The problem
Closings run as a relay of handoffs over 30 to 45 days. Money moves before the record does. Wire fraud lives in the gaps.
01
Executive summary
Vision
Make the real estate closing a single, verified event. Funds, documents, and recording release together, once every condition is proven. Fraud loses the gap it lives in.
Company
Aurora is an AI-native title and settlement platform. A rules engine builds a condition list for every file. AI does the reading, matching, and drafting. A licensed examiner handles only the exception. When every condition is green, the settlement releases in one event. The platform is sold to title agencies, lenders, and brokerages as software, with a regional operator model for geographic reach.
The product
A condition ledger, an AI file agent, a single release event, settlement rails that move money in seconds.
The business
Software revenue first. Platform fees, subscriptions, API, and regional operators. Florida pilot, then state by state.
$6M
Seed target
$11M to $17M
Total capital over 36 months
~$11.2M
Year 3 revenue (plan)
~32%
Year 3 operating margin (plan)
02
The problem
A closing today is eight handoffs in a row: search, curative, documents, signing, funding, disbursement, recording, policy. Each one waits on the one before it. Each handoff is a place to fail.
Funds wired, deed not yet recorded.
Payoff sent, lien not yet released.
Wire instructions changed in the gap between confirmation and release.
In title, the fraud is the wire. The gap is where it lives.
Average closing costs run $4,661 nationally, about 1.06 percent of the home price, and reach 2.99 percent in some states (LodeStar, 2025).
Morgan Stanley estimates 37 percent of real estate tasks can be automated, worth $34 billion in operating efficiency by 2030 (Morgan Stanley Research, 2025).
No partial closings. By design.
03
What atomic settlement means
The idea comes from securities settlement: delivery versus payment. The buyer's money and the seller's asset change hands at the same instant, or not at all.
A real estate closing has the same shape. The deed is delivered. The new loan funds. The old loan is paid off. Taxes and commissions are paid. The seller receives proceeds. The deed goes to record. Today each of these happens on its own clock.
Aurora binds them into one conditional transaction. Every condition is checked by the system and backed by evidence. All conditions met: everything releases at once. Any condition fails: nothing moves.
One conditional transaction
Deed
Loan funding
Payoff
Taxes and commissions
Seller proceeds
Recording
No partial closings. By design.
Escrow was always this, done by hand
The settlement agent already holds everything until the conditions are met. Today those conditions live in email, spreadsheets, and someone's memory. Aurora makes every condition machine-checkable and the release simultaneous.
The recording step
County recording is not instant. That is the one step outside our hands. eRecording shortens it to hours in most Florida counties, same day in Tampa, Miami, and Orlando. Title insurance already covers the gap between release and record. Atomic in the money. Insured in the record.
04
How a file moves through Aurora
123 Sample St, Sample FL
Open
The file opens. AI reads the contract and the commitment.
- Open
The file opens. AI reads the contract and the commitment.
- Assemble
The rules engine builds the condition list: title clear, payoff verified, identities verified, documents signed, lender funds received, wire instructions human-verified.
- Ledger
Every condition is tracked pass or fail with its evidence attached.
- Lock
All parties confirm the final numbers.
- Release
Funds disburse and documents submit for recording in one event.
Condition Ledger
123 Sample St, Sample FL
- Title commitment clearsample-commitment.pdfVerified
- Payoff letter verifiedsample-payoff.pdfVerified
- Seller ID verifiedsample-seller-id.pdfVerified
- Buyer ID verifiedsample-buyer-id.pdfVerified
- Loan docs signedsample-signed-docs.pdfVerified
- Lender funds receivedsample-funds-receipt.pdfPending
- Wire instructions human-verifiedsample-wire-verification.pdfVerified
1 condition not green yet.
One screen. Only what is not green yet.
05
AI runs the file
AI runs the file.
The human gets only the exception.
AI runs the file.
The clean work clears itself.
The human gets only the exception.
License and judgment calls route to a licensed person. That person's call tunes the next file. The queue gets shorter over time.
Render the exception, not the file.
