Hire an AI factory sample revision reviewer

Everything it can hand back

Your task sets the actual outputs. The specialist picks the methods that cover them.

• A line-by-line pass or fail, a score out of 100 and a go or hold
• Numbered fixes to send the factory, each with a spec and a number
• Any pack line tightened, for your approval
• On a go, your margin re-checked, and at production stage the inspection brief
• A round log, so the next sample starts from what was agreed

How it does the job well

• Start from the user problem and the evidence for it, then the solution.
• Write decisions down with the reason and the options you rejected.
• Keep scope small enough to ship and learn. Name what is out of scope.
• Define how success will be measured before building.

Mistakes it avoids

• Passing a sample from photos alone when the lines are measures or weights.
• Sending the factory a vague fix like "make it better". Give a spec and a number.
• Moving a whole order to a cheaper factory after one good sample. Test a small batch.
• Looping past three rounds. After that, the problem is the factory or the pack, not the next sample.
• Treating opinions or one loud request as proven demand.
• Specs that skip edge cases, error states or dependencies.

What a good result looks like

An example, not your task: You are having a product made and samples keep coming back wrong.

• This round. One line with product, stage, round number, score out of 100 and go or hold.
• Decision and why. The failed lines that decided it, critical ones first, with the spec and what was found.
• Fix list for the factory. Numbered fixes, each with the spec, what was found and the change needed.
• Pack changes. Vague pack lines rewritten more tightly, for your approval before anything is sent.
• If go. Your margin re-checked with the real quote, a gold standard note, and the inspection brief at production stage.
• Rounds used. How many of three rounds are used at this stage, and what happens at the cap.

How the work is checked

• Every pack line is checked, or named as not checked.
• A critical fail is a hold, whatever the score.
• A score above 90 with unchecked critical lines is not a go.
• The round is in the log, with the pack version.
• States which checks passed, failed or could not run, and names missing inputs.

These checks have not been run against a model yet. Your AI uses them to check its own work.

Capabilities

• Read the log. Opens the record of earlier rounds and counts how many are used at this stage.
• Check the sample. Checks every tech pack line against the sample and scores it, with critical lines counted heavier. A line with no measure or photo is marked not checked.
• Choose the path. On a hold with a vague line, it tightens the pack first. With a clear pack, it writes the factory fix list. On a go, it re-checks the margin.
• Loop with a cap. Waits for the next sample and checks again. It stops after three rounds at one stage.
• Stop and re-score the factories. If a critical line still fails in round three, it stops and re-scores your shortlist to name the next factory to try on a small batch.
• Log and hand back. Saves the round in the log and hands back the decision, then stops for your yes.

How the role works

1. Checks your task, context and available tools.
2. Picks the methods needed and prepares a work plan.
3. Carries out supported tasks, with each method's limits and review gates.
4. Checks the real deliverables and corrects supported faults.
5. Hands back files, evidence and unresolved gaps.

Your specialist's working setup

Download the base specialist without a task form or account. README.md and SETUP.md explain how to use the folder. SOUL.md defines its scope. INTAKE.md asks missing questions inside your AI. PLAYBOOK.md, SKILLS.md, BEST_PRACTICES.md, CHECKS.md and HANDOFF.md guide the work and its checked deliverables.

Complete associated skills, supporting files and licences are included. Choose only the methods needed for the actual task. Keep context, working files and accepted feedback in your AI project.

Use this specialist through MCP or the API

Call prepare_specialist with role_id product-sample-approval-loop only. No task is required to load it. Your AI asks what job you need done, what you want back and which relevant context it can use.

For task-led hiring, call find_specialist, explain the suggested specialists and let the user accept or change the team.

Complete ZIP: Download base specialist. API: GET /api/v1/specialists/product-sample-approval-loop/pack. The host AI supplies its own model, tools and execution; a pack does not connect accounts or run a task.

Inputs and helpful context

The specialist starts from the task brief. Only active methods require their inputs.

For the primary method, when used

• Your tech pack, the current version (a document or pasted lines)
• Photos of the sample, plus measures, weights and colour readings for the lines that need numbers
• The stage: strike-off, pre-production or production sample
• Which pack lines are critical, such as size, material, safety and legal labels (strongly advised)
• Last round's log, if there was one (a pasted copy works in a chat)
• Your quote and freight, for the margin check (optional)
• A shortlist of other factories and their answers, used only if three rounds fail (optional)

Helpful context (optional)

• The factory's quote and your freight costs
• A shortlist of other factories and their answers

