Changelog.

End-of-day 2026-04-25 sprint covering ~14 ships beyond the earlier today entries: /industries/podiatry (16th — diabetic foot recall, Medicare A5500 eligibility), /industries/urgent-care (17th — 911-redirect protocol for acute symptoms, live wait-time integration), /industries/pediatrics (18th — under-3-month fever red flag, vaccine-record lookup, well-child queue), /industries/physical-therapy (19th — MVA / Workers Comp special intake, recurring-visit cadence booking with EHR pull); Pattern Library 92→93 with medicare-flex-card-scam (CMS+FTC+AARP, ~250× volume increase since 2024 AEP) and pet-recovery-scam (BBB 2024, grief-state-targeting); 4 new compare pages (/compare/goodcall, /compare/numa, /compare/bland-ai, /compare/retell) bringing total to 11 vendors; /security page with HIPAA BAA process + 8-row sub-processor table + vuln disclosure safe-harbor; /about page with founder profile + Q1 2025→Q2 2026 timeline; /press kit page with 4 story-angle panels + brand assets + 3 attribution-ready quotes; /resources content discovery hub aggregating 22 internal links across 6 categories; /shield page Recent threat additions credibility band; /pricing FAQ expanded 5→10 entries + FAQPage schema for rich snippets; homepage hero refresh adding 19-vertical badge + scam-screening differentiation in subhead + 19 inline above-the-fold links; 2 new blog posts (how-to-pick-ai-receptionist-2026 buyer's guide; senior-living-ai-call-screening-procurement-guide-2026 targeting Shield buyer); footer Library + Company columns get new /resources, /about, /press, /security, /blog links across all marketing pages. Net: 19 vertical landing pages, 93-pattern Shield library, 11 competitor comparisons, 5 blog posts, full trust-page set (security/about/press), sitemap at 245+ URLs.

Five ships extending the launch-week revenue surface: (1) /industries/pharmacy (13th vertical) — 7 call patterns including DEA-compliance no-verbal-Rx-via-AI boundary, 8 dispensing-system integrations (PioneerRx, McKesson EnterpriseRx, QS/1, Liberty, Computer-Rx, PrimeRx), HIPAA BAA + controlled-substance posture FAQs. Targets 21k US independent pharmacies. (2) /industries/optometry (14th vertical) — 7 call patterns including VSP/EyeMed real-time benefit lookup, annual-recall outbound automation, contact-lens reorder end-to-end with Rx expiration check, surgical-coordinator routing, acute-eye-complaint protocol. Targets 35k US optometrists. (3) /blog/phone-scam-patterns-q1-2026-report — Q1 2026 field report from the 91-pattern Voipy Shield library: AI-tooled patterns now in production (voice-cloned grandchild, virtual kidnapping, SIM-swap-then-call, phantom hacker, AI-deepfake celebrity-endorsement), 7 new categories from FTC/FBI 2024-2025 (QR-code, wrong-number pig-butchering, Medicaid unwind, ISP impostor, discount-medical-plan, Airbnb host, Chinese consulate), Spanish-speaker targeting, 3-layer Shield defense posture. Targets shareable launch-week traffic + Shield conversion. (4) Homepage #industries section + nav link — closes a real conversion gap where /industries/<slug> tuned pages were unlinked from /, so Google-driven prospects would bounce. Grid of all 14 verticals with HIPAA badges where applicable + 'Don't see your vertical?' fallback. (5) Stale credibility-strip copy refreshed from 5 pilot verticals to all 14.

/industries/insurance-agency added as 12th vertical, alongside the prior /industries/therapy-practice ship — independent insurance agencies are a high-volume phone-call business (every COI request, FNOL/claim, quote-shopper, and policy-question call hits the same line), and producers waste 40% of quote time on out-of-appetite risks. New page covers 6 call patterns (new quote, COI, FNOL/claim, policy-question, mortgage-co verification, cold solicitation), 9 integrations (AgencyZoom, EZLynx, Applied Epic, Vertafore AMS360, NowCerts, HawkSoft, HubSpot, Salesforce, Zapier), and 5 FAQ entries (AMS read-only, no binding/licensure boundary, carrier-appetite filter at onboarding, FNOL routing, TCPA + state-DOI compliance posture). Both new verticals added to /trial form dropdown + verticalPrompts copy map and to /sitemap.xml so Google crawls them.