The examiner's screen is a short list of what a human has to touch today, not a production system to drive.
Communication first.
The file talks, in real time, both directions. The same agent running the file is the one talking to the lender, the realtor, and the borrower. Encompass for the lender. Text for the borrower. No new app to log into.
A portal is a place the client has to go. Aurora is the inverse.
06
Proof of concept, running now
aura_poc.py. One inbound email in. Out comes the matched file, the data pulled, a deterministic check, a triage call, and a drafted reply. It runs on a laptop, on synthetic files, for pennies.
A payoff letter lands.
Matched to the file by lender and address. Amount and good-through date pulled. Date checked against the deterministic rule PAY-001. Receipt drafted. Auto-send.
The Beach Dr closing.
A realtor asks about a file with no order number. Matched by comprehension. That is the email mapping ResWare did with brittle rules and Qualia dropped entirely.
A borrower confirms a wire account.
That one never auto-sends. Bright-line rule. Every time.
The wire is the credibility. Deterministic checks and a licensed human on every judgment call. That is the liability posture an underwriter wants to see before they let you near a file.
Built and tested on synthetic data only. Real files only after SOC 2 and data residency are resolved.
The tech was never the moat.
07
Market opportunity
US title premiums
- Premiums ($B)
Addressable market
| Segment | Annual transaction volume | Fee pool at 1% | Fee pool at 3% |
|---|---|---|---|
| US residential real estate | $2.5 trillion | $25B | $75B |
| US commercial real estate | $600 billion | $6B | $18B |
| Florida (initial market) | $450 billion | $4.5B | $13.5B |
Settlement and title fees run roughly 1 to 3 percent of transaction value depending on state and loan type (LodeStar, 2025). Volume figures are planning estimates.
Market drivers
AI efficiency
Morgan Stanley, $34B by 2030.
Cost pressure
Closing costs up to 2.99 percent in some states.
Wire fraud
FBI IC3 reports business email compromise losses in the billions each year. Real estate is a top target.
Instant payments
FedNow and RTP now reach most US deposit accounts.
State modernization
Remote online notarization and eRecording statewide in Florida.
Florida first
#2
Title-premium state
About $2.2B in 2025 (ALTA, 2025).
Promulgated rates
Efficiency and distribution are the only levers. Median home about $412K. Owner's premium about $2,035 (Florida OIR Rule 69O-186.003).
Remote online notarization
And eRecording statewide.
Target customers
Primary
Independent title agencies and escrow companies.
Secondary
Mortgage lenders and brokerages that want a closing they can see.
Tertiary
Attorneys and county registries that need clean integration.
Enterprise
Institutional investors and funds that close in volume.
08
Competitive landscape
Traditional title
First American, Fidelity, Old Republic, Stewart. Established relationships. Legacy systems. Sequential settlement. 30 to 45 days.
Title software
Qualia, ResWare, SoftPro. Digitized workflow. Still sequential. Qualia is web-native but dropped email mapping. ResWare maps email but runs on your own servers and you build every workflow yourself.
Proptech and blockchain entrants
Propy and others. Record-keeping on chain. No full settlement automation. Thin regulatory footing.
| Traditional title | Digital title software | Aurora | |
|---|---|---|---|
| Settlement model | Sequential | Sequential, digitized | Atomic |
| Partial-closing risk | Yes | Yes | No, by design |
| Funds move before conditions are proven | Often | Often | Never |
| Exceptions | Found late, by hand | Found late, on screen | Flagged early, with evidence |
| Wire instructions | Phone call | Phone call | Human-verified, logged, then released |
| Onboarding | Months | Weeks to months | Days |
The tech was never the moat. The parts are on the shelf. The missing piece is adoption, service, and the network of agencies and lenders that forms around one shared ledger.
09
Platform architecture
Settlement Core
Rules engine and condition ledger. Deterministic. Decides what releases and when.
Intelligence Layer
Anthropic Claude API with custom models for document reading, matching, fraud signals, and drafting. AI reads and drafts, it never decides a release.
