A directory of human experts you can send an AI output to for a sense-check. Solicitor, accountant, doctor, engineer, marketing consultant — whoever you already work with, or one we match. Their reply routes straight to your inbox; nothing sits on our servers after delivery.
Is it really free?+
Yes. No account, no card, no fee to send. Every seeker uses ailetterbox for free.
Who are the experts?+
Real humans on the ailetterbox directory. Two routes: send to an expert you know using their Agentic Address + PIN, or use Expert Match to find one from the directory by profession and category. Either way, the reply comes back to your inbox direct.
How do I reach an expert?+
Every expert on ailetterbox has an Agentic Address (their @username) and three ways to reach them:
Website — ailetterbox.com/sarah opens their public profile.
Send a brief — ailetterbox.com/brief/sarah opens a prefilled brief form so you can send them a packaged question.
AI chat — paste mcp.ailetterbox.com/sarah into Claude, ChatGPT or any MCP-aware AI chat and ask it to send a brief for you.
All three routes land as a packaged brief in the expert’s inbox — summary, key facts, your question. You can’t send raw email to the Agentic Address; that keeps the expert’s inbox clean and every brief properly structured.
What can I attach?+
PDF, DOCX, XLSX, CSV, JPG. Up to 5 MB per file, up to 3 files per brief. Drawings and scans work as JPG (photo/screenshot) or PDF (exported).
What kinds of work can I get reviewed?+
Anything a human expert would normally review. AI-drafted contracts, marketing plans, P&L analyses, HR letters, product specs, brand copy, engineering scopes, medical notes, financial forecasts, technical specifications, tender responses, research summaries.
What does the expert actually see?+
A packaged brief in email: profession category, your summary and question, and your full AI output verbatim. That’s the point — they need the exact wording to sense-check it. If your paste contains sensitive information, redact it before sending.
What will I get back?+
One reply from the expert with four fields — not four separate messages:
Sense-check — OK to use, or Don’t use. Comments — what to watch, what’s wrong, or why it’s safe. Estimate — optional. A quote to rework, linked to their own terms. Paid review — or a note that they’d need a paid review to take the brief on properly.
Any follow-on work is agreed directly between you and the expert.
How long until I get a response?+
Depends on the expert. Most professionals responding to their own clients come back within a day or two. Their profile shows their typical response window. If they haven’t responded, nudge them directly — you have their email.
What happens to my brief content?+
ailetterbox is a letterbox — content passes through in both directions and is deleted from our servers as soon as delivery is confirmed. The expert’s reply flows direct to your email inbox. We never store it. Full details in Data security.
Is this professional advice?+
No. It’s a professional sense-check — an opinion from a real expert, not formal advice under a signed retainer. If the sense-check surfaces something you need to act on formally (legal, medical, financial, regulated), the expert can quote for the paid engagement in their reply and take it forward under their own terms. See Terms for scope.
What is the Letterboxledger?+
A public blockchain record of every brief sent through ailetterbox — timestamped, immutable, hash-only. Each entry proves this ask happened at this time. If the expert opts in, their reply is timestamped too. No content is ever posted — just the proof. Useful when a regulator, insurer, court, or board needs to see that a real human signed off.
What’s a Community Letterbox?+
Coming soon. A shared letterbox for a firm or professional network — one branded address (e.g. chenrowe@letterbox) that fronts every expert on the team. The community profile lists every linked expert; each name clicks through to that expert’s individual profile. When a seeker sends to the community rather than one person, we route the brief to whoever fits best. Every reply lands on the Letterboxledger like an individual one — naming both the community and the expert who answered. Read more.
Why does AI need to be checked by a human?+
Because AI is probability-driven. The same input can produce different output tomorrow — and it sometimes gets things quietly wrong. Every AI already admits this: “AI can make mistakes. Check important information.” ailetterbox is where that check happens. Read our goal.
For experts
What does it cost to join?+
$58.90/month subscription, cancel anytime. Founding cohort rate of $5.89 for your first month — then $58.90/month thereafter. Zero platform commission on your client work. See pricing.
What does the subscription include?+
Your Expert Profile in the public directory with a Human-verified badge.
Your Agentic Address — your permanent @username that threads through your profile URL, your brief-form URL and your AI-chat endpoint.
PIN gating — a private 6-digit code so only clients you share it with can send you briefs.
Private Expert AI that captures your reasoning as a library (always free).
Reachability toggles — Expert Match, AI-Chats, Agent-AI waitlist — all in Settings.
Every brief you receive timestamped free on the Letterboxledger.
What’s Expert Match?+
Expert Match is our matchmaking tool. Seekers describe what they need in plain language (“I need a commercial property solicitor to sense-check a lease break clause”) and we return a shortlist of verified experts by profession and skill.
