⏰ Deadline: every task fully signed off by Fri, Nov 13 · 11:59 PM ET

Expert Brief · Enterprise Workspace Benchmark

SME ProjectBrief

Turn a set of real-world business files into an executive-grade Golden Output, and sharpen the rubric that grades AI against it.

Domain Expert Verification 5–10 hours per task

~10 min read

Project Overview

Project ID

Enterprise Workspace Benchmark

Domain

Consulting · Legal · Controllership

Entity

Human Edge

Role

Senior Domain Expert

Phase

Production · sign-off by Fri, Nov 13 · 11:59 PM ET

Why your work matters. The client is training AI models to produce senior-level business work from incomplete, ambiguous files. Your Golden Output shows the model what excellent looks like, and your rubric is the ruler that grades every AI attempt.

⏰ Every task must be fully signed off by Fri, Nov 13 · 11:59 PM ET.

The Basics

Key Concepts in 5 Minutes

No AI background needed. This page explains the 3 ideas the whole project is built on: the User Prompt, the Golden Output and the Rubric. Everything else in this brief builds on them.

1. The big picture: think of an exam

AI models improve the way students do: they get a question, write an answer and are graded. The client needs senior experts like you to write the answer key and the grading sheet for professional-grade work.

❓

In an exam: the question

User Prompt

The request a business user gives the AI, plus the files it should use.

Already written · don't edit it

⭐

In an exam: the model answer

Golden Output

The best answer a senior expert would deliver. It shows what excellent looks like.

You write it

✅

In an exam: the grading sheet

Rubric

A list of yes/no rules used to grade any answer, including the AI's.

You check and improve it

2. How the client uses your work

Your 3 pieces become a test the client runs on its AI models again and again.

1User Prompt + filesThe AI receives the request

→

2AI writes an answerA deck, a memo, a sheet…

→

3Rubric grades itEach rule: ✓ pass or ✕ fail

→

4Score shows the gapsThe model learns where it falls short

⭐

Where your Golden Output fits: it is the reference the AI is measured against. It must pass 100% of the rubric rules, which proves the test is fair and achievable.

💡

Why it has to be excellent

If your Golden Output has an error, the AI is taught that the error is right. If a rubric rule is vague or wrong, good AI answers fail and bad ones pass. Your expert precision is what makes the test trustworthy.

3. The User Prompt

A realistic request, written the way a busy manager would ask for it: often short, sometimes vague, with files attached. You'll find it on the first tab of your rubric spreadsheet.

Consulting

"I've got a call with the CEO tomorrow. Put together a short deck comparing the two expansion sites in the P&L model."

Legal

"Can you tell me what changed between the 2023 and 2024 supplier agreements before I sign the new one?"

Controllership

"Why is March payroll so much higher than February? I need a short note for the CFO."

Notice what's missing: no format details, no list of what matters, and files that may contain noise or errors. That ambiguity is intentional. A senior expert knows what a great answer needs anyway.

4. The Golden Output

The answer a top expert in your field would hand to a demanding executive. It must be:

AccurateEvery figure and clause traceable to the files CompleteAnswers everything the User Prompt asks Executive-gradeClear structure, bottom line first, client-ready GroundedNothing invented that the files don't support

✕ Weak answer

"Payroll went up in March, mostly because of overtime and some new people."

Vague, no figures, and the wrong main driver.

⭐ Golden Output

"March payroll rose $18,600 (+6.2%) vs. February. Two new Sales hires drove most of it ($11,200), followed by annual merit increases ($5,900) and overtime ($1,500)."

Exact, explains the cause, ranks the drivers, and the drivers add up to the total.

5. The Rubric

A checklist of rules. Each rule checks one thing and is answered pass or fail. Together the rules describe everything a great answer must get right.