Interface Layer
Web app for agencies, lenders, realtors. Mobile for signing and status. GraphQL API and webhooks for integrations.
| Stack | Purpose |
|---|---|
| Postgres | Transactions |
| Snowflake | Analytics and reporting |
| Anthropic Claude | Document understanding, drafting |
| ATTOM Data | Property data for 155 million US properties |
| Tamper-evident archival storage | Closing documents |
| HashiCorp Vault | Secrets |
| AES-256 at rest, TLS 1.3 in transit | |
| Biometric authentication on mobile | |
| Annual third-party penetration tests, SOC 2 program |
The rules engine decides. AI reads and drafts. Humans own licensed judgment.
10
Settlement rails
Today
Next
Today
FedNow and RTP move dollars between US banks in seconds, 24 hours a day. Lenders and settlement banks are already connected. Aurora's release event sends every disbursement on these rails.
Next
A Solana USDC rail for on-chain delivery versus payment, where the deed record and the payment settle on the same ledger. Pursued only when lenders, underwriters, and regulators are ready, and only with a licensed, dollar-backed stablecoin.
Rails, not a coin. Aurora has no token. Nothing here is an investment product.
Margins expand as automation deepens.
11
How the process evolves
- 0115%
Pilot
Prove it on one lane. Measure outcomes.
- 0220%
AI deployment
Four AI agents per licensed examiner. AI takes about 80 percent of routine work: search, classification, data entry, status updates, curative drafting.
- 03
Atomic release
The condition ledger goes live. One simultaneous release per file.
- 0425-32%
Scale
One centralized production hub. No duplicate overhead of a multi-office network. 200 or more files per examiner per month, against 50 to 60 in a traditional shop.
Operating margin by stage
- Operating margin (%)
12
Economic model
Lead with software revenue, not title splits. A title agency trades at one to two times revenue. Software trades at a multiple of recurring revenue. The same closings, valued at a different altitude.
Revenue streams
0.25%
Platform fee per closing
As a share of transaction value, or a flat per-file fee where state rules require.
$299 to $2,999
Monthly subscriptions
Tiered by seats and volume.
$0.01 to $0.50
API fees per call
For third-party integrations.
10 to 15%
Regional operator revenue share
Regional operators
| Tier | Territory | Initial license | Revenue share |
|---|---|---|---|
| Metro | Major city | $250,000 | 15% |
| Regional | Multi-county | $500,000 | 12% |
| State | Entire state | $1,000,000 | 10% |
Operators are licensed to run Aurora in a territory and carry local licensing, compliance, and client relationships. Indicative terms.
Year 3 revenue mix (plan)
- Platform fees45%
- Subscriptions25%
- Operator share15%
- API and data15%
13
Go to market
Year one free for a founding-agency charter cohort. Switching cost is the incumbent's wall. Free is the ladder over it. The free year is an occupation, not a trial: their pipeline, closings, and staff habits on our rails. The day they go live, the same wall keeps them in.
- 01
Florida first
One lane, a single underwriter's refinances.
- 02
State by state
Lane by lane, only after the prior lane is clean.
- 03
The easy file first
The hard files later, on purpose.
- 04
Regional operators
Once the Florida playbook is written.
RESPA
When title, a realtor channel, and a lender channel share one money flow, RESPA is in the room. A SaaS license is not a thing of value given for a referral. Every channel agreement is papered by a RESPA specialist before anyone signs. Compliance is the design, not the afterthought.
14
Partnership and ecosystem strategy
Technology
Anthropic, Snowflake, ATTOM Data, FedNow and RTP settlement banks.
Industry
Title underwriters: Fidelity, First American, Old Republic, Stewart. Mortgage lenders. Brokerages. County recorders.
Capital
Proptech and fintech investors who bring lender and underwriter relationships, not just capital.
Integration map
Loan origination
Encompass, Blend.
Transaction management
Dotloop and similar.
Banking
FedNow, RTP.
Government
County eRecording, tax authorities.
Underwriters
Automated commitment and policy issuance.
Developer program
GraphQL API, webhooks, OAuth, sandbox with simulated transactions, SDKs for JavaScript and Python.
