Frequently Asked Questions

Screw Genius is an AI-powered fastener engineering platform built for screw factories, mold shops, trading companies, and procurement engineers. Upload a photo of a screw or an engineering drawing (PDF, DWG, PNG, JPG) and within 5-10 seconds the AI identifies head type, drive pattern, thread pitch, outer diameter, length, and material standard (DIN, ISO, ASME, JIS, GB). Beyond identification, the platform includes a Cost Quote Calculator, Blank Diameter Calculator with plating compensation, Reverse Lookup, Die Life Estimator, Torque Analysis, and Vision Memory that learns your factory's correction patterns. The Free tier includes 15 identifications per month; paid plans unlock batch processing and advanced engineering tools.

Screw Genius uses advanced computer vision models (Gemini 2.5 Pro for high-precision engineering drawings and Flash Lite for fast photo analysis) to extract visual features from your image — head type, drive pattern, thread profile, length, and surface finish. The AI then cross-references a curated specification database covering DIN, ISO, ASME, JIS, and GB standards to determine the most likely parameters. Each identification returns a confidence score, alternative candidates ranked by similarity, and a structured output (head type, M-size, pitch, length) you can copy directly into your BOM or quote spreadsheet. For ambiguous drawings, the system surfaces the top 2-3 matches so you can pick the right one in seconds.

Under good lighting and with clear photos, AI accuracy for common screw specifications reaches 85-95%. Accuracy depends on photo clarity, lighting conditions, the screw's angle relative to the camera, and whether the screw follows standard specifications (custom or proprietary heads score lower). Engineering drawings with clear dimensions and annotations typically score above 95%, while phone photos of dusty production samples may drop to 75-85%. We recommend using AI results as an initial assessment, then confirming with physical measurement tools (calipers, thread pitch gauges) for parts entering production. The Vision Memory feature also remembers your corrections, raising accuracy on your specific factory's catalog over time.

Supported screw types include metric series (M1-M64 per ISO 4017, ISO 4762, DIN 933, DIN 912), imperial series (UNC, UNF, UN per ASME B1.1), self-tapping screws (DIN 7981/7982, ISO 7049), machine screws, hex bolts, socket head cap screws, button head, countersunk Phillips and slotted heads, plus specialty fasteners like flange bolts, Torx, security heads, and SEMS screws with attached washers. The database covers DIN, ISO, ASME, JIS, and GB standards with active expansion for ANSI and EN variants. If a head type isn't recognized, you can teach it manually through Vision Memory and the system will identify similar parts on future uploads.

The Free plan includes 15 AI identifications per month, single-drawing upload only, and access to basic engineering tools — torque analysis, die life estimation, thread engagement, and tap drill size lookup. The Pro plan at $9.9/month raises identification to 300/month and unlocks batch drawing analysis (up to 5 parallel), Screw Weight Estimator, Blank Diameter Calculator with plating compensation, Reverse Specification Lookup, and Cost Quote Calculator with material price tracking. The Enterprise plan at $49/month provides 2,000 identifications, batch up to 15 drawings, priority email support, and team-level Vision Memory sharing. Both paid tiers include a 14-day money-back guarantee. See Pricing for the full feature comparison.

Yes. Uploaded photos and drawings are sent over HTTPS, used only for the immediate AI identification call, and are not retained on our servers after processing — there is no archive of customer drawings. Account credentials and Vision Memory data are encrypted at rest in Supabase, which holds SOC 2 Type II certification. Payment information is processed entirely by Paddle, our Merchant of Record — we never see or store credit card numbers. Your Vision Memory corrections are private to your account; no other user, including industry competitors using Screw Genius, can access your factory's identification patterns or correction history. See our Privacy Policy for the full data handling specification.

AI screw identification requires an internet connection because the computer vision models run on cloud GPUs, not on your device. However, several engineering calculation tools work fully offline once the page is loaded — Torque Analysis, Die Life Estimation, Tap Drill Size lookup, Thread Engagement calculation, and Material Properties editor all run entirely in your browser using local JavaScript. If you frequently work in environments with poor connectivity (factory floors, remote sites), keep these calculator pages loaded in browser tabs before going offline. Batch identification jobs queue and resume automatically once your connection returns, so partial uploads are not lost.

Screw Genius supports 9 languages: Traditional Chinese (zh-TW), Simplified Chinese (zh-CN), English, Vietnamese, German, Japanese, Hindi, Thai, and Arabic — covering most major fastener manufacturing regions. The interface automatically detects your browser language on first visit, then remembers your choice. You can switch manually anytime from the language selector in the settings page or the top navigation. All AI identification outputs (head type names, standard codes) are localized into the active language, and our FAQ + Knowledge Base articles are fully translated for Traditional Chinese, English, and Vietnamese. Other languages currently localize the interface only, with English fallback for long-form documentation.

