• Keep compliance recommendations separate from compliance decisions.
• Record lifecycle actions and state changes for auditability.
Ready to automate lease management
Build AI-powered lease management for extraction, compliance, risk scoring, and lifecycle automation with GoML AI Matic.
.avif)
