Development Partner for AI-Curated Amazon Affiliate Platform (WordPress/OpenAI)

1. Project Overview 4indjoy.com is an AI-assisted, fully curated Amazon affiliate platform targeting the Indian market, focused on artisanal categories: spices, jewelry, natural cosmetics, leather goods, hair care, textiles & scarves, honey, essential oils, ghee, and herbal teas. Rather than a conventional product-grid storefront, the platform is built around a conversational AI search experience: customers describe what they need in natural language, and the system returns a short, engaging lifestyle-style response alongside genuinely matched products. Every completed purchase routes through Amazon via the Amazon Associates affiliate program — the platform holds no inventory and handles no fulfillment, payments, or customer support directly. This tender is for a development partner to design, build, and deliver the MVP of this platform, with a strict launch target of August 31, 2026. Business Model • Customers browse or converse with the AI search on 4indjoy.com. • Clicking through on a product redirects to Amazon, where the purchase completes. • 4indjoy.com earns an Amazon affiliate commission on qualifying purchases (governed by Amazon's standard 24-hour referral cookie). • No warehousing, shipping, returns handling, or payment processing is required on our side. 2. Scope & Phasing This is a deliberately phased build. This tender covers Phase 1 in full. Phase 2 is included for roadmap context — agencies are welcome to provide an indicative estimate for it, See attached document for the Phase 1 details and Phase 2 roadmap. Important clarification for proposals: the AI search feature must query only our curated WordPress database — never the live Amazon catalog directly — for matching products to a customer's question in Phase 1. The Amazon Creators API is used strictly for refreshing data on products we have already added, not for query-time product discovery. 3. Technical Architecture & Stack Required / Preferred Stack • CMS & Database: WordPress • Workflow Automation: Make.com (Integromat) — orchestrating data flow between WordPress, OpenAI, and Amazon • Conversational AI: OpenAI API • Commerce Data: Amazon Creators API Important — Amazon API Note Amazon's Product Advertising API (PA-API v5) has been deprecated: Amazon stopped maintaining it on April 30, 2026 and fully retired the endpoint on May 15, 2026. All integration work for this project must target the Creators API, Amazon's current replacement. Proposals referencing PA-API will be considered out of date. Agencies should note that the Creators API is access-gated: an approved Amazon Associates account needs 3 qualifying sales within its first 180 days to gain initial access, and roughly 10 qualifying sales in a trailing 30-day window to keep that access active. New credentials are also rate-limited (starting near 1 request/second and ~8,640 requests/day, scaling up with sales volume). We understand this may affect when live pricing data becomes available after launch, and expect proposals to account for this rather than assume uninterrupted live API access from day one. Frontend Approach We anticipate WordPress remaining the CMS and catalog layer, with the conversational search interface and interactive India map built as embedded components (e.g. React) inside the WordPress theme rather than through standard page templating. We are open to an alternative stack if an agency believes it meaningfully improves outcomes within the timeline — any such proposal should include clear justification, given that reusing our existing codebase (see Section 7) and keeping the platform manageable by a non-engineer team are both priorities. 4. Core Features — Detailed Requirements 4.1 Manual Product Management (Phase 1) • An admin interface within WordPress for our internal team to add, edit, and remove products without developer involvement. • Fields per product should include (at minimum): Amazon ASIN/category code, 3–5 public tags, ~20 hidden parameters (industry, mood, purpose — used by the AI for query matching), category, and status (active/inactive). • The interface should make it straightforward for a non-technical team to maintain the catalog day to day. 4.2 Amazon Creators API Integration • Automated import scoped to specific, approved Amazon category codes/ASINs — not open-ended category scraping. • Scheduled refresh of price and availability data for products already in our database, respecting Creators API rate limits. • Proposals should state a recommended refresh interval (e.g., hourly, every few hours, daily) and reasoning, rather than assuming continuous real-time fetch on every page view. • Outbound product links must correctly pass Amazon's 24-hour affiliate cookie. 4.3 Conversational AI Search • Core mechanic: the OpenAI API returns a structured JSON response per query, containing (a) answer_text — a short, engaging, conversational lifestyle-style reply, and (b) product_tags — the matched internal tags used to pull corresponding products from our WordPress database. • Frontend must display the conversational reply first, followed immediately by the matched product cards, in a clean chat-style interface. • Search must operate strictly against our curated database (see Section 2) — never a live, unfiltered Amazon catalog search. • Fallback behavior is required for queries the catalog cannot serve well — e.g., a request for an item or ingredient outside our curated categories. Rather than forcing a weak or irrelevant match, the AI should respond gracefully (acknowledging the gap, and where sensible, offering