AI Assisted E-content, review and e-commerce Website Creation

Project Summary: Content, Commerce and AI Platform
Public summary for bidders. Brand, category and design details are shared with shortlisted developers after an NDA.
1. Overview
We are building a content website with AI helpers for a consumer audience. It combines decision guides, product pages with verdicts and affiliate links, a stage-based checklist tool, short-video embeds (YouTube Shorts and Instagram Reels), a small merchandise shop, and later user accounts and AI agents. Growth comes from the owner's video channels, which send viewers to the site and later to an app. The website comes first, and a clickable prototype is shared after the NDA.
2. Delivery plan
Phase Includes Target time
1A. Website Public site, CMS and admin, guides, product pages, checklist tool, video embeds, newsletter and messaging capture, SEO, analytics [__ weeks]
1B. Shop module Merchandise shop: variants, cart, payments, orders [__ weeks]
2. Accounts and AI User accounts, tester panel, guide agent, Q&A assistant, community reviews, moderation [__ weeks]
3. App and data Installable web app (PWA) first, messaging-based reminders, permitted marketplace data if available [__ weeks]
Each phase is quoted, delivered and paid separately.
3. Technology
• Preferred: Wordpress/PHP 8 with Laravel and MySQL 8, with a REST API that an app can reuse. WordPress with custom plugins is acceptable for Phase 1if you explain the cost and speed benefit and how it grows into Phase 2 without a rebuild.
• Responsive and mobile-first, Lighthouse mobile performance of 85 or above on the home page, WCAG 2.1 AA accessibility as far as reasonable, SEO-friendly URLs and schema markup.
• Staging and live environments, SSL, daily backups, a code repository owned by the owner, and documented setup.
• Integrations: an AI API (Claude or OpenAI), newsletter (Brevo or Mailchimp), WhatsApp Business platform, payments (Razorpay or similar), Google Analytics and Search Console, YouTube and Instagram official embeds, and an optional CDN.
4. Phase 1A: Website
• Home and guides: editable decision guides with comparison tables, verdicts, "skip it if" notes, compliance notes and affiliate buy buttons, each showing a last-reviewed date.
• Products: about 1,000 products in one category, with image, price in rupees, specifications, verdict and affiliate links. Search, filters (category, brand, price, other attributes) and sorting.
• Prices: updated weekly, each showing a "last checked" date. The developer must advise how to follow Amazon's price-display rules.
• Checklist tool: the visitor picks a stage and sees Buy now, Can wait and Skip lists, editable in the admin.
• Video section: the owner pastes a YouTube Shorts or Instagram Reel link and the site embeds it using the official embeds. Platform filter, optional link to a product or guide, and lazy loading so pages stay fast.
• Capture: newsletter signup, messaging subscription and a tester-panel signup form, with consent text and unsubscribe handling.
• Admin and CMS: the owner adds and edits guides, products, categories, checklists, videos, banners and pages without a developer, with role-based access and an audit log.
• Standard pages: About, Contact (with spam protection), Privacy Policy, Affiliate Disclosure, Terms, and a cookie notice. The owner supplies the legal text.
• Analytics and SEO: Google Analytics, Search Console, sitemap, meta tags, Open Graph images, and tracking of affiliate clicks and signups.
5. Phase 1B: Merchandise shop
• Product page with a design preview that changes with colour, and type, colour and size options.
• Cart, checkout with UPI, cards and netbanking, order confirmation emails, and GST invoices.
• Admin for products, stock, prices, orders and shipping status. Fulfilment partner to be agreed.
• Returns and refund page, and clear delivery times.
6. Phase 2: Accounts and AI
Accounts and reviews. Registration, email verification, password reset, profiles, and reviews (one per user per product, editable and deletable), with a report button, spam protection and rate limits. Quality checks detect copied text, review bursts, very short generic text and one user posting many reviews. Flagged items go to an admin list with the reason, and nothing is deleted without admin action.
Rules for every AI agent
• Use an AI model through an API, swappable without a rebuild. Recommend a model and estimate the cost per use.
• Answer only from owner-approved data, with no invented facts and a source cited in each answer.
• A human approves anything published. Agents draft but do not publish.
