Updated 2026-10-29 (operator ruling). This page supersedes every sentence in pages 00–38 that frames the business as metered software. Where an older page says "plan", "per run" or "credits", read this page instead. The metered plans still exist on the signup page for evaluation access; they are not what we sell.
One founder, a garage, no staff, no raise planned. An instrument maker and optical computation company that builds everything it sells. Headline: Turn Any Desktop Computer into a 3D Metrology Lab. Subheading: Precision 3D measurement, at cost. Ultra-affordable is an engineering requirement here, not a slogan: the whole company has to work on one demo unit and deposits.
Status words, always: the software is live and available for evaluation; the hardware is designed, not built, not measured; every hardware number is a design value.
Spend $5k, get $30k back, repeat.
| Phase | What | Price | Terms |
|---|---|---|---|
| 0 | Free bounded benchmark: up to ~20 of their frames, one reference, one pass/fail threshold agreed before we run, 10 business days | $0 | exists today (page 22 Phase 0) |
| 1 | Paid pilot engineering, milestone-paid development: M1 on signing = build and bench acceptance of the demo configuration for their task; M2 at the signed readout, acceptance criteria fixed before the first capture | $4,800 = M1 $2,400 + M2 $2,400 | both milestones credited in full to the instrument |
| 2 | The instrument: bench surface profiler (H1 head + D1 stand), the one product in development | $12,900 | 50 % on order, balance on delivery |
| 2 | Second-wavelength option (sharp steps beyond the single-wavelength limit) | +$2,400 | with the bench profiler |
| — | Portfolio: inline inspection head (H1 on a belt flange), I1 transmission module, optical computing card | not priced | configurations after the first product qualifies; the card is research only |
| — | Software and 90 days of onboarding | included | every Phase 2 unit |
| — | Design Partnership, software $10,000 / hardware $25,000 | unchanged | labs that want custom work (page 22) |
Research and industrial-QA units: not certified, not medical. What the bench profiler measures (design values): lens mode = form of matte, machined, printed, anodised or ceramic parts, about 50 × 39 mm at about 0.1 mm lateral sampling, not a microscope; dock mode = polished and mirror-like parts, about 5.7 × 4.3 mm per shot at about 10 µm. Smooth surfaces resolve height below a micrometre; rougher than about 0.16 µm needs the second-wavelength option. Indoor light fine, window daylight with the shroud and its door seated, direct sun no. Designed as a Class 1 laser product in both modes (label text verified against IEC 60825-1 before the first sale): the measurement zone is closed on four sides by a light shroud with a lift-off front door, and the laser enables only when the door and back wall are seated and the nose or dock is present. Open the door and the beam is off; a robot lifts the door between captures.
Honesty gate, repeated on every price line, every surface: parts cost is a design value; the first unit has not been built; price is an offer, not a quote book.
Parts, for the arithmetic only (design values from the fleet v6 directive): H1 head ≈ $1.49k, D1 stand with its electronics pod ≈ $0.29k, so a bench profiler is ≈ $1.78k built to order. The Phase 2 price is about 7× parts and still 75–90 % below the incumbents below. That is a price comparison, never a performance claim: nothing has been measured.
Cost model: parts (design value) + founder engineering hours (to be timed on the first build) + calibration time + rework + support hours + patent prosecution (four nonprovisional applications, counsel) + commercial + hosting (unmeasured). Development cost is unproven until the first timed build; the price holds until then.
Sequential development: buy and print only what the current test lane needs, and run the lane whose failure would kill everything first.
| Rank | Test lane | Kill criterion |
|---|---|---|
| 1 | FFC core first article in dock mode on a flat mirror (sensor, pigtailed diode, coupler, cube, printed holster on a bare table; ≈ $1.4k parts) | fringe visibility < 0.5 at the detent, or ghost sideband > −20 dB after background subtraction |
| 2 | Lens-mode physics on the same core: coherence scan, ND balance, ghost residual, matte-tile form map | no coherence revival within ±32 mm, or ghost residual > 10 % of object |
| 3 | Customer self-calibration, timed: flat capture + scale grid by a non-author with a stopwatch | > 60 min, or needs the author |
| 4 | Ambient: 500 / 2,000 lx with the four-sided light shroud | contrast loss > 1 % |
| 5 | D1 mechanics: carriage droop, re-seat repeatability, Class 1 door interlock chain | droop > 1 mm; interlock defeatable |
| 6 | Step standard + repeatability on a customer-class part | outside the agreed uncertainty |
Build sequence: S1 core parts → S2 printed holster + bench test (lane 1) → S3 head shell, collar and dock (lane 3, on a desk, no stand) → S4 nose, jumper, ND and bandpass (lane 2) → S5 D1 stand, wall and pod (lanes 4–5) → S6 tray riser or belt flange only after a customer asks. A timed build sheet (hours per step, rework) is recorded at every step; that sheet is the cost model.