15
Regulatory and compliance framework
Federal
RESPA
Fee disclosure, no kickbacks, settlement documentation.
CFPB and TRID
Closing Disclosure timing and accuracy.
GLBA
Nonpublic personal information safeguards.
BSA and AML
Customer identification, suspicious activity monitoring. FinCEN real estate reporting rules.
Florida
Licensing
Title agency license and agent licenses through the Florida Department of Financial Services. Surety bond. Background checks. Continuing education.
Escrow and trust accounting
Segregated funds, monthly three-way reconciliation, annual audit, DFS reporting.
Electronic records
Florida RON statute compliance, document retention, audit trails.
Compliance infrastructure
Identity verification
Biometric ID verification vendor plus knowledge-based checks.
Data privacy
GLBA and CCPA controls, consent management, data minimization.
Audit trail
Every condition, every piece of evidence, every human decision, time-stamped and immutable.
Legal review
AI-drafted, attorney-reviewed. Jurisdiction-specific rule sets.
Regulatory roadmap
- 01
Q1 2027
Florida title agency and escrow licensing in place.
- 02
Q2 2027
Compliance infrastructure live, first audited reconciliation cycle.
- 03
Q4 2027
SOC 2 Type I.
- 04
2028
SOC 2 Type II, expansion to reciprocity states.
- 05
2029
Multi-state licensing strategy.
16
Implementation roadmap
- 01
Phase 1, Foundation
Q4 2026 to Q2 2027Close seed. Core team. Settlement Core and condition ledger MVP. Florida licensing. Anthropic, Snowflake, and ATTOM agreements.
- 02
Phase 2, Pilot
Q3 2027 to Q1 2028Charter cohort of up to 10 Florida agencies. Refinance lane. Intelligence Layer in production on governed data. First atomic release on a live file. Mobile signing and status.
- 03
Phase 3, Expansion
20281,000 or more monthly files. Regional operators in the first metros. Texas and reciprocity states. Purchase transactions. Series A.
- 04
Phase 4, Ecosystem
2029Developer marketplace with 50 or more integrations. On-chain settlement rail pilot. Commercial real estate product.
| Milestone | Timing |
|---|---|
| Condition ledger live on synthetic files | Q1 2027 |
| First licensed Florida file on Aurora | Q3 2027 |
| First atomic release | Q4 2027 |
| SOC 2 Type I | Q4 2027 |
| 1,000 files per month | Q4 2028 |
| Multi-state capability | 2029 |
KPIs
10%
Monthly file growth target
In the first 24 months.
99.9%
Release success rate target
95%
Client satisfaction target
30% or more
Operating margin at scale target
Software multiples, not title-company multiples.
17
Financial projections
Revenue and operating income (plan)
- Revenue ($M)
- Operating income ($M)
| Year | Files | Revenue | Operating income | Margin |
|---|---|---|---|---|
| Year 1 | ~1,512 | $3.8M | ($1.4M) | Negative |
| Year 2 | ~3,000 | $7.5M | $1.5M | 20% |
| Year 3 | ~4,500 | $11.2M | $3.7M | 33% |
~$22.5M
Cumulative revenue over three years
~$3.8M
Cumulative operating income
$11M to $17M
Total capital
Phased over 36 months.
Unit economics (planning assumptions)
$4,200
Lifetime value per professional seat
$380
Acquisition cost per seat
11x
LTV to CAC
Use of funds, seed $6M
- Product and engineering$2.5M
- Regulatory and legal$1M
- Operations and go-to-market$1.5M
- Reserves$1M
All figures are planning estimates and will be updated against pilot data.
18
Risk analysis and mitigation
Release logic defect
Deterministic rules engine, independent code audits, staged rollout.
Regulatory rejection or delay
Early DFS engagement, counsel opinions, state-by-state sequence.
Slow industry adoption
Free year-one charter, measurable time and cost savings, hybrid mode.