Criterion"The note states the total increase as $18,600."What is checked

Accepted values$18,600 · $18.6K · 18,600Every valid way to say it

Evaluation methodProgrammatic CheckHow it's graded (see glossary)

SeverityCritical RequirementHow much it matters

What makes a good rule? Compare these pairs:

✕ Weak rule✓ Strong ruleWhy

"The memo is good.""The memo opens with a 1-paragraph Executive Summary."Objective: two graders would agree

"The memo mentions cell F118.""The memo says the total omits Insurance ($3,000)."Business language a real expert would use

"States the total and the drivers and recommends next steps."Split into 3 rules, one per itemOne thing per rule

"States the total as $48,000." (only)"Accept $48,000, or $46,500 with the error flagged."Fair to every valid expert answer

6. Try it: grade an AI answer

Pick your field, read the AI's answer, decide whether each rule passes, then reveal the grading.

User Prompt

"Short deck for the CEO comparing the two expansion sites in the P&L model."

AI's answer (slide 2)

Site comparison
• Site A: startup $120K, payback 22 months
• Site B: startup $160K, payback 14 months
• Both sites look viable

Rubric rules

1. States Site B payback as 14 months✓ Pass

2. Slide title states a conclusion (action title)✕ Fail: "Site comparison" is a topic, not a conclusion

3. Site figures shown in a native Slides table✕ Fail: bullets, not a table

⭐ Golden Output title: "Site B pays back 8 months faster despite $40K higher startup cost", followed by a native comparison table.

User Prompt

"What changed between the 2023 and 2024 supplier agreements before I sign the new one?"

AI's answer (excerpt)

The 2024 agreement reduces the liability cap from 12 months of fees to 6 months of fees. Payment terms remain Net 30. The rest of the agreement is largely unchanged.

Rubric rules

1. States the liability cap fell from 12 to 6 months of fees✓ Pass

2. Flags the new auto-renewal clause (Section 9.2)✕ Fail: missed entirely ("largely unchanged")

3. Opens with an Executive Summary of the risk shift✕ Fail: no summary

⭐ Golden Output opens with: "The 2024 terms shift risk to us: liability exposure is halved and the contract now auto-renews unless cancelled 90 days ahead (Section 9.2)." It follows with a clause-by-clause table.

User Prompt

"Why is March payroll so much higher than February? Short note for the CFO."

AI's answer (excerpt)

March payroll increased by $18,600, mainly due to overtime. New hires and merit increases also contributed.

Rubric rules

1. States the total increase as $18,600 (accept $18.6K)✓ Pass

2. Identifies new hires as the largest driver✕ Fail: says overtime ($1,500 is the smallest)

3. Breaks the increase into drivers that add up to the total✕ Fail: no amounts per driver

⭐ Golden Output: "+$18,600 (+6.2%): new hires $11,200 · merit $5,900 · overtime $1,500." The drivers add up to the total.

This is exactly why the rubric matters: the AI got the headline number right but still failed 2 of 3 rules a senior expert would care about.

7. Your role, in one picture

Verify

Understand the [Source] files and find the errors hidden in them

→

Author

Upgrade the [Current Reference] draft into the Golden Output

→

Calibrate

Make every rubric rule correct, clear and fair, and add missing ones

8. Glossary

Terms you'll see in this brief and in your rubric spreadsheet.

[Source] filesThe documents, sheets and emails behind the task. Use them as evidence; never edit them.

[Current Reference] draftA rough first version of the deliverable. Your starting point, not the answer.

Golden OutputYour finished, expert-grade deliverable, saved as a new Google file.

Rubric ruleOne pass/fail check on an answer. The rubric is the full list.

DiscrepancyAn error hidden in the files, e.g., a total whose formula skips a row, or a clause that contradicts another document.

Fix-or-FlagA rule where a valid answer either corrects the error or shows the displayed figure while clearly flagging the error.

Programmatic CheckA rule graded automatically by software looking for specific values or patterns. Accepted values must cover every valid format.

Expert / Rubric JudgeA rule graded by judgment against its wording, so the wording must be clear and objective.

SeverityHow much a rule matters: Critical Requirement, Important Objective or Quality Polish.

Certification StatusYour final verdict on the task: Approved As-Is, Approved With Updates or Flagged.

Sign-offWhen Human Edge approves your task and delivers it to the client. Your task becomes payable here.

Productive timeTime Insightful logs while you use the project's apps or do web research for the task. Only this time is paid.

