Overview

Jobs, billing, and marketing pipeline at a glance.

Open site
Active Jobs
—
—
Revenue
—
—
Mktg Drafts
—
—
Reply Signals
—
—

Recent jobs

RouteStatusSize

Transactions

ProductAmountStatusDate

Jobs

All transfer jobs. Generate marketing from any complete job.

What this does: Lists every processing run. Complete jobs can generate marketing drafts. Failed jobs need investigation — check the error column and consider re-running with a smaller or better-formatted file.
IDRouteStatusSizeLLMCreatedAction

Billing

Transaction log, revenue tracking, and cost margin.

What this does: Tracks every Stripe payment. Each paid transaction grants a single-use token that unlocks one job creation. Refunds are logged automatically via webhook. To enable billing: set STRIPE_SECRET_KEY, STRIPE_WEBHOOK_SECRET, STRIPE_PRICE_TRANSFER, and STRIPE_PRICE_BUNDLE as Worker secrets.
Total Revenue
$—
— transactions
Paid
—
Tokens issued
Pending
—
Checkout open
Margin
$—
— est. cost
Session IDProductAmountStatusTokenJobDate

Content Queue

Approval-ready posts from real AIXPORT runs. Human control. No autonomous posting.

What to do: Go to Jobs → click "Generate marketing" on a complete job → 3 drafts appear here. Review each, edit if needed, approve, then post via the intent link or LinkedIn copy flow.
Total
—
Draft
—
Approved
—
Posted
—

Queue

ChannelStatusJobSnippetEd.
Select an item to review and act.

Replies

Find relevant conversations. Generate reply drafts. Approve and post manually.

What to do: Paste any post about switching AI tools or losing AI context into the form below. Click "Add signal", then "Generate reply drafts". Review the 3 variants (Short / Longer / Soft CTA), edit, approve, copy, and manually reply on the source platform.
Signals
—
Drafted
—
Replied
—

Add signal

Signal queue

PlatformStatusScoreAuthorSnippet
Select a signal to generate reply drafts.

Help & Guidance

Everything you need to operate AIXPORT effectively. Zero support tickets.

📦How AIXPORT Works

The flow

1
Upload — user pays, selects source AI and target AI, uploads a ZIP export
2
Extract — Worker unzips, detects format (ChatGPT / Claude / Gemini / generic), normalizes all conversations
3
Process — Claude AI analyzes conversations in chunks, extracts decisions and open questions
4
Download — user gets a ZIP with project brief, memory seed, prompt pack, decision log, and transcripts

What "continuity" means

AIXPORT does not clone chats. It rebuilds portable intelligence — structured context a new AI can actually use to continue your work. The output is not a transcript dump. It's a system prompt.

✅Best Practices (for users)

File preparation

  • Export directly from the AI platform (Settings → Export data)
  • Keep uploads under 100MB for fastest processing
  • Split large projects into focused topic bundles
  • ZIP files work best — loose .json or .md also accepted

Using the output

  • Paste memory-seed.md as your first message in the target AI
  • Use prompts from prompt-pack.txt to restore specific context
  • Check open-questions.md first — these are your blockers
  • Use decision-log.md as the authoritative record
Don't expect chat cloning. Expect portable, structured intelligence you can actually use.

🔧Troubleshooting

Job failed

ZIP contained no readable files → check the export format. Re-export directly from the AI platform. Ensure the file isn't password-protected.
Large file timeout → split the project. Recommended: <100MB per upload, <500 conversations per run.
LLM processing failed → ANTHROPIC_API_KEY may be invalid or quota exceeded. Check the Anthropic dashboard.

Output quality is low

  • Check manifest.json — was llmUsed: true?
  • If template output (no LLM), set the ANTHROPIC_API_KEY secret
  • For generic/mixed files, add clear speaker labels (User: / Assistant:)

Payment issues

  • Token already used → each payment unlocks one job. Repurchase for another.
  • Billing not enabled → STRIPE_SECRET_KEY not set. Jobs are free until billing is configured.

🎯Operator Guide — Marketing

The workflow

1
A job completes → go to Jobs tab → click "Generate marketing"
2
3 drafts appear in Content Queue: X post, LinkedIn post, Case study
3
Review each → edit if needed → click Approve
4
Post to X via intent link (opens prefilled tweet) → Mark posted
5
Post to LinkedIn via Copy → open LinkedIn feed → paste → Mark posted

Content rules

Drafts are generated from real job stats — no invented numbers. If a draft sounds off, edit it or archive it. Never post something you wouldn't stand behind.

↩Operator Guide — Replies

What to look for

Posts about: switching AI tools, losing context, ChatGPT vs Claude comparisons, "my AI doesn't remember", fragmented AI workflows. Relevance score above 60 is worth a reply.

The workflow

1
Find a relevant post on X, LinkedIn, or Reddit
2
Paste it into the Replies tab → Add signal
3
Generate reply drafts → check the relevance score
4
Pick Short / Longer / Soft CTA based on the conversation
5
Edit → Approve → Copy → reply manually → Mark replied
Use Soft CTA sparingly. Lead with helping. Mention AIXPORT only when it's genuinely the answer.

💳Operator Guide — Billing

How it works

User pays via Stripe checkout → gets redirected back → frontend verifies payment → receives a one-time token → token is included with the job upload → job runs → token marked used.

Stripe setup checklist

  • Create Stripe account at stripe.com
  • Create two products: "Project Transfer" ($149) and "Continuity Bundle" ($299)
  • Copy the Price IDs (starts with price_)
  • Set up webhook endpoint: your-domain/api/webhooks/stripe
  • Set 4 Worker secrets: STRIPE_SECRET_KEY, STRIPE_WEBHOOK_SECRET, STRIPE_PRICE_TRANSFER, STRIPE_PRICE_BUNDLE

Margin model

  • Transfer: $149 revenue, ~$0.50 LLM cost → ~$148.50 margin
  • Bundle: $299 revenue, ~$1.00 LLM cost → ~$298 margin
  • Stripe fee: ~2.9% + $0.30 per transaction

⚙️Accounts & Infrastructure

Core (required)

Cloudflare Workers ✓ Cloudflare R2 ✓ Cloudflare KV ✓ Cloudflare Queues ✓ Anthropic API ✓ Stripe (set up)

Recommended next

  • Resend — transactional email (job complete, receipts)
  • PostHog — product analytics, free tier
  • Sentry — error tracking, free tier

Later (when you need auth)

  • Clerk — email + magic link auth, free tier
  • D1 — Cloudflare's SQL DB for user records
  • Upgrade from KV to D1 when you have user accounts

Prompts

Pipeline step usage counts.

What this does: Tracks how many times each processing step ran. Extraction + Normalization run on every job. Chunk Analysis + Synthesis only run when LLM processing is enabled (ANTHROPIC_API_KEY set). Marketing steps only run when you generate content from a job.
StepUses

Customers

Source → target flow distribution.

What this does: Shows which AI-to-AI migration routes are most popular. This tells you where to focus marketing and where to improve the extraction quality.
RouteJobs