Changelog
Release notes for Apiway, newest first. Covers the platform-build period in early 2025, the AI-fashion pivot and public beta in late 2025, and ongoing 2026 launches — new templates, infrastructure, marketplace, billing, and structured-data improvements.
Last updated: . 17 entries.
Apiway now ships dedicated `/for/<slug>` pages for Shopify clothing brands, Amazon FBA sellers, Etsy clothing shops, DTC fashion brands, multi-brand marketplaces, and photographers transitioning to AI. Each page emits its own Service + WebPage JSON-LD with audience-typed `audience.audienceType`, recommended workflows as a hasOfferCatalog, and the shared legal-FAQ block on commercial-use rights.
The pricing page now emits seven atomic Product entities (Free, Starter, Basic, Pro, Studio, Business, Scale) wrapped in an ItemList, with priceValidUntil, eligibleRegion, hasMerchantReturnPolicy, and additionalProperty fields per Offer. AI shopping engines can now cite specific plans and credit allocations directly instead of resolving the page as a single bundle.
Anton Viborniy, founder and CEO of Apiway, now has a dedicated profile page at /team/anton-viborniy with full Person + ProfilePage JSON-LD, sameAs links to LinkedIn and Instagram, and a bio explaining the 2026 pivot from B2B workflow automation to AI fashion photography. All 100+ blog posts now reference Anton as the named author with rel='author' linking back to the profile.
The /glossary page ships canonical definitions for ghost mannequin, virtual try-on, AI fashion model, White Studio shot, hybrid AI pipeline, plastic look, creator marketplace, right of publicity, DMCA, and 7 more. Each term is wrapped in DefinedTerm JSON-LD inside a DefinedTermSet so AI answer engines (ChatGPT, Perplexity, Gemini) can cite Apiway as the source for 'what is X?' queries.
Apiway now ships individual comparison pages versus 25 competitors — including FASHN AI, Higgsfield, Photoroom, Freepik, Figma Weave, and the underlying engines (Google Gemini, ChatGPT Image). Every comparison includes a feature table, audience scenarios, an FAQ, the Apiway differentiator block (fashion-only, hybrid pipeline, creator marketplace), and a shared Commercial-rights & legal section covering copyright and right-of-publicity exposure.
Batch Creation lifted the per-session cap from 20 to 50 garments. The change makes Apiway viable as the operating mode for marketplace catalog onboarding, Q4 ramp-ups on Amazon FBA, and weekly 50-SKU drops on Shopify Plus. Sessions still run unattended — submit a batch, return to a finished set.
Edit with Paint now supports up to 5 mask regions per pass for region-level iteration on existing shots — sleeve recolor, label removal, background cleanup, accessory swap, all in one generation. Pairs naturally with hero ad creative iteration without scheduling a re-shoot or a Photoshop pass.
The AI Photoshoots / White Studio template now ships 80 curated poses across 7 framings (Full Length, Three-Quarter, Half Length, Leg Crop, Close-Up, Seated on Chair, Seated/Floor), 9 aspect ratios (Shopify 4:5, Amazon 1:1, social 9:16, lookbook 16:9, and more), and supports up to 16 poses in a single generation. Output is a true RGB(255,255,255) background — Amazon-spec verbatim — with no Photoshop pass.
Virtual Try-On combines a person photo with a garment image to preview how the garment looks on the subject. Particularly powerful for inclusive sizing and modest-fashion shops where buyers want to self-identify with the body type before purchase. Available across the Starter through Scale plans.
Apiway ships its public beta after a 10-month pivot from the predecessor B2B workflow-automation platform. The beta opens with five core templates — Ghost Mannequin, White Studio (AI Photoshoots), Reference Photoshoots, Image Creation, and Edit with Paint — and a credit-metered billing model where 1 credit = $0.01 USD across all plans.
Ghost Mannequin and AI Photoshoots now run on a dedicated Docker generation-worker pool (10 containers, 64GB RAM, AWS S3 storage) instead of the Render web process. The Next.js app and the worker communicate via a Supabase generation_jobs table — a clean rollback path is preserved by toggling the OFFLOAD_TEMPLATES environment variable. Result: 5–8× faster batch throughput on the heaviest templates.
First version of Batch Creation goes live for internal alpha testers. Initial cap is 20 garments per session — enough to validate the unattended-batch operating mode against real catalog workflows. Public beta planned alongside the Q4 product launch.
The Creators marketplace concept ships in a private cohort: photographers and models can upload curated photo sets, set a per-generation price in credits, and earn payouts as brands generate AI imagery against the sets. Each generation pays the creator at the listed price; balances are pegged 1:1 to US cents (1 credit = $0.01).
The Reference Photoshoots template — Apiway's hybrid pipeline where a real photographed creator is the visual anchor and AI generates only the garment overlay — enters alpha. Internal QA confirms the hybrid pipeline avoids the 'plastic look' that pure-AI human generation produces.
First end-to-end working build of the White Studio template — segmentation, recomposite onto RGB(255,255,255), and a tone-correction pass produce true Amazon-spec output. Alpha confirms the design hypothesis: dedicated post-processing eliminates the grey-cast problem that ships from raw image LLMs.
Apiway's billing layer wires up to Stripe with seven tiers (Free, Starter, Basic, Pro, Studio, Business, Scale) and a unified credit accounting model: 1 credit = $0.01 USD, deducted before generation via try_deduct_credits and refunded on failure. Plan transitions and pro-rations are owned by Stripe; the app derives plan from the active subscription's price_id.
The first end-to-end ghost mannequin pipeline runs against a small internal garment set. The prototype validates the core thesis — phone-quality flat-lay or hanger photo in, publishable invisible-mannequin shot out — and seeds the architectural decisions that become the production templates a year later.
