CaseClear — intake desk for small dental labs

1. One-liner

Pulls every case from every scanner portal and email, reads the Rx, and flags what’s missing before it hits the bench.

2. Trend signal — why now?

Dental labs are drowning in intake fragmentation. A lab today receives cases across iTero’s manager, Carestream’s web portal, Medit Link, Cerec/DDX, 3Shape Communicate, 3M, EasyRx — plus roughly half of all daily cases arriving by plain email and Dropbox. One lab on Dental Lab Network described running “every matching portal” while ~50% of cases still came in by email; staff report it “spirals out of control.” Someone has to open all of them every morning, download each case, and eyeball each prescription for completeness.

And the prescriptions are bad. Industry data: only 30% of dental technicians find written prescriptions “clear” and fully understandable, and in up to 66% of cases technicians had to call the dentist to clarify details — missing shade, undefined margin design, no material spec, ambiguous tooth number. Each gap is a “delay domino”: production pauses, the case gets pulled off the bench, a phone call goes out, and the tech waits — sometimes a day — for a callback that often doesn’t come.

Two things changed in the last 12 months. First, vision/OCR models now read messy handwritten medical prescriptions at ~81% exact-match accuracy (12.5% word error rate) — and a dental Rx is far more structured than a free-text drug script (fixed fields: tooth #, shade, material, margin, pontic design), so extraction is easier and missing-field detection is reliable. Second, the dental software market is compounding ~10.6% a year ($3.2B → ~$8B by 2033), with Evident publicly shipping a new product every month and AI money pouring into the category (Pearl, Overjet, 3Shape). The wallet is opening; the intake bottleneck is the part nobody has neutrally solved.

Provenance:

• Signal 1 (demand): “Only 30% of technicians find written prescriptions clear; in up to 66% of cases technicians had to call the dentist to clarify”; labs juggle 7+ scanner portals + ~50% of cases via email — https://www.evidentdigital.com/blog/dental-lab-bottlenecks / https://dentallabnetwork.com/forums/threads/multiple-digital-impression-portals.30695/ — 2026-06-19
• Signal 2 (feasibility): AI-driven OCR reads handwritten medical prescriptions at ~81.3% exact match / 12.5% WER; dental Rx is more structured than drug scripts — https://jcbi.org/index.php/Main/article/view/1054 — 2026-06-19
• Signal 3 (economic): Dental software market $3.2B (2024) → ~$8B (2033) at 10.6% CAGR; Evident shipping monthly; AI investment accelerating across the category — https://www.precedenceresearch.com/dental-software-market — 2026-06-19 Category: Workflow automation

3. The opportunity

The gap is the neutral aggregator. The incumbents that touch intake are structurally conflicted: 3Shape Unite/Communicate exists to keep cases inside the 3Shape ecosystem; iTero, Medit, Carestream each want labs glued to their portal; EasyRx is strong but leans orthodontic and only auto-ingests cleanly when the dentist also runs EasyRx — which most don’t. None of them has any incentive to build the one thing the lab actually needs: a single place that vacuums cases out of all their rivals’ portals, plus email and Dropbox, and standardizes them.

So the small lab does it by hand. That manual intake — log into seven systems, download, transcribe, and read each Rx for completeness — is 1–3 hours of an experienced person’s morning, every day, before a single crown gets designed. CaseClear collapses that to a single triaged queue: every case in one inbox, the Rx already parsed into structured fields, and the incomplete ones flagged with exactly which field is missing and a pre-drafted clarification message ready to send to the dentist. The 10× isn’t “AI designs crowns” — it’s “the case is bench-ready the moment a human looks at it, instead of after a phone-tag round trip.”

4. Target market

• Primary customer: Owner or office manager of a US crown-&-bridge / removable dental laboratory, 2–15 technicians, taking cases from 30–100 dentist accounts. The family-owned <10-technician segment manufactures ~$6B of US lab output — these are the labs with the worst portal sprawl and no IT staff to fix it.
• Why they buy (their words): “I’m logging into seven portals every morning.” “The doctor didn’t write the shade and won’t call me back.” “Half my cases come in by email and I’m copying them into my system by hand.” Remakes and stalled cases are margin they can’t get back.
• Rough TAM reasoning: US dental labs number ~4,200–7,200 by the conservative counts (NAICS 339116 ≈ 5,000 establishments). Target the ~3,000–4,000 small/mid labs with real multi-channel intake pain. At ~$3,000/yr ACV, even 1,500 labs is ~$4.5M ARR — a clean sub-$5M business without leaving the US.
• Why now for them: Scanner-portal count crossed the threshold where manual intake genuinely doesn’t scale, margins are squeezed, and — for the first time — cheap vision models can read the handwritten and PDF Rx that make up half their inflow.