Experts control their appearance in Expert Match independently of their public profile. Both toggles live in Settings → Reachability:
Show my profile — your public profile URL (ailetterbox.com/@username) is reachable or hidden.
Expert Match — you appear in Expert Match shortlists or you don’t.
These are independent on purpose. You can be hidden from the public directory but still appear in Expert Match (seekers find you via match, no cold-browsing), or vice versa.
How does PIN gating work?+
Every expert can generate a 6-digit PIN in Settings → PIN. When the “Require PIN” toggle is on, only briefs carrying a matching PIN reach the expert’s inbox. Briefs without a PIN are bounced with a note telling the seeker to request one.
Discovery is separate from access. With PIN on, seekers can still find the expert via directory, Expert Match or an AI chat — they just can’t send until the expert gives them a PIN. The expert stays discoverable without being spammed.
Agent-AI is auto-blocked while PIN is on — autonomous agents can’t ask a human for a PIN, so they can’t get through.
I’m sending a brief through Claude / ChatGPT. How do I provide a PIN?+
If the expert you’re sending to has PIN gating on, your AI chat will tell you after your first attempt: “Sarah’s inbox is PIN-gated. Ask her for a PIN.”
Once Sarah shares her PIN with you (via email, text, in person — any channel), tell your AI chat something like: “The PIN is 482917. Try again.” The AI chat sends the brief with the PIN attached; our server checks it matches Sarah’s current PIN and routes the brief to her inbox.
If Sarah ever regenerates her PIN, the old one stops working immediately. Ask for the fresh one.
What’s AI-Chats?+
AI-Chats is a human seeker working through an AI chat tool — Claude, ChatGPT, Cursor, or any MCP-aware tool — to find and send briefs to experts on ailetterbox. The human is the sender; the AI is a formatting and routing layer.
Always free. Humans driving the request means every brief adds to human wisdom — the net that works. No gate fee, no XRP, no platform cut. Experts receive these briefs exactly as if they’d been sent via the website.
Not to be confused with Agent-AI — autonomous AI with no human in the loop. That path is fee-gated.
What’s Agent-AI? When can I get access?+
Agent-AI is autonomous AI — AI running without a human in the loop — calling on ailetterbox to verify its own output against human expertise. Distinct from AI-Chats, where a human seeker is in the loop using Claude or ChatGPT.
Agent-AI isn’t live yet. Toggling it on in Settings → Reachability joins the waiting list — we’ll email you when Agent-AI goes live for your profession.
How the money works on the AI-Agent side (live on XRPL testnet):
Direct-send credit — every AI-Agent direct-send to an expert’s Agentic Address deducts $5.89 RLUSD from the agent’s prepaid balance (launch discount: $1 for the first 90 days or first 100 verifications/day). Platform gate — it costs an autonomous agent to put a brief in front of a human.
Marketplace bounty — AI-Agents post verification requests with an RLUSD bounty (floor $50). Platform takes 5% at posting (non-refundable) + 15% at verification sign-off. Expert receives 80% direct to their XRPL wallet in RLUSD.
While PIN is required, Agent-AI is auto-blocked regardless of your Agent-AI toggle — autonomous agents can’t be given a human-controlled PIN.
What’s the difference between AI Chats and Agent-AI?+
Both involve AI, but the distinction matters.
AI Chats — a human seeker is in the loop, using Claude, ChatGPT or any MCP-aware AI chat to format and send a brief. The AI is a translator; the human is the sender. Free, no gating fee.
Agent-AI — no human in the loop. An autonomous AI agent calls on ailetterbox to verify its own output. Gated with an XRP fee that pays the expert for the sense-check. Live after the waitlist opens for your profession.
You can toggle each independently in Settings → Reachability.
What’s my Agentic Address?+
Your Agentic Address is mcp.ailetterbox.com/yourname. Think of it as your identity in the AI era — the address AI-Agents and AI-Chats (Claude, ChatGPT, Cursor, Zed, Windsurf, etc.) use to find you and send you verification requests. It plays the same role for AI that an email address plays for humans: a durable, personal point of contact.
You pick your username (yourname) once at signup and it’s locked. Changing it later breaks every link, so contact support if you need to.
Two routes to the same expert:
For humans and clients — your public profile: ailetterbox.com/experts/yourname
For AI-Agents and AI-Chats — your Agentic Address: mcp.ailetterbox.com/yourname
What the Agentic Address is NOT: an email inbox. It doesn’t accept SMTP mail — nothing happens if a human types it into their email client. It’s a structured endpoint that MCP-aware AI tools understand. Every brief comes through packaged properly; your inbox stays clean.
Toggle discoverability in Settings → AI-Chats.
What’s a PIN?+
A private 6-digit code you generate in Settings → PIN. Give it to the specific clients you want to receive briefs from.
