Hands-on labs, the full verdict archive, a watchlist, and worksheets, all based on linked public documentation. Checkout is closed, so you can't buy it yet.
See plans: from $19/mo What's insideAlready a member? Sign in to the library.
7 reader labs and 11 source-based guides. You run the steps. We did not run the products for you.
Worksheets for a tool pick, a subscription bill, a weekly catch-up, a local model, a price, a model card, and a release. Plus a cost calculator.
The full verdict archive, with category filters and a log of the first snapshots, and a watchlist with notes and a next-check date in this browser.
Pro also includes courses: AI Literacy, Claude Certified Architect Decoded, and Build Your First AI Agent. Three other tracks are named and not written. The hub lists all six. The lessons use the same sign-in as the library. Checkout is still closed.
Ready means the guide is written from public sources. It doesn't mean we tested or timed the product ourselves. If a lab asks you to run something, you need your own permission to run it.
| Title | Type | Status | What it does |
|---|---|---|---|
| Pick a coding plan from your actual requirements | Workflow playbook | Ready | An offline scorecard for business-owned plans, with a documented decision and exit path. |
| Cost per passing output: a reader evaluation worksheet | Workflow playbook | Ready | Define acceptance, compare authorized outputs and calculate cost per pass without invented results. |
| Specify an internal tool before an agent builds it | Workflow playbook | Ready | A request-intake specification, permission matrix and release checklist based on Cursor cloud-agent docs. |
| Build an invoice extraction specification | Workflow playbook | Ready | A complete schema, synthetic fixture and review checklist, without an unverified model recommendation. |
| Design a bounded, non-interactive code review | Workflow playbook | Ready | A least-privilege CLI recipe and failure policy from official Claude Code documentation. |
| Read a model card before you trust the adjective | Workflow playbook | Ready | A reader worksheet for separating a vendor's model-card claims from what the card does not say. |
| Take a price snapshot you can compare later | Workflow playbook | Ready | A dated worksheet for copying a public price page, so a later change is a second row and not a guess. |
| Draw the permission line before the agent runs | Workflow playbook | Ready | A worksheet for listing what a public doc says an agent may touch, and what it does not say. |
| Pick the tool for the job in front of you | Workflow playbook | Ready | A worksheet for choosing one AI tool for one named job, using public pages and constraints you write down first. |
| Cut the duplicate bill without sharing the login | Workflow playbook | Ready | A worksheet for separating chat subscriptions, API invoices, and usage caps, using the numbers printed on official plan pages. |
| A 30-minute catch-up that actually ends | Workflow playbook | Ready | A weekly routine for reading five briefs, checking one source, and updating one watchlist note before the timer runs out. |
| Run one open model on the machine you already have | Workflow playbook | Ready | A step-by-step for installing Ollama, pulling one published local model, and recording the license and the disk size from the official pages. |
| Caching math without imaginary savings | Cost comparison | Ready | A read/write/uncached-token worksheet with transparent illustrative arithmetic. |
| Five-seat coding budget worksheet | Cost comparison | Ready | Compare seat, usage and migration costs without stale plan names or unsupported list prices. |
| API cost comparison: bring your own verified rates | Cost comparison | Ready | A transparent workload table and interactive calculator; examples are hypothetical, not current provider quotes. |
| Audit a deprecation notice before changing your stack | Switch guide | Ready | A source-verification and replacement checklist without an unconfirmed deadline. |
| Switch models without a blind cut-over | Switch guide | Ready | A model-neutral migration runbook with routing, acceptance and rollback decisions. |
| Source evidence and decision record | Template | Ready | A copyable record that separates documentation, vendor claims, unknowns and reader-run results. |
Hands-on labs, the full verdict archive, a watchlist, and worksheets, for people who want to put AI tools to work, not just read about them.
What is on the site today. Pro is a planned price. Team has no confirmed price.
| Feature | Free | Pro | Team |
|---|---|---|---|
| Price | $0 | $19/month or $190/year | Coming soon |
| Coverage | |||
| Tuesday and Friday briefing | Yes | Yes | Yes |
| Tool profiles | 73 | 73 | 73 |
| Company profiles | 44 | 44 | 44 |
| Published issues | 3 | 3 | 3 |
| News briefs | 56 | 56 | 56 |
| Public guides | 8 | 8 | 8 |
| Learn explainers | 10 | 10 | 10 |
| Pro library | |||
| Pro reader labs | No | 7 in the preview | Coming soon |
| Other Pro library pieces | No | 11 in the preview | Coming soon |
| Worksheets and checklists | No | 7 files in the preview | Coming soon |
| Tools | |||
| Watchlist saved in this browser | No | Yes, on the Pro preview | Coming soon |
| Cost calculator | No | Yes, on the Pro preview | Coming soon |
| Team | |||
| Team seat list, stack board, policy template | No | No | Local preview. Not sold. |
$0
Tuesday and Friday briefing, plus the public directory, guides, explainers, and news briefs. No Pro library or tools.
$19/month or $190/year
Everything in Free, plus Pro library labs, worksheets, a watchlist, and the cost calculator. Planned at $19/month or $190/year.
Coming soon
Coming soon. Same library for a team, plus a seat list, stack board, and policy template. Local preview. Not sold.
ShipReady Lab is a separate preview with 8 modules. It is not included above, and it is not for sale. The public directory, issues, guides, and explainers stay free.
Pro is $19/month or $190/year when checkout opens. This form joins the existing briefing list. It does not start a paid plan.
Team is coming soon. This page does not list a price. This form joins the same briefing list.
No. Checkout stays off until we have verified the launch checklist, access controls, payments, and refunds. The prices on this page are a preview, not a live offer.
18 source-based guides, including 7 reader labs, a cost calculator, switch checklists, the full verdict archive with category filters, a watchlist with notes stored in this browser, and worksheets for a tool pick, a subscription bill, a weekly catch-up, a local model, a price, a model card, or a release. Illustrative numbers are labeled. We do not invent model performance results.
No. There is no fixed weekly paid-content quota and no instant-alert service. Updates are dated in the library. The free briefing is still the news.
Our recommendation for a specific kind of reader, based on public sources. It does not mean we tested the product ourselves. Watch means an important question is still open. Ignore means it is a bad fit for that reader.
If Pro doesn’t save you at least one hour in your first month, email us within 30 days of your first payment and we’ll refund it in full. No forms, no argument.
No. Sponsors buy a clearly labeled ad. Payment does not buy a verdict, a ranking, or coverage.
Up to 10 named seats, including one admin, a shared stack board, an AI usage policy template, a rollout checklist, and a monthly stack-review worksheet you run yourselves. The template is not legal advice. It does not include a consultant or a custom monthly analysis.
Before sales open, we have to check receipts, cancellation, when access ends, and refunds in Stripe test mode. There is no working billing portal yet.
A self-paced project where you build a request-board app from public documentation: eight modules, starter files, worksheets for the database, auth, deploy, monitoring, and release steps, plus a launch and rollback checklist. Planned price: $249, or $199 for active Pro members. Hosting and other outside services are not included. It is not for sale yet.
Email hello@loadedtokens.com. A person reads it.
The source comes with the story.
Your inbox can handle one more.