• No medical advice. Health or safety-critical questions are redirected to a professional with a standard message.
• Protect against prompt injection, log all agent actions, set daily cost limits, and show a clear "AI-generated" label.
• Do not use user data to train models, and send personal data to AI services only when necessary and disclosed.
Agent 1: Guide agent. The visitor enters a few details (stage, home type, budget). The agent builds a personalised shortlist from approved guides and products, shows its steps, explains its reasons, links to buy buttons, and says when data is missing.
Agent 2: Tester panel agent. It matches testers to product tests by their profile, sends invitations and reminders (messaging or email), collects structured forms and voice notes, transcribes voice notes (English and Indian languages where practical), flags suspicious responses, and drafts a summary of pros, cons and best-for. The owner approves before publication. Free test items are disclosed.
Agent 3: Q&A assistant. A chat helper that uses only approved content, cites the review or video behind each answer, and declines to guess outside its data.
AI admin. Preview, edit, approve or reject every draft, view logs and costs, and switch any agent off.
7. Privacy, safety and compliance
• Follow India's data protection law (DPDP Act): clear consent, data minimisation, deletion on request and secure storage. The owner will take legal advice on final policies.
• Sensitive personal details are collected only with consent and in the minimum form needed.
• Affiliate and sponsored-content disclosures near links and in videos, as Indian advertising rules require.
• Security: prepared statements, input validation, CSRF and XSS protection, hashed passwords, rate limiting, HTTPS only, role-based admin access, and API keys kept out of public code.
• Show only compliance or certification claims that the owner has verified, with the source and date.
8. Data sources
• Allowed: the owner's own content and transcripts, owner-entered products, manufacturer specifications, public safety and regulatory data, affiliate links and feeds the owner is eligible for, user submissions, CSV imports, and permissioned third-party content.
• Not allowed: scraping or collecting data in ways that break a site's terms, third-party video transcripts without permission, and paid commercial review-scraping providers. We will not sign a waiver accepting such risk.
• Amazon API access depends on affiliate sales not yet in place, so affiliate links start as manual links. Build each external source as a separate connector that can be switched off.
9. Possible scope (priced separately)
Native iOS and Android apps, languages English only at launch, price tracking, location-based discovery, and live marketplace review data.
10. Ownership and confidentiality
• All code, design, content and data belong to the owner, with an IP assignment clause.
• The NDA is signed before the brand, prototype and full specification are shared.
• The owner holds the domain, hosting, database, repository, payment, AI, email and analytics accounts. The developer gets temporary access, removed at handover.
• All communication, files and payments stay on the freelance platform.
11. Acceptance criteria
• Every feature in the agreed phase works on the live site on mobile and desktop.
• The owner can add a guide, a product, a video and a checklist, and edit and approve an AI draft, without help.
• Search, filters and sorting return correct results on a supplied test set.
• Each agent answers only from approved data, cites sources, declines health questions, and is stopped by its cost limit in testing.
• Checkout works end to end in test mode and then live (Phase 1B).
• A backup restore is demonstrated.
12. Deliverables
Live site, source code in the owner's repository, database schema, admin guide or recorded walkthrough, API and AI setup notes with prompts, and 180-365 days of free bug-fix support after each phase.
13. Budget and payments
Quote build cost and monthly running cost separately for each phase. Build: Phase 1A [₹ ___], Phase 1B [₹ ___], Phase 2 [₹ ___], Phase 3 [₹ ___]. List monthly running costs (hosting, AI usage, email, messaging, payments, tools) line by line. Suggested split per phase: 30% at start, 40% at demo, 30% on acceptance.
14. What each bid must include
1. Links to 2-3 live sites you built, ideally with user accounts, and one AI or agent project if possible.
2. Fixed price and timeline per phase, and monthly running costs.
3. Recommended framework, hosting and AI model, with reasons.
4. How you would build and test the three agents safely.
5. Confirmation that you will sign an NDA and IP assignment.
6. Availability for a short kickoff call.
15. How bids will be judged
Fit with the specification (30%), safe and realistic AI plan (25%), relevant portfolio (20%), clear price and timeline (15%), communication (10%).