| Instrument (class) | Vendor | Approximate list | Where the figure comes from |
|---|---|---|---|
| ZeGage Pro, benchtop coherence-scanning profiler | Zygo | $100k+ class | repo docs/research/hardware_form_factor_survey_2026-10-19.md (from the landscape doc); spec: https://smlouvy.gov.cz/smlouva/soubor/34936175/0.1%20ZeGagePro-Specs.pdf |
| Nexview NX2, 3D optical profiler | Zygo | $100k–$150k+ class | quote-only on the vendor page; class estimate, operator to verify: https://www.zygo.com/products/metrology-systems/3d-optical-profilers/nexview-nx2 |
| DHM R-series, lensed reflection digital holographic microscope | Lyncée Tec | "$100k or more" class | repo docs/design/ApertureAPI-main.md §DHM; vendor page quote-only: https://www.lynceetec.com/reflection-dhm |
| VR-6100 / 6200, at-line structured-light profiler | Keyence | quote-only; $40k–$60k class, operator to verify | https://www.keyence.com/products/3d-measure/3d-scanner/vr-6000/models/vr-6100/ |
| WI-5000, in-line white-light interferometric head | Keyence | quote-only, operator to verify | https://www.keyence.com/products/measure/laser-2d/wi-5000/specs/ |
| S lynx 2, compact benchtop profiler | Sensofar | quote-only, operator to verify | https://www.jhtechnologies.com/wp-content/uploads/2024/06/BR46-05A-EN-Brochure-S-lynx-2-version-A.pdf |
Vendors sell on quote; the figures above are the classes our own research files recorded, not vendor quotations. Before any figure reaches a customer, the operator verifies it against a dated public source and writes the date on this page. Compare prices only; never say "replaces your Zygo" (fleet v5 §0 correction stands: field, lateral resolution and accuracy are unmeasured).
Founder does the selling, one conversation at a time, by email and the 15-minute call. No ads, no booths, no lists bought. A placement is a signed Phase 1 or a Phase 2 deposit, nothing softer.
| Minutes | Agenda | Output |
|---|---|---|
| 0–15 | Their task: the part, the surface, the expected height and slope range, the reference method, what they use today and what the queue costs them | one paragraph, agreed |
| 15–35 | The exact configuration (dock mode for polished parts, lens mode for the form of matte/machined/ceramic parts; one wavelength or two), the reference artifact, the data manifest, the numerical acceptance criteria. Classify every piece of evidence as physical measurement, synthetic test, mechanical concept or planned work | the signed configuration sheet |
| 35–50 | The work statement with its two milestones (M1 build and bench acceptance; M2 signed readout), data handling, the readout date, and the purchase the readout would unlock | the work statement |
| 50–60 | Decisions, open items, owners, dates | the day-one minute |
The work: one head, one sensor, one wavelength (or two), one fixture. Record firmware and pipeline versions, head identity, exposure, geometry, calibration and reference information, raw captures. Measure in the shroud and in representative indoor light, with repeats and with removal and replacement. Capture failure cases on purpose (clipping, low contrast, displaced reference, wrong head profile) and show that the output marks them unusable.
The readout package: raw and processed data with the configuration record; reference comparison, repeatability, valid-coverage and failure maps, an uncertainty discussion sized to the task; a short demonstration of install, reference check, loading and output inspection; the complete-unit parts list; a written recommendation: buy, repeat one bounded test, or close.
/docs/gtm/private/, sales-flock sign-in and passcode). Nothing on the private build is a
public claim.marketing/sales/claims-boundary.md govern
every sentence; no accuracy figure, no frame rate, no shipped-instrument implication, no
"patent pending" for unfiled applications. The P1 card is research with no revenue row.