5. Product sketch (MVP)

• One unified case inbox that pulls in cases from email + Dropbox/shared folders on day one, and from the major scanner portals (iTero, Medit, 3Shape, Carestream) via account connection or forwarding as integrations land.
• Rx reader that parses each prescription — typed, PDF, or photo of a handwritten slip — into structured fields: patient/case ID, tooth number(s), restoration type, material, shade, margin design, pontic, due date.
• Completeness triage: every case lands tagged ✅ bench-ready or ⚠️ missing-info, with the specific gaps named (“no shade specified,” “material ambiguous: ‘zirconia’ but no translucency”).
• One-click clarification: a pre-drafted message to the dentist’s office requesting exactly the missing fields, sendable by email/SMS, with the reply auto-attached to the case.
• Standardized work ticket: a clean, printable/exportable case sheet so the bench works from one consistent format regardless of where the case came from.
• Per-dentist memory: learns each account’s defaults and recurring omissions (“Dr. Lee never writes margin — always asks for chamfer”) to pre-fill and pre-flag.
• Daily intake digest: “14 cases in, 9 bench-ready, 5 need info — 3 clarifications already drafted.”

6. AI angle — what’s load-bearing

Remove the AI and there is no product — just another portal to log into. The AI does three irreducible jobs: (1) reads heterogeneous, often-handwritten, often-photographed prescriptions across dozens of clinic formats into one schema; (2) reasons about completeness — not just “is the shade field blank” but “this is an anterior crown with no shade and no stump shade, which for this material is a stop”; (3) drafts the right clarification in the dentist’s-office register, asking only for what’s genuinely missing. Rule-based parsing breaks on the first handwritten slip or non-standard PDF; that’s exactly the input that dominates real lab inflow.

7. Localization angle (if any)

N/A — this is a US-first play. The wedge is the specific US scanner-portal landscape (iTero/Medit/3Shape/Carestream/EasyRx) and US dental-lab economics. The same intake-fragmentation pain exists in the EU/UK and could be a phase-2 geography, but there’s no localization wedge that beats nailing the US first; language and payment rails aren’t the constraint here — portal coverage and Rx-reading accuracy are.

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

• Pricing: $199/mo base (single location, email + Dropbox intake + Rx reader) → $399/mo (multi-portal connections + per-dentist memory + clarification automation). Add-on: $0.50–1.00 per case over a monthly cap for high-volume labs.
• ACV: ~$3,000–3,600/yr blended.
• To $1M ARR: ~280 labs at $300/mo. That’s <10% of the addressable small-lab base.
• To $5M ARR: ~1,400 labs at ~$300/mo, plus per-case overage on the busiest accounts. Requires solid portal coverage and word-of-mouth in a tight-knit industry — both plausible, neither guaranteed.
• Expansion path: seats are irrelevant (it’s per-lab), so growth comes from per-case volume tiers, adding portal connectors (each new connector raises willingness-to-pay), and an eventual outbound layer — once you’re the system of record for intake, you can offer dentist-facing digital Rx forms that feed in pre-validated.

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

• NADL + state lab associations: ~1,000 NADL member labs, plus state guilds, are a directory of exactly the buyer. Sponsor/exhibit at a regional lab meeting, run a “how much of your morning do you lose to portals?” live demo. These owners talk to each other constantly.
• Dental Lab Network forum (where the complaints already are): the multi-portal and intake threads are full of named pain. Show up with a tool that ingests their seven portals, offer 20 founding labs a free 60-day pilot, convert on time-saved.
• Loom-to-owner cold outreach: scrape lab listings, send a 90-second personalized video showing their public Rx form’s gaps auto-detected. Small, identifiable list (~3,000–4,000 labs); a 3–5% reply rate is a full pipeline.
• Scanner-agnostic angle as the hook: every other tool asks “which scanner?” CaseClear’s pitch is “all of them, plus email.” That single line differentiates in every conversation.
• Referral loop inside DSOs/group labs: land one lab in a group, expand to its sister locations.

10. Build complexity — justification

Medium. The Rx reader, completeness reasoning, unified inbox, and clarification drafting are all off-the-shelf vision/LLM APIs plus standard web stack — a strong solo builder ships that core in ~8–10 weeks. The complexity (and the moat) is the portal connectors: email/Dropbox ingestion is trivial, but iTero/Medit/3Shape/Carestream each need account-linking or forwarding glue, and APIs are inconsistent with ugly edge cases (resubmissions, remakes). Sequence it: ship email + photo-Rx first (covers the ~50% email volume immediately), add connectors one at a time. v1 in 3–4 months for a small team.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Processing case metadata/Rx; minimal PHI, standard BAA where needed. No clinical decision-making.
Ethical — no harm / dark patterns	✅	Reduces errors and rework; pro-quality, pro-patient.
Market exists (evidence above)	✅	Sourced demand + portal sprawl + clarification-call frequency.
1–5 person team can build this	✅	Off-the-shelf AI + staged connectors.
Launchable with <$50K / ₹40L	✅	Inference + dev time; no hardware, no inventory.