➡️

Ready?

Next you'll see the 9-step process from onboarding to sign-off. After that, Your Task walks you through the work itself.

⏰ All 9 steps, including sign-off, must be complete by Fri, Nov 13 · 11:59 PM ET.

Assignment

From Onboarding to Sign-off

Your task moves through 9 steps. Click any step to highlight it. Turnaround times for each step are shared when your task is assigned.

1

Onboarding

You

2

Platform setup

You

3

Creation

You

4

Automatic QA check

Human Edge

5

Revision 1

You

6

Peer review

You + a peer

7

Revision 2

You

8

Final QA check

Human Edge

9

Sign-off & delivery

Human Edge

Your actionDone by Human EdgePeer review

✎ Your Action

Step 1

Onboarding

Sign your Terms of Engagement (sent through a Dialectica platform). Then read this brief end to end.

⏱ Before you start

✎ Your Action

Step 2

Platform setup

Join Slack, download Insightful from your 2 invite emails, and open your task folder in Google Drive. See details below ↓

⏱ Before you start

✎ Your Action

Step 3

Creation

Complete modules A to E: build the Golden Output, calibrate the rubric and fill in the sign-off. See Your Task.

⏱ Counts toward the 10 h cap

⚙ Done by Human Edge

Step 4

Automatic QA check

Human Edge checks your work and sends you a plain-language QA report of what to fix.

⚙ Turnaround shared at assignment

✎ Your Action

Step 5

Revision 1

Fix every point in the QA report. You get one revision round.

⏱ Counts toward the 10 h cap

⇄ You + a peer

Step 6

Peer review

Another expert in your field comments on your Golden Output and rubric. You do the same for another expert's task.

⏱ 2 h cap for your review

✎ Your Action

Step 7

Revision 2

Address the peer reviewer's comments. You get one revision round.

⏱ Counts toward the 10 h cap

⚙ Done by Human Edge

Step 8

Final QA check

Human Edge confirms every fix is in and your task is consistent and client-ready.

⚙ Turnaround shared at assignment

⚙ Done by Human Edge

Step 9

Sign-off & delivery

Human Edge signs off and delivers your task to the client. From this moment, your task is payable.

🏁 By Fri, Nov 13 · 11:59 PM ET

Your time budget

Two separate hour caps. Time beyond a cap isn't paid. If you see you might go over, tell us on Slack early.

10 hmax

Creation + Revision 1 + Revision 2

■ STOP

0 h5 h10 h

Most tasks take 5–10 h (green). Time beyond 10 h isn't paid.

2 hmax

Peer review of another expert's task

■ STOP

0 h1 h2 h

Comments on their Golden Output and rubric. Time beyond 2 h isn't paid.

Step 2 in detail: your 3 tools

New to these apps? No problem. Each one takes a few minutes to set up.

📩

Expect 2 invite emails

One from Slack and one from Insightful, each with a link to join. Can't find them? Check your spam folder, then tell us.

Invite email

💬 Slack

What it is: A team chat app, like WhatsApp for work.

Why: It's our main communication channel for questions, updates and QA reports. Use Slack before email; we reply fastest there.

How: Open the invite email, click the link and follow the steps.

Invite email

⏱ Insightful

What it is: A time-tracking app you download and install on your computer.

Why: You're paid only for time Insightful logs as productive time.

How: Open the invite email, download and install the desktop app from the link, sign in, and track every work session.

▶ Watch the setup demo ↗

Shared by Human Edge

📁 Google Drive

What it is: Where your task files live and where you deliver.

Why: The client needs native Google Docs, Sheets and Slides.

How: Human Edge shares your task folder with you. Work only inside it.

🚫

Anti-AI Policy — Strictly Enforced

You may use AI tools (ChatGPT, Claude, Gemini or similar) only to brainstorm. Everything you create and deliver must be written by you: the Golden Output, rubric changes, justifications and peer review comments. Any AI-generated content is flagged, not delivered to the client and not paid.

⏰ Creation and both revisions must fit within 10 hours, with sign-off by Fri, Nov 13 · 11:59 PM ET.

