Title + PSA + FSBO offer workflows · CRM-agnostic

The intelligent document layer for brokerages and FSBO sellers.

Unify deal documents, extracted fields, and transaction memory in one place — without replacing your CRM, flat-fee listing platform, or transaction tool. ClawQL classifies title commitments and buyer offers, routes Schedule B exceptions to human review, and answers diligence questions with grounded citations.

Book a demo Technical docs

Reference packs and demos use synthetic data only — not legal, title, or brokerage advice. Third-party CRM, storage, transaction, and FSBO listing products are not affiliated with ClawQL; integration is as an intelligent document layer, not a CRM or MLS replacement.

Overview

ClawQL for real estate

Residential transactions generate dozens of PDFs per deal — purchase agreements, buyer offers, title commitments, appraisals, HOA disclosures — often split across a brokerage CRM (pipeline), cloud storage (folders), and a transaction platform (e-sign). FSBO sellers get the same PDFs without a coordinator. ClawQL is the intelligent document layer: shipped reference packs for title commitments, PSAs, and FSBO buyer offers demonstrate parse → classify → extract → HITL → vault threading with deal_id metadata. Financed purchases can reuse clawql-lending mortgage tools; Coneshare VDR closes the loop when external parties access sensitive files.

Shipped reference workflows mirror the lending W-2 pattern used for See The Greens LOS — parse, classify, extract, HITL, vault recall — applied to transaction coordinators at brokerages and to FSBO sellers comparing multiple offers without a coordinator seat.

Industry context

The gap brokerages and FSBO sellers share

CRM and flat-fee listing tools compete on pipeline and MLS access — nobody classifies title commitments or compares buyer offers with grounded citations.

Major brokerages standardize on different CRM operating systems — Keller Williams (Command), eXp (BoldTrail/kvCORE), Compass and Anywhere brands (Home Platform), and thousands of high-volume teams on Follow Up Boss — but none of them classify title commitments, extract Schedule B exceptions with citations, or answer cross-deal diligence questions from indexed transaction files. FSBO sellers on Houzeo, Beycome, or DIY MLS listings face the same gap at smaller scale: multiple buyer offers arrive as PDFs with no semantic index. AI investment in 2026 skews toward lead gen, listing copy, and routing; the document layer — where deals actually close — still runs on Drive folders and manual re-reads. Teams and sellers that add intelligent document processing first compress hours per file; those that wait inherit a compounding disadvantage.

Who it's for

Two audiences, one document engine

Same classify → extract → recall pipeline — positioned for transaction coordinators at brokerages and for FSBO sellers comparing offers without a coordinator seat.

Brokerages & transaction teams

Document intelligence alongside Command, BoldTrail, FUB, and Dotloop.

Transaction coordinators re-read PDFs because CRMs track pipeline, not document content. ClawQL adds classify → extract → HITL → vault on top of your existing stack — the same pattern as the shipped title and PSA reference packs.

System	Role
Brokerage CRM	Contacts, pipeline — Command, BoldTrail/kvCORE, Follow Up Boss, Compass, Lofty.
Cloud storage	Transaction folders — Google Drive, Dropbox, SharePoint.
Transaction / e-sign	Dotloop, SkySlope, DocuSign Rooms — forms, signatures, broker compliance.
ClawQL	Classify, extract, HITL, semantic search, vault memory, VDR share.

Title commitment intake: Schedule B HITL with shipped clawql-realestate-title-ingest workflow.
PSA field extraction: Cross-check purchase price vs title policy amount via memory_recall.
Cross-deal memory: Wikilink [[Deal {id}]] notes — stop asking "how did we handle this easement last time?"

Forward this to your TC lead: ClawQL sits on top of Drive and your CRM — when a title commitment lands, we extract Schedule B exceptions with citations and route ambiguous reads to Label Studio. No CRM migration. Synthetic demo in 15 minutes.

FSBO sellers

Understand offers and title docs without re-reading every PDF.

For Sale By Owner sellers receive competing offers and title commitments with no transaction coordinator. ClawQL extracts price, earnest money, and contingencies with citations so you can compare offers side by side — complementary to Houzeo or Beycome for MLS and forms, not a replacement.

System	Role
Flat-fee / FSBO platform	Houzeo, Beycome, ISoldMyHouse — MLS listing, forms, optional coordination.
Title / escrow	Commitment, closing — ClawQL helps you read Schedule B, not replace title insurance.
ClawQL	buyer_offer extract preset, offer comparison, title commitment intake, vault recall.

Compare multiple offers: Extract financing, inspection, appraisal, and sale-of-home contingencies — ask which offer closes soonest.
Title commitment review: Same title pack as brokerages — understand Schedule B exceptions before closing.
Advisor-friendly vault: Attorney or friend helping you sell? Share grounded summaries, not raw PDF hunts.