You can cancel anytime from the Account Settings page — no email or support ticket required. After cancellation, you retain access to all paid features until the end of your current billing period, then automatically revert to the Free plan with 15 identifications per month. No data is deleted on downgrade; your Vision Memory, identification history, and saved quotes remain intact. New subscribers to Pro or Enterprise plans qualify for a 14-day money-back guarantee — request a refund within 14 days of your first charge and we will process a full refund through Paddle. See our Refund Policy for the exact terms and refund processing timeline.

Email service@screwgenius.com for technical issues, billing questions, feature requests, or feedback on AI identification accuracy. Free-tier users typically receive a response within 48 hours; Pro and Enterprise users get priority queue treatment with a 24-hour response target. For account-specific issues (subscription, refunds, login), please include your registered email and account ID (found on the Settings page) so we can look up your account quickly. You can also use the Contact form for general inquiries or partnership questions. We do not currently offer phone or live chat support, but we read every email.

For best identification accuracy: (1) Shoot under bright, even lighting — natural daylight or a white-balanced LED desk lamp; avoid harsh shadows from overhead fluorescents. (2) Place the screw on a plain, contrasting background (white sheet for dark screws, dark cloth for galvanized parts). (3) Make the screw fill at least 60% of the frame — don't shoot from far away. (4) Take two photos: one from the side showing the full thread profile, one from above showing the head pattern. (5) Avoid glare on the head — angle slightly to break direct reflection. (6) For engineering drawings, ensure all dimension lines, callouts, and thread notations are crisp and legible. Blurry text drops accuracy by 15-25 percentage points.

Yes — the Pro plan supports uploading up to 5 drawings or photos in parallel, with each one analyzed concurrently and results returned in a single dashboard view. The Enterprise plan raises this to 15 parallel uploads, making it ideal for screw factories and trading companies processing daily quote requests with dozens of part numbers. Batch results can be exported as CSV with M-size, pitch, length, head type, and material columns for direct import into ERP, MRP, or quotation spreadsheets. Batch identification typically completes in 10-30 seconds for 5 drawings depending on image complexity. This workflow saves roughly 90% of the time spent on manual specification entry for high-volume quoting.

Yes — Vision Memory is the feature that records every correction you make on AI identification results. When a similar drawing or part is uploaded next time, the system automatically applies your factory's specific preferences (preferred head type naming, specific tolerance interpretations, material defaults). Accuracy on your specific catalog grows measurably with each correction — typical users see 5-10 percentage points of accuracy improvement after their first 50 corrections. Vision Memory data is completely private to your account: no other user, including direct industry competitors who also use Screw Genius, can read or copy your correction patterns. Enterprise plans add team-level Vision Memory sharing so all factory engineers benefit from collective corrections.

Reverse Lookup lets you input a blank (wire) diameter — for example, 5.5mm of cold-heading wire stock — and the system automatically cross-references all standard screw specifications that can be rolled or formed from that diameter, sorted by deviation from smallest to largest. The output shows M-size, pitch, equivalent imperial size, finished outer diameter, and the percentage stock utilization. This is critical for screw factories deciding which orders can be fulfilled from existing wire inventory versus requiring new material procurement. Combined with the Blank Diameter Calculator (which works the opposite direction — finished spec to required blank), Reverse Lookup pays for the Pro subscription in inventory savings within the first month for most factories.

The Blank Diameter Calculator includes built-in coating thickness data for the most common plating types: Fe/Zn 5-12μm (electroplated zinc), electroless nickel 8-25μm, hot-dip galvanizing 40-80μm, Dacromet 8-12μm, Geomet 8-10μm, and trivalent passivation variants. The system automatically calculates plating impact on pitch diameter using the standard formula ΔD ≈ 4 × coating thickness (accounting for buildup on both flank surfaces), then subtracts it from the recommended blank diameter so the finished screw meets specification after plating. You can override the default thickness for specific plating houses with non-standard tolerances. This compensation is critical for high-strength fasteners — even a 10μm error on M5 threads can push the part out of class 6g tolerance.

Die Life Estimation calculates expected production output based on screw geometry (M-size, thread pitch, head height ratio), screw material grade (C1022, SCM435, SUS304, SUS316, 10B21, and other common cold-heading steels), heat treatment specification, die material (DC53, SKH51, M2 tool steel, and other premium die steels), and whether nitriding surface treatment is applied. The algorithm combines published wear-rate research with the founder's 10+ years of hands-on thread rolling die R&D experience in Taiwanese fastener factories, calibrated against real production data from M3-M16 die runs. Estimates fall within typical production ranges and help factories plan die replacement cycles, quote pricing realistically (die amortization per piece), and identify when material/coating upgrades will pay back through extended die life.