Description
Summary I am looking for an automation engineer to build a single, reliable workflow that turns a customer purchase into a personalised PDF, generated by an AI model and delivered by email. The building blocks already exist: the prompt, the PDF template, the data structure and a full written specification are all prepared and will be provided. This is an integration and wiring job, not a design or content project. I am looking for someone who builds clean, well-tested automations and communicates clearly, not a large agency. In one line: when a customer pays through Stripe, the system pulls their form answers, calls an AI model to generate their personalised content, renders it into a supplied HTML and CSS template as a PDF, holds it for a quick human review, then emails it to them. WHAT YOU WILL BUILD - A Stripe payment webhook that triggers the workflow and identifies the customer by email. - A Make.com (or similar) scenario that looks up the customer record from a Google Sheet, calls an AI model API (Anthropic Claude) with a supplied prompt, and validates the JSON it returns. - A render step that injects the returned content into a supplied HTML and CSS template and produces a pixel-faithful PDF. Custom web fonts and inline SVG graphics must survive the conversion. - A human-in-the-loop step: an automatic acknowledgement email to the buyer, the draft PDF placed in a review queue for approval, and an approve-to-send action that emails the final PDF. - The transactional emails, with correct sender authentication (SPF, DKIM, DMARC), and the start of a short follow-up sequence. DELIVERABLES - The complete working automation, end to end, from payment to delivered PDF. - The review-queue and approval mechanism. - A short handover document and a walkthrough recording so the workflow can be operated by a non-developer. - Tested against at least three sample records, with evidence. MUST-HAVE SKILLS - Advanced Make.com (or n8n) experience: multi-step scenarios, webhooks, HTTP and JSON modules, error handling, data stores. - Integrating an LLM API (Anthropic Claude, or OpenAI) into an automation over HTTP: sending prompts, parsing and validating the JSON response, handling failures and fallbacks. - Stripe: payment links or checkout, verifying webhook events, reading customer email and metadata. - HTML-to-PDF rendering with high visual fidelity (DocRaptor, PDFShift, Puppeteer or similar), including web fonts and inline SVG. - Google Sheets as a data store, with keyed lookups by email; Apps Script a plus. - Transactional email and deliverability: an ESP or SendGrid, and setting up SPF, DKIM and DMARC. - Comfortable working to a written specification and mapping tokened data into templates. NICE TO HAVE - Experience building human-in-the-loop approval or review queues. - Serverless functions (Vercel or Cloudflare) as an alternative to a no-code platform. - A background in generating personalised documents or reports from form or CRM data. TIMELINE I would like the build completed and tested within one to two weeks of starting. This is a fixed-scope project; propose milestones if you prefer. TO APPLY, PLEASE INCLUDE 1. Two or three examples of comparable automations you have built, ideally involving payments, an API and generated documents. 2. Which automation platform you would use for this and why. 3. How you would keep the rendered PDF faithful to a supplied HTML and CSS template. 4. Your fixed price and an estimated timeline. A few quick questions to answer in your proposal: - Have you integrated the Claude or OpenAI API into an automation? Describe what you built and how you handled the response. - Describe a Stripe-triggered workflow you have built end to end. - How would you approach a step where a human must approve a generated file before it is sent to the customer?