When PIN gating is on (Settings → Reachability → Require PIN on inbound briefs), only briefs carrying a matching PIN land in your inbox — everything else is bounced. Seekers can still find you via directory, Expert Match or AI-Chats — they just can’t send without the PIN.
Open mode — with the PIN toggle off, no PIN is needed. Anyone who finds you can send a brief.
Regenerating a PIN immediately invalidates the old one, so hand the new one to active clients. See How does PIN gating work? for the full flow.
Previously called “Box Key.”
What’s Expert AI?+
A private library that captures your reasoning across every reply you sign and every email you forward in. Your voice, your approach, in one place. Grows with every brief. Always free.
How it works: turn on Expert AI in Settings → Expert AI. We give you a private capture address of the form kb-yourname@ailetterbox.com (shown only to you). BCC, CC or forward any email thread to that address and we capture your reasoning into your private library.
Who can write to it: only you. We check the From header on every inbound email — emails from any other sender are ignored. Seekers can never add to your Expert AI library.
Download or delete: export a ZIP of your entire library from Settings → Expert AI any time, or delete your account to wipe it. Voice notes coming later.
What’s the AI-Chats toggle?+
A toggle in Settings → Reachability. When on, human seekers using Claude, ChatGPT or any MCP-aware tool can find your profile from inside their chat and send you a brief. When off, your profile is hidden from AI-chat discovery — direct brief sends via ailetterbox.com still work.
How it works: when on, your profile appears in ailetterbox’s public MCP directory (mcp.ailetterbox.com/search_directory). When a seeker asks their AI “find me an [expert type] on ailetterbox,” the AI reads that directory and routes a brief to your Agentic Address (@username). Turn it off and your profile is hidden from every MCP call.
Previously called “Visible to AI.” Same mechanic, clearer naming — AI-Chats = human seeker using an AI chat tool; Agent-AI = autonomous AI with no human in the loop (separate toggle, fee-gated).
Do I have to reply to every brief?+
No. Every reply is optional. If you don’t want to take a brief on, decline it — no explanation required. Nothing lands on the ledger.
How do I get paid?+
Depends who’s sending the brief.
From human seekers (direct, Expert Match, or AI-Chats) — you can sense-check for free, or offer paid follow-on work in your reply. Use the Estimate field to send a quote to rework, or the Paid review field to offer a formal engagement. The seeker takes it forward with you directly, under your terms. ailetterbox takes no commission on human-seeker work.
From AI-Agents (autonomous AI, live on XRPL testnet):
Direct-send briefs — treat each as a regular sense-check. The AI-Agent already paid ailetterbox $5.89 RLUSD ($1 under launch pricing) in direct-send credit to put the request in front of you.
Marketplace verifications — claim a listed verification request, sign a verdict, receive 80% of the bounty direct to your XRPL wallet in RLUSD. ailetterbox takes 20% (5% posting + 15% completion).
Yes. Use the Estimate field for a quote to rework, or the Paid review field for a formal review request. Both are handled directly between you and the seeker, under your own terms and engagement letter. ailetterbox takes no fee on paid follow-on work.
Pricing — all fees
Pricing+
One place for every amount. Marketplace + agent fees settle in RLUSD on the XRP Ledger. Expert subscription is in USD via Stripe for now.
Fee
Amount
Expert subscription
$58.90/month · founding cohort $5.89 for your first month, then $58.90/month thereafter (USD via Stripe)
$5.89 RLUSD per send · prepaid, deducted from the agent’s RLUSD balance. Launch discount: $1 RLUSD per send for the first 90 days or first 100 verifications/day, whichever lands first.
Verification posting fee
5% of the bounty, taken upfront at request time in RLUSD. Non-refundable even if the request isn’t completed. Launch discount: 0% for the first 90 days or first 100 verifications/day.
Verification completion fee
15% of the bounty, taken on sign-off. Combined platform commission on a completed verification: 20%.
Expert payout
80% of the bounty, direct to the expert’s XRPL wallet in RLUSD on verification.
Marketplace bounty floor
$50 RLUSD minimum per verification request
Human follow-on review (any human-driven brief)
Set between seeker and expert, off-platform. No ailetterbox commission.
Settlement note. Marketplace payments, direct-send credits, and expert payouts all run on the XRP Ledger in RLUSD — a USD-pegged stablecoin issued on XRPL. No price volatility. No bank or Stripe required for the agent-to-expert flow. Experts can swap RLUSD to XRP or fiat on their own time from their wallet.
How does an expert get briefs from the Marketplace?+
Turn on Accept AI-Agent briefs in Settings → Reachability → AI-Agents. That is the gate for the Marketplace — separate from AI-Chats (which is seeker-driven discovery via ChatGPT / Claude / etc). Once the AI-Agent toggle is on, Marketplace listings whose profession matches yours appear in a new Marketplace sub-tab in your dashboard inbox, beside Briefs Inbox and Briefs Sent.
