Veriti AI: Pre-Seed Memo

1.The problem and the proof

Precision CNC shops run on judgment that lives in one or two heads. A senior estimator knows which tolerances matter, which vendors to trust, when material substitutions are safe. That judgment doesn't make it into the systems that run the shop. When the estimator retires, the shop loses pricing discipline, quote consistency, and margin control with them. Quoting is where the loss becomes visible to the customer first: quotes slow down, assumptions vary by shift, margin leaks.

Veriti captures that judgment as the shop works. A drawing goes in; a source-linked feature map, an estimator-correctable quote, and a record of why the price came out the way it did come out. Every correction goes into the shop's growing configuration; every assumption traces back to its source in the drawing.

The locked benchmark on one of Dun Rite's real customer drawings (a 7.50 × 20.75 17-4 PH H1150 stainless drive shaft, 3-unit lot) captures 19 of 20 pricing-critical fields and lands at ~$10.3K against Dun Rite's target band of $9–11K.

By M12: 6–8 paying customers and $150–225K ARR, with 3–4 outside the initial design-partner network. Quoting in production with paying customers running daily workflows; purchasing workflows in design-partner discovery with one customer in alpha. 4–6 closed-loop field deployments, 50%+ onboarding automation, and 55%+ SaaS gross margin after onboarding (path to 65%+ by M18).

2.The product

The wedge is 2D-drawing-native quoting. A shop uploads a customer drawing. Veriti reads it through a multi-pass extraction pipeline, produces a source-linked feature map, and applies the shop's pricing rules to compile a quote. The estimator reviews in a chat-driven workstation: drawing viewer with numbered-circle overlay, quote page with operations × hours × rate and material/margin breakdown, setup panel for overrides.

Veriti maintains the decision trail for whyeach quote came out the way it did: what the drawing said, what the shop assumed, why a material was selected, how a vendor price was used, what a human approved. It works alongside the shop's existing ERP rather than replacing it.

Veriti stores the shop's quoting decisions, assumptions, corrections, and approvals in one shared context layer that future workflows can build on.

Roadmap. Quoting M0–M9 from calibration to production. Quoting hardening + purchasing workflows design-partner discovery M9–M12. Purchasing workflows alpha → production M12–M15. Routing workflows later.

3.Why now

The strongest pain in precision CNC is workforce: recruitment, retention, and the looming retirement of senior estimators and owners. Quoting is the first place that pain hits the customer-facing edge of the business.

Drawing extraction, reasoning, human review, and source-linked evidence now combine into a practical workflow that ships in months. Each capability crossed the threshold within the last 12–18 months. Earlier extraction couldn't read GD&T reliably; reasoning models couldn't chain pricing logic; chat workstations weren't trusted by domain experts.

4.The judgment gap

Manufacturing software historically captures transactions, not judgment. ERPs record what happened after decisions were made. CAM systems define toolpaths after process selection. MES systems track execution after routing is approved. The highest-value operational knowledge in a precision shop lives upstream of all of them, in quoting assumptions, manufacturability judgment, vendor selection, tolerance interpretation, routing tradeoffs, and estimator corrections.

Veriti starts at the first decision point, before the ERP, CAM system, or MES is involved. Every correction the estimator makes, every rule the shop establishes, every vendor preference and tolerance call goes into the shop's decision trail at the moment the decision is made, not after it's been executed.

5.From wedge to platform

Veriti is not a quoting company. Quoting is how Veriti gets in the door.

Purchasing workflows and routing workflows ship into the same customer base on the same architecture, expanding ACV without re-acquisition. Each workflow captures a why: quoting why the price was set, purchasing why the vendor was chosen, routing why the work order looked the way it did.

Day-one is 1,170 precision machine shops in the Great Lakes corridor; aligned precision categories (precision turning, tool & die, fabricated metal manufacturing) bring the corridor total to 5,309. At quoting ACV in the Great Lakes alone, the 5-year target is $5–10M ARR. (NAICS 332710, 332721, 333514 plus aligned codes, per U.S. Census County Business Patterns.)

6.Where Veriti fits

The bet.Frontier models will commoditize drawing extraction within 24 months. Every quote a shop runs through Veriti deepens what's inside: estimator corrections train future quotes, pricing rules become the shop's working spec, vendor and material decisions become context the next quote uses. Veriti becomes how the shop quotes.