Scope — Section 1

Your Task, Step by Step

Everything happens in your task folder in Google Drive. Follow modules A to E in order.

📥 In your task folder

The rubric spreadsheet [00 Task Brief & Grading Rubric]: a Google Sheet with 2 tabs (see the map below).

[Current Reference] draft: a rough first version of the deliverable.

[Source] files: the documents, sheets and emails behind the task.

→

📤 What you deliver

1. Golden Output: a new native Google file, saved in the same folder next to the [Current Reference] draft.

2. Calibrated rubric: filled in on the second tab, Grading Rubric Rules.

3. Sign-off: filled in on the first tab, Task Brief & Sign-Off.

Map of the rubric spreadsheet

The [00 Task Brief & Grading Rubric] Google Sheet has 2 tabs. Read the grey items; fill in the green ones.

First tab: Task Brief & Sign-Off Modules A & E

User Prompt (Do Not Edit): what the client's user asked forRead

Step 1, Step 2, Step 3: instructions specific to your taskRead

Certification StatusFill in

Link to Finalized Reference DeliverableFill in

Verified Reconciled vs. Raw Figures SummaryFill in

Summary of Rubric Rules Edited or RemovedFill in

Second tab: Grading Rubric Rules Module D

Rule ID to Rule Rationale: the existing rule and how it's gradedRead

Expert Action: KEEP AS-IS, EDIT or REMOVEEvery rule

Updated Criterion or Acceptable Alternative ValuesIf EDIT

Expert Justification / Notes: one line explaining your decisionEvery rule

🧩

The files are messy on purpose

Some [Source] files hold what you need; others are noise or contradict each other. That's deliberate: the client wants models that perform well even when information is incomplete and context is thin. Your expert judgment is the point. Decide what matters and resolve conflicts the way you would for a real client.

Your Task

Verify, Author & Calibrate

~5–10 hours

#	Component	What you need to do	Evaluation Criteria
A	Understand the task First tab: Task Brief & Sign-Off	▸Read the User Prompt — What is asked, for whom and in which format. Don't edit it. ▸Read Step 1, Step 2 and Step 3 — Your task-specific instructions for modules B, C and D. ▸Skim the second tab — Learn what will be graded. Don't judge yet.	Clear on the askClear on the format
B	Understand the [Source] files First tab: Step 1	▸Read every file in depth — Know what each one says and why it's there. ▸Map how they connect — Which sheet feeds which deck or memo, and which figures should match across files. ▸Verify the math yourself — Check formulas, ranges, totals and clauses. Most tasks hide a cross-file or formula discrepancy. ⚠ Never edit the [Source] files	Sources understoodDiscrepancies found
C	Build the Golden Output First tab: Step 2	▸Create a new native Google file — In the same folder, next to the [Current Reference] draft. ▸Upgrade the draft — Use it as a starting point, verify every fact and rewrite it to executive grade in your own words. ▸Show discrepancies clearly — Give the corrected figure, the displayed figure and the cause. ▸Office is fine for drafting — Convert to Google format and re-check everything (see Module C below).	AccurateCompleteExecutive-gradeNative Google file
D	Calibrate the rubric Second tab: Grading Rubric Rules · First tab: Step 3	▸Decide on every rule — Choose KEEP AS-IS, EDIT or REMOVE in the Expert Action column. ▸Justify every decision — One line in the Expert Justification / Notes column, for every rule. ▸Write the new wording for each EDIT — In the Updated Criterion or Acceptable Alternative Values column. ▸Add missing rules — As new rows at the bottom of the tab. ⚠ Never invent facts to make a rule work	Every rule decidedEvery rule justifiedNothing invented
E	Self-check & sign off First tab: Task Brief & Sign-Off	▸Grade your own Golden Output — Against your final rubric. It must pass 100% of the rules. ▸Fill in the 4 sign-off fields — At the bottom of the first tab (details below). ▸Tell Human Edge on Slack — As soon as you finish, so the automatic QA check can start.	Passes 100% of rulesSign-off complete

Module C: building the Golden Output

The first tab's Step 2 tells you exactly what to build for your task. The standards below apply on top.