Three revenue-track ships: (1) New /integrations page surfaces the 8 webhook revenue features already in main — CRM connectors (HubSpot/GHL/Salesforce), immutable audit-log webhook, real-time transcription webhook, AI disposition tagging, missed-call rescue + attribution, outbound reminder calls, review-request SMS, Shield threat-intel feed — with pricing tiers, architecture diagram, auth + delivery semantics, and Schema.org Service JSON-LD. Cross-linked from pricing tier features and Product footer columns. Targets the API/Enterprise wedge (50 × $2k/mo = $100K MRR path). (2) Pattern Library 90→91 with ai-celebrity-endorsement-scam (FBI IC3 + FTC 2025 imposter-investment fraud + SEC 2024 advisory; AI-deepfake video ads of Musk/Cuban/Buffett endorsing fake crypto/AI-trading funds; fastest-growing subset of $5B+ FTC reported losses). EN + ES openings, full pattern card with FINRA BrokerCheck + SEC IAPD shield citations, JSON-LD ItemList entry. (3) Demo page conversion improvements: email lead-magnet reframed as family-share for adult-children-of-seniors target buyer; new 'Email this to mom or dad' mailto: viral-share button on completed demos (each demo can fan out to 1-3 family forwards); homepage FAQ section with 7 high-objection answers + matching FAQPage Schema for Google rich snippets; stale 81/83/84/89/90 count drift across 6 surfaces resynced to live 91.

Library grew 83→90 with new FTC/FBI/HHS-flagged gaps: discount-medical-plan-scam (FBI PSA250430, fake 'health plans' = discount cards), chinese-consulate-impostor-scam (FBI 2018/2022/2024, $100M+ Chinese-American diaspora), wrong-number-pig-butchering-scam (FTC 2024 $3.9B, SMS→crypto friendship hook), isp-impostor-scam (FCC/FTC 2024 $42M, Comcast/Spectrum/Xfinity), qr-code-callback-scam (FTC 2024 quishing, parking-meter/USPS/IRS QR stickers), medicaid-unwind-recertification-scam (HHS-OIG post-COVID unwind exploit), airbnb-host-impostor-scam (BBB/FTC 2024 off-platform Zelle pressure). Also: 5 flagship multi-turn dialogues (grandchild / virtual-kidnapping / sim-swap / phantom-hacker / irs-arrest) with session-sticky A/B variant shuffle, featured block on /demo, 27+ scenarios rewritten natural first-person, classifier keywords expanded 20→50+. 60+ commits, production stable at 20-concurrent p95 2.6s.

The same classifier powering Voipy Shield is now available as a metered REST API. POST /api/shield/classify with Bearer auth returns BLOCK / QUARANTINE / HOLD / ALLOW decisions + rich signals (claimed_organization, impersonation_suspicion, deepfake_or_synthetic_suspicion, asks_for_sensitive_info, …). Per-customer keys with 60 req/min default rate limits, monthly usage CSV for invoicing, 99.9% SLA. Public reference at /docs/shield-api includes a zero-signup 'Try it now' widget (4 one-click scam scenarios: IRS suspension, grandchild-in-jail, account-verify phishing, legit appointment). Pricing: 2-week pilot free (10k req cap) → $0.005/req metered → Custom enterprise with dedicated/on-prem deploy option. Instrumented: new /metrics counters voipy_shield_api_requests_total / errors_total / keys_active for 24h-window observability. Shipping as beta — email anton@sdnbros.com for a pilot key.

Library +4 scenarios (ssn-suspension-scam per SSA-OIG #1 phone scam, senior-official-clone-scam per FBI PSA 250515 AI-voice-clone, medicare-dme-scam per HHS-OIG DME fraud alert, phantom-hacker-scam per NY AG $1B+ coordinated 3-phase ruse). Production UX audit resolved: mobile-nav hamburger across index/shield/pricing (≤1024px), card-required copy unified, plan names Solo/Practice/Enterprise consistent, placeholder testimonials replaced with credibility stats, Command Center dashboard mockup (pure-CSS 7-day bar chart + 5-row call log with action pills), section padding tightened 120→88px, ROI Calculator transparency with live formula + 34% benchmark citation, demo redesign (auto-select+prominent Start, ghosted transcript sample, collapsed accordions by default, post-call summary card with intent/spam-score/action + trial CTA, how-it-works ⓘ tooltip, sticky transcript panel on desktop), 4-col footer propagated to all 4 pages, Shield nav now signals 'senior-living' subproduct. New lint: scripts/lint_scenario_tokens.py catches scenarios that would exceed Qwen 7B 2048 ctx in ES mode (would have prevented the 99e2d37 IDIOMA-expansion regression before it reached production). ES sweep quality: 77→81/83 clean (98%), 2 persistent leaks fixed via surgical Example: rewrites. TLS cert: self-signed → Sectigo. 35+ commits, 0 extended outages.

The library grew from 14 to 79 voice-call patterns over the last day and covers every major FTC / FBI IC3 / SEC / CFTC / FCC / FEMA / CFPB / ALTA / state-PUC advisory category across 9 catalog sections: government impersonation (15), financial & account takeover (16), tech support (3), business/B2B fraud (4), employment & income (6), online-purchase (3), everyday-volume (12), emotional/rapport (11), and debt/investment/recovery (11). Every scenario ships in English + Spanish with native-speaker voices. Sweep latency p50 1.0s / p95 2.1s / p99 2.5s; vLLM continues-batching tested to 20 concurrent calls without degradation.