The segment. Existing quoting tools in this tier are template- and rules-based, not AI-native. The wedge is AI-native quoting without the multi-month implementation or the dedicated configuration owner.

7.Team

David Blank, CEO / Product.Marine veteran (Aircraft Mechanic, USMC), awarded the Navy and Marine Corps Achievement Medal for building repeatable systems. Three years on the shop floor at Tejas Tubular and Lippert as EHS Manager, converting operational complexity into dashboards, 5S programs, and process systems. Eight years at Lippert across Salesforce, conversational AI, and senior systems roles. Quoted in Forbes on enterprise AI agent adoption. Building Veriti's codebase.

Ryan Banks, COO / Sales.President of Dun Rite Machining, Veriti's design partner. Five years at Lock Joint Tube (Lerman Enterprises, $85M facility) where I drove the operational turnaround that grew EBITDA from $500K to $8M peak between 2019 and 2022, with a 35% throughput improvement. 25+ years running precision and heavy-industrial plants as PM, VP Ops, GM. Navy nuclear veteran.

AI Engineer hire triggers on signing of paying customer #2 (non-Dun Rite); in seat ~month +1 (codebase ownership, field on edge cases). Marketing Operator hire triggers on signing of paying customer #3 (non-Dun Rite); in seat ~month +1 (operator-channel content, events, capture kit, case-study flywheel).

8.Go-to-market

Day-one geography is the Great Lakes precision-manufacturing corridor. Founder network, design-partner work, and dense regional shop geography concentrate market access. A founder in Michigan can run six shop-floor deployments in a quarter without spending company cash on flights; a founder trying to cover California, Texas, and Pennsylvania at once cannot. Density makes referrals compound; geography is the unfair advantage. West/Southwest Michigan and Northern Indiana first; Chicago/Northern Illinois, Detroit/Southeast Michigan, and Northern Ohio next.

Primary ICP. U.S. precision CNC shops, 10–100 employees, $5–20M annual revenue, 50+ RFQs per month, estimator bottleneck or owner/senior-estimator knowledge-transfer risk.

Pricing includes paid implementation. LAND ($15K ARR) is base quoting for one shop. MID ($30–40K ARR) adds tariffs, vendor pricing, DFARS trail, material substitution. PURCHASING DESIGN-PARTNER ($40–55K ARR, alpha-grade pricing only) adds purchasing workflows. Field deployment is $5–15K one-time.

LAND at $15K clears a shop owner's signature without procurement review and trains the buyer to use the product daily. Free pilots train shops to expect free; $1K/month pilots train them not to use it.

Quarterly tracking of implementation hours per customer, percentage of onboarding steps automated, gross margin after onboarding, and deployment payback; all must improve materially versus baseline by M12.

9.Use of funds

$1.0M at $5.0M post-money. 20% new ownership. 12-month proof plan with a 3-month extension reserve.

10.Risks

1. Frontier-model commoditization of extraction. As foundation models improve, raw drawing extraction becomes commodity. Mitigation: the defensible layer is per-shop, not extraction itself.
2. Competitive response from Paperless. Paperless could move down-market with a stripped-down tier. Mitigation: build the estimator-correction loop and quote-to-purchase decision trail fast enough that down-market Paperless feels like a different product, not a cheaper version of the same one.
3. Wedge fails to convert outside Dun Rite.Hires are gated on paying customers #2 and #3 (non-Dun Rite). If the wedge doesn't bite outside Ryan's network, hires don't fire and the founder team carries the M12 work. Mitigation: founder-led embedded deployments in the corridor; pipeline through regional manufacturing-association outreach; scope-down rather than hire-against-hope if conversion stalls.
4. Implementation-tax risk.Deployment hours don't drop fast enough to clear the 55%+ gross-margin gate at M12. Mitigation: the quarterly deployment gate forces scope-down before headcount.
5. ITAR / CUI exposure. Aerospace/defense shops handle export-controlled drawings; Veriti is not yet ITAR-eligible. Mitigation:the path (self-hosted open-weights model on AWS GovCloud, NIST 800-171 self-attestation, ITAR registration + Tech Control Plan, tenant isolation) is scoped post-PMF on a 3–4 month Vanta-accelerated timeline at $80–150K. Not on the pre-seed critical path; the M12 plan doesn't commit to ITAR customer counts.