All five pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	16/20	Daily, felt by every multi-account lab; 66% clarification-call rate is hair-on-fire-adjacent. Not life-or-death, so not a 20.
Demand evidence	15	12/15	Strong sourced signals (clarity %, portal sprawl, market CAGR). Direct “I’d pay for this” quotes still to gather → not 14+.
Build feasibility	15	11/15	Core is easy; portal connectors are the gnarly, sequence-dependent work.
Distribution clarity	15	11/15	Tight, identifiable buyer base (NADL, the forum, ~4k labs). Tight-knit but slow-moving, conservative buyers.
Revenue mechanics	15	11/15	Pricing benchmarks to existing $100–400/mo lab tools; ACV credible; ARR path needs connector coverage.
Time to first revenue	10	8/10	Email + photo-Rx MVP can be pre-sold to forum pilots in <8 weeks.
Defensibility	10	6/10	Moat = connector library + per-dentist memory + system-of-record lock-in. Copyable, but incumbents are conflicted out of building it.
Total	100	75/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required — needs someone who can wrangle vision/LLM extraction and either has a dental-lab insider or recruits one fast, because the connector edge cases and Rx vocabulary are learned, not Googled.

Key assumptions to validate (3–5)

1. Assumption: Small labs will pay $200–400/mo for intake triage specifically (not bundled into full lab-management). How to test: 20 owner interviews + a pre-sell landing page; target ≥30% “take my card for a pilot.”
2. Assumption: Email + photo-Rx alone (before any scanner connectors) saves enough time to be worth paying for. How to test: 10-lab pilot measuring minutes-saved/day on email cases only; need ≥45 min/day.
3. Assumption: Rx-reading accuracy is high enough that techs trust the missing-field flags rather than re-checking everything. How to test: Run 500 real anonymized Rx through extraction; target ≥90% field-level precision on shade/material/tooth#.
4. Assumption: Portal connectors are buildable without scanner-vendor cooperation (forwarding/account-link, not blessed APIs). How to test: Build the iTero + Medit connectors first; if both require vendor partnership, the moat and the GTM both shift.

Risk flags

1. Platform dependency: Scanner vendors could close off forwarding/scraping or ship their own aggregator. Mitigate by owning the email/photo lane (which they can’t close) and the dentist relationship.
2. Conservative, slow buyers: Dental lab owners are not early adopters; sales cycle may stretch. Mitigate with free pilots and time-saved proof, not feature lists.
3. Accuracy trust gap: One wrong “bench-ready” tag on a missing shade = a remake the lab blames on you. Triage must err toward flagging, and surface confidence.

14. Structured verdict

Score:                  75/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder + a dental-lab insider advisor (or co-founder)
Time to revenue:        6–10 weeks to first paid pilot (email + photo-Rx MVP)
Capital to launch:      ₹8–18 lakh / $10–22K (mostly inference + 3–4 months of build)
Top 3 assumptions to validate first:
  1. Labs pay $200–400/mo for intake triage standalone — 20 owner interviews + pre-sell page
  2. Email + photo-Rx alone saves ≥45 min/day — 10-lab measured pilot
  3. Rx extraction ≥90% field precision on shade/material/tooth# — 500-Rx accuracy run
Kill criteria:
  - Abandon if <20% of 30 cold-contacted labs agree to a free pilot
  - Abandon if extraction can't clear 90% field precision after two model/prompt iterations
  - Abandon if a scanner vendor ships a neutral multi-portal aggregator before your v1

15. Next step — 1-week validation sprint

• Day 1–2: Pull 30 small-lab contacts from NADL/state directories. DM/call 30; book 12 owner conversations. Ask: how many portals, how long is morning intake, how often do you call dentists for missing info.
• Day 3–4: Collect 50–100 real (anonymized) prescriptions — photos and PDFs — from 2–3 friendly labs. Run them through an off-the-shelf vision model for field extraction. Measure field-level precision on tooth #, shade, material.
• Day 5: Decide. Go if ≥6 of 12 owners describe intake as a top-3 daily pain and extraction hits ≥85% precision on the test set. No-go if owners shrug (“it’s fine”) or extraction can’t read the handwritten slips that make up the hard half of the inflow.

The falsifiable result: owner-stated pain ranking (top-3 or not) and a hard precision number on real prescriptions — both measurable in five days.