Coinbase / crypto-exchange fraud alert (seed-phrase & 2FA phishing), grandchild voice-clone bail scam (FBI IC3 fastest-growing AI fraud), and SIM-swap attack on a carrier fraud-verification line ship this week — bringing the library to 14 voice-call patterns. Plus: EN/ES language toggle on /demo (same scenarios, Google Neural2 Spanish voices) doubles addressable demographic at zero cost increase. Plus: OpenGraph + Twitter cards on /demo, /scam-catalog, /smishing with a branded shareable OG image. Post-call 'What Shield caught' tells panel emits 2-3 cited red flags on every ended call + shareable copy/X buttons.

The Library is published as a browsable catalog now: 12 voice-call patterns grouped by category (Government impersonation, Financial & account, Tech support, Everyday-volume, Emotional/rapport, Debt & investment) + 7 SMS-only patterns at /smishing. Each pattern links to the live /demo, cites its FTC/FBI/AARP/SEC/FDCPA source, and names the specific tells Shield uses. Next release (2026.05) adds grandchild-in-jail voice-clone variant, Medicare Open Enrollment impersonation, Coinbase-exchange impersonation, and international-courier redelivery.

New page at /smishing walks through the top SMS scam patterns (unpaid toll, USPS/UPS delivery, bank fraud alert, wrong-number pig-butchering, fake job offers, IRS refund, Amazon account compromise) with real example texts from FTC/AARP case reports and a sender-number decoder chart (5-digit short code vs 10-digit spoof vs international prefix). Part of the Shield Pattern Library content.

Scam scripts mutate every 4–8 weeks (2026 Social Security COLA twist, new bank 'safe account' variants, pig-butchering platform pivots). Starting this month, every Shield account — free tier included — gets new signatures the month they land. First drop covers Microsoft tech-support, Social Security / COLA, bank safe-account, Amazon 'unauthorized order', romance→crypto, and 'you've won' prize fraud. Sources cited per pattern (FTC Consumer Sentinel, FBI IC3, AARP Fraud Watch).

Pick a scenario, click a dialog line, hear the real LLM answer live — no signup, no phone. Eight scenarios across the Receptionist and Shield product lines including Microsoft tech-support, bank 'safe account' fraud, Social Security COLA, Amazon impostor, romance/pig-butchering, and prize/sweepstakes scams.

Plumbing/HVAC/electrical contractors and hair salons now have their own /industries page — pain points, call patterns, and FSM/PMS integrations tuned to each trade. That's 10 verticals covered.

/shield now has the same 'try the AI' widget the receptionist landing has, with facility-flavored presets (grandparent scam, Medicare suspension, IRS warrant, legit family, pharmacy callback). Results show Shield's decision (Block / Escalate / Allow) in under 300ms.

Three new /compare pages for the AI-native receptionist competitors prospects actually evaluate against Voipy. Factual-only: pricing model, time-to-live, what's included turnkey vs what you build yourself.

Home page language was still dental-specific ('missed patients', 'booked cleanings'). Updated to reflect the 8 verticals Voipy actually serves — dental is still one of them, but so are vet, law, med-spa, accounting, chiropractic, real-estate, auto-repair.

Every paying customer now has a unique share link. When a friend signs up via your link and converts to paid, you both get a month's credit on your next invoice.

Customers now get a Monday-morning summary email: for receptionist, how many calls we answered + appointments booked; for Shield, scam calls blocked + high-risk alerts. Reply to the email to change cadence.

If you sign up and don't complete setup in the first 24 hours, we send one short email with the 3 things to do to get to 'live'. Opt out by finishing setup or by replying.

Landing pages tuned per vertical — per-industry call patterns, PMS integrations, and intake scripts. Pick the one closest to your practice for a more honest sense of fit.

When a customer cancels, they now get one email from Anton asking what didn't work. No coupon-bait, no 'wait wait we'll give you 30% off', no drip sequence. Replies go to his direct inbox.

Stripe fires a warning 3 days before your trial converts to paid. We now forward that to you as a short email with a one-click Cancel/Extend link — so nobody gets surprised by a charge.

Voipy Shield (AI scam + deepfake call screening for senior-living facilities) moved to its own landing at /shield. We ship an authenticity model that runs in parallel with transcription — see the engineering note on the blog.

Text 1-866-879-8984 to find a barber, dentist, auto shop, or restaurant near you. No app, no browser. Each directory listing is a page on /d that business owners can claim and upgrade to a Voipy phone line.