DrawClear

1. One-liner

Tells a private lender which construction draws are complete enough to wire today, and chases the one missing waiver.

2. Trend signal — why now?

Construction draw administration is the last spreadsheet-run process sitting on top of billions of dollars, and 2026 is the year the cost of that became legible.

The money is stuck, and the delay is measurable. Built’s own transaction ledger puts the typical trade partner on a commercial job at about 50 days waiting to get paid after submitting a pay application (57 days average). Billd’s 2026 National Subcontractor Market Report found an average of 51 days. The gap isn’t the owner — owners pay GCs in a median of 17 days. The 32-day spread is pure paperwork friction, and Built names lien waiver collection across tiers as the first of four causes.

Missing paper, not missing money, is what stops a draw. Industry guidance is blunt: missing or incorrectly timed lien waivers are the #1 cause of funding holds and draw rejections, and a single missing waiver from one sub or supplier can hold an entire draw. Manual review of a draw request runs 5 to 10 business days, driven by document reconciliation across disconnected systems.

The small lender is the one still doing it by hand. Lenders who outsource draws to a national servicing firm face ~10-day delays; those who handle draws in-house often wire within 24–48 hours of inspection. Meanwhile construction loan administrators remain tethered to an “Excel legacy” — shared email inboxes, disconnected spreadsheets, and manual data entry to manage millions in capital. The incumbent software (Land Gorilla, Built, Rabbet) aims at banks and midsize financial institutions: Land Gorilla’s reviewer base is 54% midsize businesses, 39% banking.

That’s the whole thesis. The duty to verify a draw packet is identical whether you’re a $40B bank or a lender with 60 active construction loans. The software is priced and shaped for the bank.

Provenance:

• Signal 1 (demand): Missing/mistimed lien waivers are the #1 cause of draw rejections; a single missing waiver holds the whole draw; manual draw review takes 5–10 business days — https://getbuilt.com/blog/lender-construction-draw-request/ — observed 2026-09-05
• Signal 2 (economic): Subs wait a median 50 days (57 avg) post-pay-app while owners pay GCs in a median 17 days; Billd 2026 survey shows 51 days and 64% of subs say their GC pays slowly — https://getbuilt.com/blog/subcontractor-payment-delays/ — observed 2026-09-05
• Signal 3 (feasibility/structural): Outsourced draw servicing = ~10-day delays vs 24–48h in-house; small lenders run on spreadsheets and email while incumbents target midsize banks (Land Gorilla: 54% midsize, 39% banking) — https://nfronline.com/hard-money-lender-inspections/ + https://www.capterra.com/p/147212/Construction-Loan-Manager/ — observed 2026-09-05 Category: Underserved niche

3. The opportunity

Every construction draw is the same question asked badly: is this packet complete enough that I can wire money without taking on lien risk?

Answering it means opening a folder of PDFs — a G702/G703 pay application, conditional and unconditional lien waivers from each sub and supplier, invoices, inspection photos, sometimes a title update — and cross-checking them against a budget line schedule and against what was waived last month. It is tedious, entirely rule-based, and today it’s done by a human squinting at attachments in Outlook for 5–10 business days.

The incumbents solved this for banks. Land Gorilla, Built and Rabbet are real products with real customers, and if you’re a community bank with a construction portfolio you should buy one. But they’re sold on annual commitments through a demo-and-procurement motion, they’re built around bank workflows (LOS integration, regulatory exam artifacts), and their reviewer base skews midsize-and-up. A private lender running 40–150 active construction loans out of an office of six people is not the buyer those products were designed around — and Trimble Pay and Built’s free GC tier don’t help the lender either; they’re aimed at the trades.

What I’d build is deliberately narrower than a construction loan management system. It’s a completeness verdict on one draw packet: ingest whatever the borrower emailed, extract who signed what for how much through which date, diff it against the budget and the prior draw, and return one of three answers — clear to fund, fund partially at $X, or blocked, and here is the exact document from the exact party you’re missing. Then it sends that chase email itself.

The wedge isn’t “manage your construction loans.” It’s “cut your draw turnaround from nine days to two, starting with the packet sitting in your inbox right now.”