1

Create the fileA new Google Doc, Slides or Sheet in your task folder, next to the [Current Reference] draft.

2

Upgrade the contentStart from the draft, verify every figure against the [Source] files and rewrite it in your own words.

3

Call out discrepanciesShow the corrected figure, the displayed figure and the exact cause (e.g., a SUM range that skips a row).

4

Link itPaste the file's link in Link to Finalized Reference Deliverable on the first tab.

Strategy & Management Consultants

Executive deck in Google Slides

4–6 slides, max 110 words and 5 bullets per slide
Action titles: the bottom line first
Comparisons in native Slides tables, never pasted text tables
An audit callout with the corrected figure, the displayed figure and the cause

Legal Counsel & Controllers

Audit memo in Google Docs

Native heading styles (Heading 1, 2, 3), never typed # symbols
Opens with an Executive Summary: 1 paragraph plus key takeaways
Comparisons in native Docs tables, never pasted text tables
Both the reconciled and the displayed total wherever a file has an error

🔁

Prefer Word, Excel or PowerPoint?

Draft there if you like. Then upload the file to your task folder and open it as a Google Doc, Sheet or Slides file (in Drive: File › Save as Google Docs / Sheets / Slides). Re-check accuracy, content and format line by line, since conversion can break formulas, numbers, tables and layouts. The Golden Output you link must be the native Google version.

Module D: deciding on each rubric rule

Ask these questions in order for every rule on the second tab, Grading Rubric Rules.

1. Is the rule a duplicate of another rule, or does it grade a document or tab the User Prompt didn't ask for?

Yes → REMOVE

No ↓

2. Does it need a fix so that it matches the [Source] files and every valid expert answer passes?e.g., accept the corrected figure or the displayed figure with the error flagged; accept plain business wording instead of cell references like "F118"; accept standard formats like $5.79K or ($3,190)

Yes → EDIT

No ↓

KEEP AS-IS

Finally, across the whole rubric: is anything essential not tested? → ADD a new row at the bottom of the tab.

✍️

Every rule gets a justification

Whatever you choose (KEEP AS-IS, EDIT or REMOVE), write one line in Expert Justification / Notes. For EDIT, also write the new wording in Updated Criterion or Acceptable Alternative Values.

Examples

📎 Fictional task: "Write a memo for the CFO reconciling the Q3 marketing budget." The budget sheet shows a $46,500 total because its SUM skips the Events row ($1,500). The real total is $48,000.

Rule"The memo names the Events line ($1,500) as the row missing from the sheet's total."

Expert ActionKEEP AS-IS

JustificationMatches the sheet's formula, specific and easy to grade.

Rule"The memo states the Events line as $1,500.00."

Expert ActionEDIT

Updated criterion"Accept $1,500, $1,500.00 or $1.5K."

JustificationStandard formats were being rejected.

Rule"The memo states the Q3 total as $48,000."

Expert ActionEDIT

Updated criterion"Accept $48,000, or $46,500 when the memo flags that the sheet's total skips the Events row."

JustificationAn expert could validly present either figure, as long as the error is called out.

Rule"The memo corrects the $2,000 travel line."

Expert ActionEDIT

Updated criterion"The memo explains the $1,500 gap between the sheet total and the real total."

JustificationNo travel line exists in any [Source] file. Rewritten to test the real discrepancy instead of inventing one.

Rule"The memo includes speaker notes."

Expert ActionREMOVE

JustificationThe User Prompt asks for a memo, so speaker notes are out of scope.

New row"The memo recommends fixing the sheet's SUM range before the Q4 close."

Expert ActionADD (new row)

JustificationA senior controller would insist on it, and no existing rule tests it.

Module E: the sign-off on the first tab

At the bottom of the first tab, Task Brief & Sign-Off, fill in all 4 fields:

Certification Status: Approved As-Is, Approved With Updates or Flagged (see below)Fill in

Link to Finalized Reference Deliverable: the link to your Golden OutputFill in

Verified Reconciled vs. Raw Figures Summary: each corrected figure next to the displayed oneFill in

Summary of Rubric Rules Edited or Removed: which rules you changed or added, and whyFill in

Approved As-IsThe rubric needed no changes, and your Golden Output passes every rule.