Tools and access

• Claude Code: It can keep the sample folder and the log for each product.
• A measuring tape and a scale: Real measures and weights are what turn a photo check into a real one.
• A leak, fit or wear test you can run at home: Critical lines such as a cap seal or a zip need a hands-on test. A photo cannot pass them.
• Any good chat AI: Any chat AI that reads images can check photos against your pack. Attach every angle.
• A kitchen scale and a tape measure: Give real weights and lengths. The AI cannot measure what you did not.
• Any good chat AI: Any chat AI that reads images can draft the pack from your photos.
• Figma, Canva or Notion: Hold the drawings and specs in one shared place with a version history.
• A freelance designer: A freelance product designer can draw it up. The source puts the price near $200, which you should re-check for your product.
• Any good chat AI: Any chat AI can add the layers. Paste your numbers and check one sum by hand.
• Shopify: Holds cost per item for every product, so profit per order and per day use the landed cost.
• A customs broker or your country's tariff lookup: Gives the duty rate for your product code and route. The skill needs it from outside.
• A third-party inspection company: Sends someone to open random units, measure and photograph them. Ask which sampling standard they use.
• Any good chat AI: Any chat AI can turn your order and pack into the brief.
• Any good chat AI: Any chat AI can score the answers and draft the messages. Paste what each factory said.
• Alibaba: Supports filters for verified suppliers and trade protection. Treat it as research, not the final choice.
• A sourcing agent or inspection platform: Can visit a factory for you when you cannot go. Check the agent first.

Functions and deliverables

• Factory sample revision reviewer: Compare each factory sample with your product spec. Get go-or-hold reasons, clear fixes and an inspection brief when supported.
  • A line-by-line pass or fail, a score out of 100 and a go or hold
  • Numbered fixes to send the factory, each with a spec and a number
  • Any pack line tightened, for your approval
  • On a go, your margin re-checked, and at production stage the inspection brief
  • A round log, so the next sample starts from what was agreed
• Product sample approval reviewer: Check a product sample against your tech pack line by line. Get pass or fail on each line, a score out of 100 and a go or hold on bulk.
  • A pass or fail for every line of the tech pack
  • A score out of 100 and how it was worked out
  • What to fix before the next stage, and a go or hold on bulk
  • Which sample to keep as the gold standard, a test batch plan and a 90 day return watch
• Factory product brief writer: Turn your product photos and notes into a tech pack a factory can build from, with every vague word rewritten as a measurement.
  • A tech pack: drawing or photo callouts, materials with weights and finishes, hardware and trims, measurements for every size, color codes, labels, packaging and care text
  • Name, version and date on every page
  • A list of every word that left a decision open, rewritten as a measurement
  • One quote request that asks every factory for the same answers
  • A list of measures and weights your quality checks can use
• Landed cost calculator: Work out your landed cost per unit and margin from a factory quote. Add freight, duties and fees, and get a go or no-go score out of 100.
  • Cost per unit by layer, from ex-factory to courier
  • Cost per sellable unit after defects and spare units
  • Your margin at your price, and the price you need for your target margin
  • Scenarios: air or sea, small or big order, each at its own price break
  • A go or no-go, a score out of 100, and how it was worked out
• Pre-shipment inspection brief writer: Share the order, approved sample and product specs. Get an inspection brief, proposed decision rules and a payment checklist.
  • An inspection brief: sample size, measurements, packaging, labels, carton weights and photos required
  • A pass, rework or reject rule, written before the report arrives
  • A payment release plan where the balance is paid only on a pass
  • The spare-units rule, if inspection is not booked
• Supplier shortlist vetter: Vet a factory shortlist before you pay a deposit. Give it your spec and each factory's answers. You get scores out of 100, red flags and a pick.
  • A shortlist of factories to contact and the message to send
  • Video-call questions and a red-flag list
  • A score out of 100 for each factory, and how it was worked out
  • A dependency map with a risk rating, a backup supplier plan and terms to ask for
• Product context file writer: Share product facts, audience and positioning. Get a reviewed context file that other marketing skills can reference.
  • A product marketing context file
  • A summary of what the file already holds
  • A change log of updates
  • Gaps you still need to fill
  Source: Corey Haines.

Tasks you can use it for

• You are having a product made and samples keep coming back wrong.
• A new sample arrived and you need to decide whether it can move to the next stage.
• You want a written record of what each sample got wrong and what was asked.
• You want a clear point to stop, so a bad factory does not eat months.

Limits to check before starting

• You have no tech pack yet. Write one first with the Factory product brief writer.
• You are checking finished goods before they ship. Use the Pre-shipment inspection brief writer, and have an inspector check the real goods.
• You are still choosing between factories and no sample exists. Use the Supplier shortlist vetter.
• You want one sample checked once, with no round log. Use the Product sample approval reviewer.

Best way to use this role

Give it one clear outcome, the actual files and a definition of done. Keep the work in one AI project so it can reuse checked context and earlier files.

• Add the agent to your AI. In Claude Code, add it as a skill or save it in .claude/agents. In Claude, ChatGPT or Gemini, paste it at the start of a chat and paste last round's log if you have one.
• Attach the tech pack and the first sample's photos and measures. Say the stage. Use a tape measure and a scale, because a photo cannot show a weight or a size.
• Read the decision. If it is a hold, check the fix list and send it to the factory yourself.
• When the next sample comes, run it again with the log. It checks only what is new.
• On a go at production stage, take the inspection brief to your inspector before the goods ship.

Read it as data

Read the complete role setup as JSON. Connect your AI to the AI Blocks MCP library and read aiblocks://roles/product-sample-approval-loop for its role instructions and pinned sources.

This is a prepared AI role. No task has run. Your AI carries out authorized work using its available tools and reports the actual deliverables and checks.