Selling your home without an agent? When buyer offers arrive, ClawQL reads each PDF and surfaces price, closing date, and contingencies in plain language — "Offer 2 is cash with no appraisal contingency." Same engine brokerages use for title commitments. Per-transaction pricing fits one sale; no $299/mo coordinator seat required.

Forward to your team

One-paragraph pitch

Copy and send this to transaction coordinators, team leads, or tech evaluators before a demo.

Every major brokerage runs a CRM for contacts and pipeline, a transaction tool for e-sign and compliance, and Google Drive or Dropbox for transaction folders — but nothing connects them intelligently. When a title commitment or PSA lands in a deal folder, ClawQL classifies it, extracts purchase price and Schedule B exceptions with citations, routes low-confidence reads to a coordinator for review, and indexes everything for semantic search — so teams stop re-reading the same 200-page diligence folder. Your CRM keeps contacts and compliance; your storage keeps the files; ClawQL is the layer that understands what is in the documents and remembers it across deals. Works alongside Command, BoldTrail, Follow Up Boss, Compass, Dotloop, or SkySlope — no migration required. Live demo uses synthetic data only; early access includes founder-led setup.

Challenges

Problems agents solve in this vertical

These are the operational friction points real-estate transaction samples targets — on top of ClawQL Core search, execute, memory, audit, and the IDP pipeline.

CRM knows the deal — folders hold the files

Brokerage CRMs track contacts and pipeline stages, but Drive folders do not classify, extract, or link back to the deal record. Coordinators manually match filenames to transactions and re-open PDFs for every question — at every franchise.

FSBO sellers compare offers by re-reading PDFs

Multiple buyer offers differ on price, contingencies, and closing dates — Houzeo and flat-fee tools help list, but nothing extracts and compares offer terms with citations. One missed contingency clause costs more than the software.

Schedule B exceptions need human judgment

Title commitments arrive with exceptions that require curative review — not fully automatable. classify_document and extract_document surface Schedule B items; hitl_enqueue_label_studio routes ambiguous exceptions to coordinators (or seller advisors) before they enter the vault.

Cross-deal memory lives in someone's head

"How did we handle this utility easement last time?" has no answer in CRM or storage. memory_recall threads title exceptions, cleared conditions, and party notes across sessions with wikilinks per deal_id.

Platform

Shared ClawQL capabilities

Every vertical package composes these horizontal layers — security, memory, and document intelligence.

Shipped title commitment + PSA + FSBO buyer offer reference workflows
CRM-agnostic — works with Command, BoldTrail, Follow Up Boss, Compass, or custom stacks
classify_document labels: title_commitment, purchase_agreement, buyer_offer, appraisal, hoa_disclosure
extract_document presets: title_commitment, purchase_agreement, buyer_offer (contingencies + PSA fields)
Eight-vendor IDP pipeline — Gotenberg normalization, Stirling PII redaction, Onyx search
Coneshare VDR share links with viewer activity webhooks for external counsel and lenders

Domain tools

Tools from real-estate transaction samples

Planned or shipping MCP tools from modularization v2.1 — registered when the vertical package is enabled via CLAWQL_ENABLE_* or Operator flags.

classify_document()

Route PSA vs buyer_offer vs title commitment vs appraisal before extraction schema selection.

extract_document()

Grounded fields: purchase_price, contingencies (buyer_offer), Schedule B exceptions (title).

workflow()

Submit clawql-realestate-title-ingest or clawql-realestate-psa-ingest with deal_id.

hitl_enqueue_label_studio()

Schedule B exception review or PSA field confirmation with auto-resume.

memory_recall / memory_ingest()

Thread deal notes — [[Deal {id}]], [[FSBO — {address}]], [[Offer N — {buyer}]].

loan_archive()

Financed purchases: tag transaction packages when mortgage module is enabled.

Document types

What the IDP pipeline processes

Representative inputs agents classify, extract, redact, and archive in this vertical.

Purchase and sale agreements
Buyer offers (FSBO)
Title commitments (Schedule A + B)
Appraisals and inspection reports
HOA and condo disclosure packages
Closing disclosures and settlement statements

Use cases

Why ClawQL for real estate

Vertical packages extend the same Agentic Gateway — search, execute, memory, IDP, audit — with domain tools from real-estate transaction samples.

Title commitment intake

Parse title commitments with Docling, classify Schedule B routing, extract policy amount and exceptions with citations, HITL when confidence is low — vault note links to the deal_id used in your folder naming convention.

PSA field extraction

Extract purchase price, earnest money, closing date, and parties from purchase agreements — cross-check against title policy amount via memory_recall before closing.

FSBO offer comparison

Extract buyer_offer contingencies from multiple PDFs — compare financing, inspection, appraisal, and sale-of-home terms without re-reading each contract line by line.