Approved With UpdatesYou edited, removed or added rules, and the task is now correct.

FlaggedAn issue remains that you can't resolve within the files. Explain it in your summary.

⚖️

The golden rule: judgment yes, invention no

Where the files are ambiguous, use your expert judgment. Where a rule or the draft refers to something the [Source] files don't contain, don't invent it: EDIT the rule to test what the files actually show, and explain why in Expert Justification / Notes.

📌

Submission Platform

Work only in your task folder in Google Drive. Never edit the [Source] files or the User Prompt. When modules A to E are done, tell Human Edge on Slack right away.

⏰ Both revisions and the final QA check must be done in time for sign-off by Fri, Nov 13 · 11:59 PM ET.

Scope — Section 2

Quality & Review

Every task is checked twice by Human Edge and once by a peer before it reaches the client.

CreationStep 3

→

Automatic QA checkStep 4 · Human Edge

→

Revision 1Step 5

→

Peer reviewStep 6

→

Revision 2Step 7

→

Final QA checkStep 8 · Human Edge

→

✓ Sign-offStep 9 · delivered

✕ If issues remain at the final QA check, the task can't be signed off. It isn't delivered to the client and isn't payable.

Steps 4 & 8

What Human Edge checks

At a high level, your task is checked for:

Accuracy: every fact matches the [Source] files
Completeness: the Golden Output answers the whole User Prompt
Rubric quality: every rule is decided, justified, clear and gradable
Consistency: the Golden Output passes every rule
Format: executive-grade, native Google file, sign-off complete
Originality: your own work, with no AI-generated content

After Step 4

What you get back

A short, plain-language QA report listing what to fix, most important first.

Fix every point in one revision round (Revision 1)
Unclear point? Ask on Slack before you revise

Step 6

Peer review

You review another expert's task, and another expert reviews yours. The review covers 2 things only: the Golden Output and the rubric.

Leave comments directly in their files (Google Docs / Slides / Sheets comments)
Comment; don't edit their work
Be specific: what's wrong, where, and how to fix it
2-hour cap for your review

✓

Revision Rounds

1 revision round only after the automatic QA check (Revision 1) and 1 after the peer review (Revision 2). Make each one count.

⏰ Only tasks signed off by Fri, Nov 13 · 11:59 PM ET are delivered and payable.

Scope — Section 3

Payment & Time

Compensation

You're paid your hourly fee agreed with Human Edge for the productive time Insightful logs, on every task that reaches sign-off.

A task becomes payable at Human Edge sign-off, the moment it is delivered to the client. Payment arrives by bank transfer within 15 business days of sign-off.

Your time budget

10 hmax

Creation + Revision 1 + Revision 2

■ STOP

0 h5 h10 h

Most tasks take 5–10 h (green). Time beyond 10 h isn't paid.

2 hmax

Peer review of another expert's task

■ STOP

0 h1 h2 h

Comments on their Golden Output and rubric. Time beyond 2 h isn't paid.

When you get paid

Fill out the tax formBefore sign-off

→

✓ Human Edge sign-offTask delivered

→

Share bank detailsWhen Human Edge asks

→

Bank transferWithin 15 business days of sign-off

✓ Payment Trigger

✓Your task passes the final QA check and receives Human Edge sign-off
✓Sign-off happens by Fri, Nov 13 · 11:59 PM ET
✓Your time is logged in Insightful as productive time and within the caps
✓Your work contains no AI-generated content

📋 Your Responsibilities

✓Log every work session in Insightful
✓Stay within 10 h (creation + revisions) and 2 h (peer review)
✓Submit the tax form before sign-off — payments cannot be processed until this is on file
✓Share your bank details when Human Edge requests them at sign-off

Task passes all stages and is signed off by the deadline

Payable

Work time not logged as productive time in Insightful (e.g., Insightful off, or time in AI apps)

Not paid

Creation + revisions go beyond 10 h, or peer review beyond 2 h

Extra time not paid

Issues remain at the final QA check, so the task isn't signed off

Not payable

AI-generated content is found

Not delivered · not paid

Expert abandons assignment after starting and/or does not meet the delivery deadlines

Not payable

ℹ️

Insightful

Start Insightful whenever you work on the task, including revisions and peer review, and stop it when you finish. Productive time covers the apps this project needs and web research for your task; AI apps are flagged. Need help setting it up? Watch the setup demo ↗

⏰ Reminder: every task must be fully signed off by Fri, Nov 13 · 11:59 PM ET.