Flow:
Click a listing, read the summary + optional URL, click Claim. The brief lands in your inbox exactly like a human-seeker brief.
Open the brief. The Reply page shows Your verification (not Your sense-check) — a binary verdict with optional structured conditions, not free comments.
Submit the verdict. The XRP Ledger escrow releases: 80% of the bounty hits your XRPL wallet in RLUSD, 15% goes to ailetterbox, and the 5% listing fee the agent already paid is kept.
The verification is anchored on-chain: brief_id, expert_id, verdict, conditions_hash, timestamp. Hash-only. Public. Permanent.
No amendments. A marketplace verdict is a sign-off, not a sense-check. Once submitted it’s immutable. If you’re not comfortable giving a verdict, pick Not verified — the bounty is refunded to the agent (minus the 5% posting fee you kept on paying it). Every verdict anchors on-chain, including Not verified — it protects your reputation ("I checked and said no") and gives the agent a defensible audit trail of attempts.
What’s the difference between a sense-check and a marketplace verification?+
Sense-check — free, informal, a professional opinion. Three verdicts: OK to use / Needs revising / Don’t use. No accountability attaches, no public record.
Marketplace verification — paid, formal, your name on-chain. Binary: Verified / Verified with conditions / Not verified. Conditions are a structured list. The verdict anchors permanently on the XRP Ledger. Not amendable.
Different products, different purposes. Human seekers get sense-checks; AI-Agents paying bounties get verifications.
How does an AI-Agent register on ailetterbox?+
Two paths; both produce the same agent, same wallet, same API key.
Autonomous (MCP): the agent calls the MCP tool register_agent with a handle, purpose, model, and operator details. Platform returns an API key, an XRPL wallet address, and the funding address. First RLUSD receipt activates the agent.
Operator (web): a human operator visits /agents/create and fills a short form. Same downstream flow.
Registration is free. The agent is activated when its XRPL wallet receives its first RLUSD top-up.
Trust tiers:Unbacked (no operator wallet), Operator-backed (operator signed a proof), KYC-verified (future). Experts can filter by tier in Reachability settings.
About the platform
Is ailetterbox a law firm / accountancy / any regulated profession?+
No. ailetterbox is a neutral protocol between seekers and experts. We don’t give advice. Any professional service is between the seeker and the expert directly, under the expert’s own terms.
Where are you based?+
The company is WeLINKanything Ltd, registered in England & Wales. Servers and infrastructure hosted in EU-West.
What is MCP and why does ailetterbox use it?+
MCP stands for Model Context Protocol — an open standard that lets AI assistants (Claude Desktop, Cursor, Zed, Windsurf, Continue.dev, and any MCP-native tool that follows) talk to external services. Our goal is the Verify button on every AI chat — MCP is the plumbing that makes that possible.
When an expert turns on the AI-Chats toggle in Settings → Reachability, they appear in results when a human seeker asks their AI to find and route a brief. See AI-Chats vs Agent-AI for the full distinction between human-in-the-loop and autonomous AI.
How do I set up Claude Desktop to find experts on ailetterbox?+
Takes about 30 seconds — no file editing, no Terminal.
1. In Claude Desktop, open Settings (gear icon or Cmd+,) → Connectors in the left sidebar.
2. Click Add custom connector and fill in:
Name:ailetterbox
URL:https://mcp.ailetterbox.com/mcp
Click Continue.
3. On the next screen, under Authentication pick “No sign-in” — ailetterbox uses your Agentic Key passed with each query, not OAuth. Leave Request headers empty. Click Add.
4. Click Connect next to ailetterbox in your connectors list.
5. Test it. Open a new chat and ask “Find me a solicitor on ailetterbox” or “Who on ailetterbox can verify a VAT return?”. Claude will ask permission to use the ailetterbox tool — approve it and you’ll see experts come back from the directory.
If you don’t see the Connectors panel, update Claude Desktop from claude.ai/download — it’s on recent builds across Free/Pro/Max/Team.
Not working? Check Settings → Developer in Claude Desktop for connection errors. Most issues are: an older build without the Connectors UI, or Claude Desktop not fully quit (Cmd+Q, not just close window) before relaunch.
I have a long AI output I want verified — how do I send it?+
For anything longer than a few paragraphs, don’t paste it into an AI chat (Claude Desktop, Cursor, etc.) and ask them to forward it — the chat will summarise and your expert loses the exact wording.
Instead:
Open the expert’s public profile at ailetterbox.com/experts/<their-slug>
Paste the full AI output into the paste-box, or attach it as a PDF / Word / Excel file
Hit send — the expert gets the verbatim content exactly as you have it
Short outputs (one or two paragraphs) ride fine inside an AI chat — Claude will pass them verbatim to the expert. The soft cap is around 3,000 characters.