CRM + storage without the re-read loop

ClawQL does not replace your CRM or cloud folders. It makes transaction PDFs intelligible to agents: "What is the earnest money on 123 Main?" returns a grounded answer with char_interval, not a Ctrl+F hunt.

Secure external sharing

After redaction, Coneshare VDR share links with viewer webhooks notify Slack when buyer's attorney or lender opens the title package — coordinators follow up in real time.

Examples

Example agent workflows

Representative MCP tool sequences with step-by-step detail — production deployments add tenant classifiers, RBAC, and your compliance policies on top.

Shipped reference path — synthetic title commitment through classify, extract, HITL, vault.

Title commitment — Schedule B HITL

The real-estate-title pack demonstrates the pattern transaction coordinators need: parse, score confidence on Schedule B, human review for curative exceptions, persist to vault linked to deal_id.

1Parse layoutexecute docling_convert_file on fixtures/synthetic-title-commitment.txt.
2Classifyclassify_document routes title_commitment — threshold 0.90 for HITL gate.
3Extractextract_document schema_preset title_commitment — property_address, policy_amount, schedule_b_exception with char_interval.
4HITL + vaultLow confidence → hitl_enqueue_label_studio (Schedule B UI) → memory_ingest "Title — {deal_id}" with wikilinks.

execute()classify_document()extract_document()hitl_enqueue_label_studio()memory_ingest()

Shipped reference path — purchase agreement fields with cross-doc recall.

PSA intake with deal threading

The real-estate-psa pack extracts contract fields and links vault notes to the title commitment note for the same deal_id.

1Parse + classifyclassify_document label purchase_agreement on synthetic-psa.txt fixture.
2Extractschema_preset purchase_agreement — purchase_price, earnest_money, closing_date, buyer_name, seller_name.
3Submit workflowworkflow submits clawql-realestate-psa-ingest with deal_id demo-deal-123-main.
4Cross-checkmemory_recall "purchase price and title policy for {deal_id}" — verify PSA vs title alignment.

execute()classify_document()extract_document()workflow()memory_recall()

Shipped reference path — buyer_offer preset with contingency extraction.

FSBO — compare two buyer offers

The real-estate-fsbo pack extracts competing offers so sellers (or advisors) compare price, closing date, and contingencies from grounded fields — not manual PDF review.

1Parse offersdocling_convert_file on synthetic-buyer-offer.txt and synthetic-buyer-offer-alt.txt.
2Classifyclassify_document label buyer_offer on FSBO offer text.
3Extractschema_preset buyer_offer — purchase_price, financing_contingency, inspection_contingency, sale_of_home_contingency.
4Comparememory_ingest [[FSBO — 456 Oak Lane]] + per-offer notes → memory_recall "which offer is cash and closes soonest?"

execute()classify_document()extract_document()memory_ingest()memory_recall()

Production-style workflow templates with explicit suspend step.

Argo suspend / resume for coordinator HITL

Same pattern as lending W-2: workflow hits suspend at hitl-review; coordinators complete Label Studio; webhook auto-resumes.

1Submitclawql-realestate-title-ingest or clawql-realestate-psa-ingest with confidence_threshold.
2SuspendTemplate waits at hitl-gate — deal state preserved in Argo, not chat context.
3ReviewCoordinator completes Label Studio task; CLAWQL_HITL_WEBHOOK_TOKEN validates webhook.
4Finalizeworkflow resume writes title-result.json or psa-result.json artifact with deal_id.

workflow()hitl_enqueue_label_studio()audit()

POC environment via Docker Compose — same Docling + classifier + Label Studio as lending.

Local demo stack (reuse lending Compose)

Point agents at synthetic fixtures; bootstrap Label Studio with real-estate-title or real-estate-psa label-studio-config.xml.

1Bootstrapdocker compose -f lending.compose.yml up -d — see docker/compose/README.md real estate section.
2Label StudioPaste label-studio-config.xml from deployment/samples/real-estate-title/ or real-estate-psa/.
3Connect agentCursor or OpenClaw at http://localhost:8080/mcp.
4Smoke flowRun title, PSA, or FSBO offer path from deployment/samples/real-estate/README.md.

search()execute()classify_document()

Security & compliance

Built for regulated workflows

Industry pages summarize platform capabilities — your legal, compliance, and security teams should review docs.clawql.com/security before production data.

PII redaction via Stirling before documents enter Onyx or archive indexes
Merkle audit trail on agent processing steps — not just activity logs
Self-hosted option keeps transaction docs on your infrastructure
Synthetic fixtures only in reference packs — no real client data in demos
VDR viewer webhooks for external access audit without replacing broker compliance workflows

Resources

Ready to demo ClawQL for real estate?

Self-host the real estate compose stack today, or join the managed waitlist for hosted MCP, vault, and IDP tailored to real estate workflows. Start your 14-day trial or self-host free with npm or Helm — full Apache 2.0 stack, no license fee.

Book a demo Modularization roadmap