Support

Questions & Contacts

Stuck? Ask on Slack first.

Slack is the project's main communication channel, and we reply fastest there. Use email only if Slack isn't working for you.

Email the PMs ↗

Lead Project Manager

Esteban Toro

esteban.toro@humanedgetech.ai

Project Manager

Thomas Manser

thomas.manser@humanedgetech.ai

Can I use AI tools (ChatGPT, Claude, Gemini) for any part of this work?⌄

Only to brainstorm. Everything you create and deliver (the Golden Output, rubric changes, justifications and peer review comments) must be your own work. Any AI-generated content is flagged, not delivered to the client and not paid.

How many revision rounds do I get?⌄

Two in total: 1 after the automatic QA check (Revision 1) and 1 after the peer review (Revision 2).

When and how will I be paid?⌄

By bank transfer, within 15 business days of Human Edge sign-off (when your task is delivered to the client), for the time logged in Insightful. Submit the tax form before sign-off; Human Edge will ask for your bank details at sign-off.

What happens if I miss a deadline?⌄

Every task must be fully signed off by Fri, Nov 13 · 11:59 PM ET. Turnaround times for each step are shared when your task is assigned. A task that isn't signed off by the deadline isn't delivered to the client and isn't payable.

Who do I contact for questions during the project?⌄

Message us on Slack first, since we reply fastest there. If Slack isn't working, email Esteban Toro (Lead PM) at esteban.toro@humanedgetech.ai or Thomas Manser (PM) at thomas.manser@humanedgetech.ai.

Which time is paid?⌄

Only time Insightful logs as productive time: work in the apps this project needs plus web research for your task. AI apps are flagged. Time with Insightful switched off can't be counted, so start it every time you work on your task, its revisions or a peer review.

Where exactly do I work?⌄

In your task folder in Google Drive. The rubric spreadsheet [00 Task Brief & Grading Rubric] has 2 tabs: the first, Task Brief & Sign-Off, holds the User Prompt, your task-specific instructions and the sign-off fields; the second, Grading Rubric Rules, is where you calibrate the rubric. Your Golden Output is a new Google file next to the [Current Reference] draft.

What is the Golden Output?⌄

The best-possible version of the deliverable the User Prompt asks for: accurate, complete and executive-grade, as a native Google Doc, Sheet or Slides file. You create it by upgrading the [Current Reference] draft.

What if a rule asks for something that isn't in the [Source] files?⌄

Don't invent it. EDIT the rule so it tests what the files actually show, and explain why in Expert Justification / Notes. If the issue can't be resolved within the files, set the Certification Status to Flagged and explain it in your sign-off summary.

Can I edit the [Source] files or the User Prompt?⌄

No. Use them as evidence only.

Can I draft in Word, Excel or PowerPoint?⌄

Yes. Then convert the file to Google format in your task folder and re-check accuracy, content and format line by line, since conversion can break formulas, numbers, tables and layouts. The Golden Output you link must be the native Google version.

What does the peer review cover?⌄

Only the Golden Output and the rubric. Leave comments directly in the other expert's files, without editing them, within the 2-hour cap.

What if I might not finish within the 10-hour cap?⌄

Tell us on Slack as early as possible, before you reach the cap. Time beyond the cap isn't paid.

✓ Acknowledgment

By beginning work on your task, you confirm that you have read this brief, understand the quality standards, time caps and payment conditions (including that only productive time logged in Insightful is paid), agree to the Anti-AI Policy, accept the limit of one revision round after the automatic QA check and one after the peer review, and commit to the deadline of Fri, Nov 13 · 11:59 PM ET. If anything is unclear, contact the project team on Slack before starting.