4. Target market

Primary customer: Draw administrator / loan servicing manager at a US private, hard-money, or fund-based construction lender — 3–30 staff, roughly $20M–$300M loans under management, 30–200 active construction loans. Also fits family-office lenders and the servicing arm of fix-and-flip lenders who added ground-up.

Why they buy: Because the draw is a cash-velocity problem for them, not just their borrower. Undrawn commitment doesn’t earn interest, and slow draws lose repeat borrowers to the lender down the street who wires in 48 hours. From the field, in their own words:

• “I’ve been dealing with construction loans lately, and honestly, it can get messy fast.” — trebel73, Custom Home Building Forum
• “I keep a simple spreadsheet to track each draw, date requested, amount approved, and when it actually hits my account… It’s kinda tedious but at least I know where every dollar goes.” — trebel73
• “I’ve tried a couple budgeting apps before, but they didn’t really simplify things much — just added another layer to manage.” — charleswhite285, same thread
• “most apps weren’t flexible enough to handle the irregularities of construction spending — unexpected costs, changing timelines, different contractors needing payment on different schedules” — golfplayer25, same thread
• “construction loans are just messy by nature” — charleswhite285

And on the borrower side of the same transaction, the reason lenders lose deals: one developer in Dallas hit a lender that only inspected during the first week of each month, and when a mid-month inspection was needed the wait was three weeks — the choice was delay work or float nearly $800,000 to keep contractors on site. A separate account describes a draw sitting three weeks because a bank employee was out, with nobody able to explain the holdup.

Rough TAM reasoning: The American Association of Private Lenders has 800+ members and is the largest national body for private money lenders and fund managers. That’s the association-registered core, not the whole market — the broader private/hard-money lending population is several thousand firms, of which a meaningful slice writes ground-up or heavy-rehab construction paper that requires draw administration. I need roughly 105 paying accounts at blended ACV for a $1M ARR business — around 13% of the AAPL directory. That’s a share I can defend on a spreadsheet.

Why now for them: Rate environment has made undrawn-commitment drag expensive, borrowers openly shop on draw speed, and the document-extraction problem that made this unbuildable in 2021 is now a solved commodity. The competitive pressure is real: professional lenders wiring in 24–48 hours are setting the expectation that a 10-day turnaround now visibly loses business.

5. Product sketch (MVP)

• Packet intake by email. Borrower or builder emails the draw packet to a per-loan address. Everything attached gets parsed — no portal adoption required from the borrower on day one.
• Waiver ledger per loan. Who signed what, conditional vs unconditional, through what date, for what amount, at what tier. The running answer to “which party has an unwaived balance right now.”
• Completeness verdict. Clear to fund / fund partially at $X / blocked, with the blocking item named down to the specific party and document.
• Budget-line diff. This draw’s requested amounts against the schedule of values and against everything drawn to date; flags line overdraws and percentage-complete jumps that don’t match the inspection.
• Auto-chase. Generates and sends the request for the missing waiver or invoice to the specific sub/supplier, tracks the reply, and re-runs the verdict when it lands.
• Two-minute funding memo. A one-page PDF the credit person signs off on: what’s being funded, what’s held back, what changed since last draw.
• Turnaround clock. Days-to-fund per loan and per borrower, so the lender can prove the 48-hour promise they’re now selling against.

6. AI angle — what’s load-bearing

Remove the AI and this is a filing cabinet. The load-bearing work is extraction and reconciliation from unstructured, non-standard documents: lien waiver forms vary by state (many states have no statutory form at all, so subs use whatever their attorney or a Google template gave them), G702/G703s arrive as scans, photos of signed paper, and every builder’s Excel variant. Pulling signer identity, waiver type, through-date, and amount out of that heterogeneous pile — and then reasoning about whether the set is complete against a budget — is precisely the task that was uneconomic to build before cheap document models.

The verdict layer is the second AI-dependent piece: “this waiver covers through 8/31 but the pay app claims work through 9/15, so the tail is unwaived” is a judgment made across documents, not a field lookup. That’s what the lender is paying a human 5–10 days to do.

What is not AI: the ledger, the diff math, and the emails. Those are ordinary software, and they should be — a hallucinated funding verdict is a lien-risk event, so the model extracts and flags, and deterministic rules decide. Every verdict cites the page and line it came from.