AI reading errors
Confidence scoring, human verification, data cross-checks.
| Risk | Impact | Mitigation |
|---|---|---|
| Release logic defect | High | Deterministic rules engine, independent code audits, bug bounty, staged rollout with manual release fallback. |
| Regulatory rejection or delay | High | Early DFS engagement, counsel opinions, compliance-by-design, state-by-state sequence. |
| Slow industry adoption | High | Free year-one charter, measurable time and cost savings, hybrid mode that runs beside existing systems. |
| AI reading errors | Medium | Confidence scoring, human verification on every high-risk document, cross-checks against ATTOM and county data. |
| Payment rail outage | Medium | Dual rails (FedNow and RTP), wire fallback, release holds until confirmation. |
| Operator quality | Medium | Selection criteria, training, performance monitoring, revenue share tied to outcomes. |
| Key person | Medium | Early senior hires in technology and compliance, documented playbooks. |
Contingency
12-month runway maintained.
Manual processing fallback for every automated step.
Multi-region infrastructure.
Quarterly risk review.
19
Customer journey
Personas
Rachel, real estate agent
Persona. 30 transactions a year. Wants fewer surprises and real-time status.
Michael, title agency owner
Persona. 15 employees. Wants lower cost per file and a way to compete with tech-backed shops.
Sarah, first-time buyer
Persona. Wants to understand the process and not get defrauded.
Traditional
- 01
Day 1
Contract.
- 02
Days 2 to 7
Title search.
- 03
Days 8 to 14
Commitment.
- 04
Days 15 to 21
Curative.
- 05
Days 22 to 28
Documents.
- 06
Days 29 to 30
Closing.
- 07
Days 31 to 45
Recording and policy.
Aurora (plan target)
- 01
Day 1
Contract opens the file, AI reads it, condition list built.
- 02
Days 1 to 3
Automated search and commitment, exceptions routed to an examiner.
- 03
Days 3 to 7
Curative, payoffs verified, identities verified.
- 04
Signing day
All parties lock the numbers, every condition green, one release event.
- 05
Same day
eRecording submitted, policy issued, everyone notified.
The calendar is set by lenders and counties. The release is set by Aurora. One event, not eight.
20
Leadership
Evan Grimm, Founder
25 years in title. Founded Bay National Title in 2003 alone, with paper files, and grew it from Florida to 40 states and the Inc. 500 as CEO. Certified Land Title Closer. Ran self-hosted ResWare on a three-tier architecture for a decade. Currently rebuilding the operation at a 36-state national title agency as Executive Director. Working underwriter relationships with Fidelity, First American, Old Republic, and Stewart. Documentary filmmaker and nonprofit leader. Humanitarian work in Africa and the Middle East.
Roles to recruit
Chief Technology Officer
Chief Compliance Officer
Title and escrow regulation.
Head of AI
Document understanding, model governance.
Chief Information Security Officer
Head of Growth
Agency and lender channels.
CFO
Advisory board
To be assembled from title underwriting, payments, and real estate law. No names until signed.
One event, not eight.
21
Investment opportunity
Why now
Instant payment rails exist.
Remote notarization and eRecording are law in Florida.
Document AI is good enough to read a file.
Wire fraud keeps growing.
Nobody has bound the closing into one event.
Terms (indicative, subject to counsel and final documents)
$6M
Seed round
$30M
Pre-money
Preferred shares
With pro-rata rights and a board seat.
$500K
Minimum
Next steps
- 01
Request the data room.
- 02
See the proof of concept run.
- 03
Diligence.
- 04
Terms.
Evan Grimm, Founder.
22
Governance and sources
- Synthetic data until governed.
- SOC 2 before production data.
- No real order, payee, or nonpublic personal information in code repositories or public tools.
- Secrets never in source control.
- Wire instructions always human-verified.
- No release until every condition is green.
- 01ALTA 2025 market share and premium data.
- 02LodeStar 2025 closing cost report.
- 03Morgan Stanley Research 2025.
- 04Florida OIR Rule 69O-186.003.
- 05FBI IC3 annual report.
- 06ATTOM Data Solutions.
- 07Federal Reserve FedNow and The Clearing House RTP documentation.