our closest genuinely relevant alternative). Proposals should describe the intended approach to this fallback. • Holiday/religion redirect: the platform does not carry religious or classic-holiday-themed products. If a customer asks for something in that space (e.g. "a birthday gift" or "something for Christmas"), the AI should smoothly redirect toward a relevant non-holiday alternative rather than declining unhelpfully or returning nothing. • "More inspiration for you": a secondary set of suggestions should always appear beneath primary AI results, so the customer isn't left with a dead end if the top matches aren't quite right. Proposals should describe the intended logic (e.g., same-category suggestions for MVP, with more advanced matching as a possible future enhancement). • Tone: responses should read as warm, positive, and genuinely helpful — closer to a knowledgeable friend's recommendation than a generic chatbot reply — and may include light educational/cultural detail where relevant. 4.4 Website Architecture and Filters • Schema.org Product Markup: To ensure the platform is the best in the market, the backend must highlight key data for Google SEO optimization. The architecture must include max. 5 key specifications (e.g., price, brand, availability, weight, organic certification) so that Google can index them perfectly for rich snippets. • Dynamic Faceted Search: Once the user is on 4indjoy.com, we want to provide an expansive filtering experience offering up to 20 detailed specifications. The backend must support dynamic custom attributes/metafields so that our team can flexibly add highly specific filters based on category needs (e.g., Vegan, Gluten-Free, Packaging Type, Spice Level, Region of Origin, etc.). 4.5 Interactive India Map • A visual, clickable map overlay (opens on partial screen, closable) where selecting a state surfaces regional artisan products, with the AI assistant panel acknowledging the selection. 4.6 Category Browsing • Standard top-menu category browsing (e.g., Spices, Leather) should trigger the same assistant-panel behavior as the map — consistent experience across both discovery paths. 4.7 WhatsApp Sharing • One-click share button generating a pre-filled WhatsApp message with a link back to our own product page — never a direct Amazon affiliate link, per Amazon's program rules on private-channel link sharing. 4.8 Compliance • A clearly visible Amazon affiliate disclosure statement in the site footer ("As an Amazon Associate, we earn from qualifying purchases"). • No hardcoded prices anywhere in AI-generated or static text — all pricing displayed must trace back to the refreshed database values described in Section 4.2. 5. Existing Assets • A partial codebase from a prior project (including a working map-of-India element) is available for review. The selected agency will assess which components are worth adapting versus rebuilding from scratch. • Final logo and brand guidelines will be provided at kickoff. 6. Timeline & Milestones Target MVP launch: August 31, 2026. We expect proposals to include a week-by-week delivery plan against this date, along with any risks or dependencies (on our side or the agency's) that could affect it. Milestone Payment Trigger Milestone 1 30% Deposit, upon contract signing, code handoff, and project kickoff. Milestone 2 30% Delivery of a fully functional staging demo (conversational search + Creators API integration working end to end). Milestone 3 40% Successful deployment to production, flawless operation, and handover of all assets. We operate on a strict milestone payment structure to maintain project momentum. No 100% upfront payments will be considered. 7. Agency Requirements & Deliverables • A proven portfolio demonstrating previous OpenAI API integrations and WordPress development. • At least two examples of past projects utilizing OpenAI (or comparable LLM) API integrations. • Demonstrated ability to work within a milestone-based payment structure and commit to firm deadlines. • A single accountable point of contact for the duration of the engagement. 8. Questions Your Proposal Must Address In addition to cost and timeline, please directly address the following in your response: 1. Your proposed refresh interval/strategy for Amazon Creators API data, and how it stays within API rate limits as the catalog grows. 2. Your proposed approach to fallback behavior when a customer's query falls outside our curated catalog (e.g., an ingredient or product type we don't yet carry). 3. Your proposed logic for the "more inspiration" suggestions shown beneath primary AI search results. 4. If you are recommending a technology stack other than WordPress + Make.com + OpenAI + Amazon Creators API, your justification and how it affects reuse of our existing codebase. 5. Your approach to keeping the AI's tag-matching accurate and fast as the catalog grows beyond the initial ~100 products. 6. Any dependencies you anticipate on our side (e.g., Amazon Associates account status, asset delivery timing) that could affect the August 31 date. 9. How to Apply Please provide: • Your proposed total cost for the Phase 1 MVP as scoped in this document. • Your estimated week-by-week timeline to hit the August 31 deadline. • Answers to each question in Section 8. • At least two examples of past projects utilizing OpenAI API integrations. • Optionally, an indicative estimate for Phase 2 (AI-driven curation, described in Section 2) for future planning purposes — not required for this tender. We are happy to arrange a call for any agency requiring further information before submitting a proposal.