7. Localization angle

N/A — this is a US-first play. Lien waiver law, the G702/G703 convention, and the private-lending market structure are US-specific, and the value is entirely bound up in those artifacts. Australia and Canada have analogous progress-claim regimes and are plausible year-two markets, but the wedge does not translate by translation.

8. Business model — path to $1M–$5M ARR

• Pricing: $400–$1,200/mo per lender by active-loan count (tiers at 50 / 150 / 400 active construction loans). Land it at the low tier; the count grows on its own.
• ACV: ~$9,600 blended (mix of $400 and $1,200 tiers, some annual prepay discount).
• To $1M ARR: ~105 lenders at $9,600. Alternatively 200 lenders at the $400 entry tier plus a handful of larger accounts. Both are inside the AAPL directory.
• To $5M ARR: ~450–520 lenders, which means expanding past AAPL into fund administrators, credit unions doing residential construction, and the servicing shops that administer draws on behalf of lenders — that last group is a channel, since one servicer covers many lenders.
• Expansion path: Per-draw overage above tier, then a paid borrower-facing portal, then inspection-vendor scheduling as a marketplace take. The honest expansion is loan count, not seats — these are small teams.

Gross margin is fine: document extraction on a draw packet costs cents, and a busy lender submits a few hundred packets a month.

9. Go-to-market wedge — first 100 customers

1. Work the AAPL member directory. 800+ members, public, structured, and filterable to the ones writing construction paper. Not a cold email — a completed audit of one of their real draws. Ask for one recent packet under NDA, run it, send back the waiver ledger and the gaps their analyst missed. That is a two-day sales cycle when it lands.
2. AAPL Annual Conference, Las Vegas, November 2026 (800+ attendees). One event where essentially the entire buyer population is in a room. Go with the audit offer, not a booth demo — book 30 packet audits on-site and close them in December.
3. Third-party draw inspection firms as a channel. These firms (NFR and peers) already touch every draw for dozens of small lenders and get blamed for delays that are actually document delays. Offer white-label: their inspection report arrives with the waiver completeness verdict attached. One inspection firm relationship = 20–40 lenders introduced warm.
4. BiggerPockets private lending forum + Scotsman Guide. The private-lending forum is where these operators publicly compare servicing headaches. Answer draw-administration questions with real answers for 90 days; the operators self-identify by complaining.
5. Borrower-side pull. Builders who’ve waited three weeks on a draw will name their lender. Publish a “how fast does your lender fund?” benchmark from anonymized data and let borrowers forward it upstream.

The first 100 is channels 1–3. That’s a directory, a conference date, and roughly a dozen inspection firms.

10. Build complexity — justification

Medium. Document extraction is off-the-shelf; the state-by-state waiver taxonomy and the reconciliation rules are the real work, and they’re knowledge work rather than research. Email intake, a ledger, a diff, and templated chase mail are standard web-app construction. Estimate 12–16 weeks to a v1 a design partner can run real draws through, for a technical pair — most of that spent on the waiver taxonomy and on being conservative in the verdict logic. No integrations are required for v1 (email in, PDF out), which is what keeps this out of High.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Document review tooling; the lender’s credit officer still makes the funding decision. Not legal advice, not a licensed activity.
Ethical — no harm / dark patterns	✅	Speeds payment to subcontractors at the bottom of the chain. The incentive alignment is unusually clean.
Market exists (evidence above)	✅	Funded incumbents serving banks, 800+ AAPL members, sourced 50-day payment lag.
1–5 person team can build this	✅	Two people, 12–16 weeks.
Launchable with <$50K / ₹40L	✅	Inference and hosting only; the cost is founder time.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	16/20	Felt every draw, multiple times a week, and it directly gates cash. Not quite hair-on-fire because the lender’s own money isn’t at risk — it’s velocity and borrower retention, not survival.
Demand evidence	15	12/15	Strong: sourced payment-lag data, named #1 rejection cause, funded incumbents, an 800-member association. Docked because I found no verbatim lender-side complaint — the quotes I have are borrower/builder-side.
Build feasibility	15	11/15	12–16 weeks for a pair. Waiver taxonomy across 50 states is grinding work, and verdict conservatism takes iteration.
Distribution clarity	15	13/15	A public 800-member directory, a dated November conference, and an inspection-firm channel. The packet-audit offer is concrete and fast.
Revenue mechanics	15	12/15	Pricing well below incumbent annual commitments, ~105 accounts to $1M. The $5M path needs a second segment, which is the soft spot.
Time to first revenue	10	7/10	8–10 weeks realistically — the audit motion presells, but lenders will pilot on real loans before signing.
Defensibility	10	5/10	The waiver taxonomy and accumulated verdict corrections compound, but a funded incumbent could aim down-market at this. Execution and focus moat, not a structural one.
Total	100	76/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · sales-heavy

Document reconciliation is the technical core; the buyer is a relationship-driven private lender who closes at conferences and on referrals. Construction lending domain knowledge is a strong advantage but can be bought with an advisor — this is not domain-expertise-required in the way a clinical product would be.

Key assumptions to validate (3–5)

1. Assumption: Small private lenders genuinely take 5–10 days on draw review, not the 24–48 hours the in-house ones claim. How to test: Ask 20 AAPL-directory lenders one question — “what’s your median days from packet received to wire?” — and separately ask five of their recent borrowers the same thing. Trust the gap.
2. Assumption: A completeness verdict from software is trusted enough to shorten review, rather than becoming a second opinion the analyst re-checks by hand. How to test: Run 10 historical packets through the v0 and have the lender’s analyst mark every disagreement. Adoption requires the analyst stops re-checking, so measure re-check rate, not accuracy.
3. Assumption: Willingness to pay $400–$1,200/mo when the alternative is an unpaid intern’s time. How to test: Price the packet audit at $250 in the pilot. Anyone who won’t pay $250 for one audit won’t pay $400/mo for unlimited.
4. Assumption: Waiver-form variance across states is tractable rather than infinite. How to test: Collect 200 real waivers from five lenders and measure how many distinct form patterns cover 90% of volume.

Risk flags

1. Incumbent down-market move: Built and Land Gorilla are funded and could ship a lightweight tier. Mitigation is speed and a segment they find unattractive to serve — but this is the real risk, and it’s why defensibility scores 5.
2. Liability perception: If a lender funds on a green verdict and eats a lien, they will blame the software regardless of contract language. The product must be architected around flagging, with the human signing, and marketed that way. Get this wrong and the sales cycle triples.
3. Free-tier squeeze: Trimble Pay is free for subs and Built has a free GC tier. Neither serves the lender today, but a free lender tier appearing would compress pricing fast.
4. Concentration in a rate-sensitive market: If construction lending volume contracts, active-loan-count pricing contracts with it. Revenue is directly levered to a cyclical market.

14. Structured verdict

Score:                  76/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical pair, one of whom will do conference/relationship sales; construction-lending advisor on equity
Time to revenue:        8–10 weeks from launch
Capital to launch:      $8–12K (₹7–10L) — inference, hosting, one conference ticket
Top 3 assumptions to validate first:
  1. Median days packet-to-wire at 20 AAPL lenders vs what five of their borrowers report
  2. Analyst re-check rate on 10 historical packets — adoption needs it to fall to near zero
  3. $250 paid packet audit converts; refusal at $250 predicts refusal at $400/mo
Kill criteria:
  - Abandon if fewer than 5 of 20 AAPL lenders report median packet-to-wire above 4 days
  - Abandon if analysts still re-check 100% of verdicts after 10 packets in pilot
  - Abandon if Built or Land Gorilla ships a sub-$500/mo lender tier before v1 ships

15. Next step — 1-week validation sprint

• Day 1–2: Pull the AAPL member directory, filter to firms writing ground-up or heavy-rehab. Email 40 of them one question: “What’s your median time from receiving a complete draw packet to wiring?” No pitch. Target 12 answers.
• Day 3–4: Get 3 lenders to send one real historical draw packet under NDA. Hand-build the waiver ledger and completeness verdict for each — no code, just me and the PDFs, timed. Send each back with the gaps found and ask: “would you have caught all of these, and how long would it have taken you?”
• Day 5: Offer the $250 paid packet audit to all 12 respondents.

Falsifiable outcome: Go if ≥5 of 12 report a median above 4 days and ≥2 of 12 pay $250 for an audit within the week. If the reported medians cluster at 48 hours, the pain I’ve priced doesn’t exist at this segment and the idea dies here — not after a 14-week build.