Sign in to continue
dlivrd Academy · dlivrd Technology Inc.
dlivrd Academy
Welcome back
🤖
AI Tools
dlivrd × Claude
AI Access Qualification Program. Complete 13 steps to earn your Claude seat.
13 steps
~45 min
Not started
+
More programs coming soon
Check back for new training
dlivrd Academy · dlivrd Technology Inc.
Access temporarily locked
You need to wait before retaking.
A failed attempt was recorded from this device. You must wait 7 days before trying again.
Time remaining
Calculating...
Questions? sam.hart@dlivrd.io
Welcome back
Pick up where you left off.
Your progress has been saved.
Your progress
dlivrd is adopting AI.
Here's what that means for you.
We're rolling out Claude, an AI assistant by Anthropic, to help our teams work smarter. This isn't about replacing people. It's about giving our best people better tools.
~45
Minutes
4
Quizzes
80%
To pass
You can pause and resume anytime. Progress is saved automatically so you never lose your place.
Why now?
AI is changing how work gets done across every industry. dlivrd is being intentional, so we move fast, without moving recklessly.
Why Claude?
Claude is built by Anthropic, one of the world's leading AI safety companies. It's designed to be helpful, honest, and safe, with enterprise-grade privacy controls.
Why selective access?
Access is earned, not given. We want people who understand Claude and have a real use case. That's what this program is for.
Your 9-step journey to Claude access
Step 1, Welcome & overview (you are here)
Step 2, Video modules & prompting cheat sheet
Step 3, dlivrd usage policy + policy check
Step 4, Learning module (5 lessons)
Step 5, Foundation quiz
Step 6, Register for access
Step 7, dlivrd scenario briefing
Step 8, Scenario quiz (90% to pass)
Step 9, Use case submission
Step 2 of 13, Video modules ▼ Watch on YouTube if video doesn't load ↗ Watch on YouTube if video doesn't load ↗ ▼ Watch on YouTube if video doesn't load ↗ ▼ Watch on YouTube if video doesn't load ↗ ▼
Watch these first.
3 video modules covering Claude foundations, Co-work, and prompting. Complete all 3 in order, each one unlocks the next.
1
Module 1, Claude foundations & interface
1
Claude foundations
The official Anthropic intro: prompting framework, models, Projects, Artifacts, and Research mode.2
Full beginner guide to Claude
Models, sidebar, Projects and memory, Connectors, Co-work intro, mobile app, and free vs Pro vs Max.2
Module 2, Claude Co-work
1
Getting started with Claude Co-work
Organize files, build spreadsheets from receipts, create reports from notes, and set up scheduled tasks.3
Module 3, Prompting techniques
1
Claude in practice
Real examples of Claude being used in professional workflows, not just demos.★
Bonus, Prompting styles cheat sheet
There's no single "right" way to prompt. Different tasks need different techniques. Bookmark this, it's your go-to reference when Claude isn't giving you what you need.
01
Role prompting
When: you need expert-level depth or a specific perspective
"Act as a senior logistics operations manager with 15 years of last-mile delivery experience. Review this SLA and flag any clauses that are unrealistic for urban routes."
Why it works: assigning a role shifts Claude's vocabulary, assumptions, and depth of analysis immediately.
02
Few-shot (show examples)
When: you need Claude to match a specific format, tone, or style
"Here are two examples of how we write update emails to clients: [Example 1] [Example 2]. Now write one for this week's delay situation: [context]."
Why it works: examples remove ambiguity. Claude reverse-engineers your pattern and replicates it precisely.
03
Chain of thought
When: the task involves logic, decisions, or multi-step analysis
"Think through this step by step before answering: We have 3 delivery routes, each with different cost and time tradeoffs. Which should we prioritize for peak season?"
Why it works: asking Claude to reason before concluding dramatically reduces errors on complex problems.
04
Constraint-based
When: you need tight, usable output, not a wall of text
"Summarize this partner agreement in exactly 5 bullet points. Each bullet must be under 20 words. Focus only on obligations and penalties."
Why it works: constraints force prioritization. Claude can't ramble when you've defined the box it must fit into.
05
Audience targeting
When: the same content needs to land differently for different readers
"Explain our new tracking feature. Version A: for a non-technical operations manager. Version B: for an engineer integrating our API."
Why it works: Claude adjusts vocabulary, assumed knowledge, and framing based on who's reading.
06
Iterative refinement
When: the first draft is close but not quite right
"Good start. Now make it 30% shorter, change the tone from formal to direct, and move the risk section to the top."
Why it works: Claude holds context within a conversation. You don't rewrite, you redirect.
07
Devil's advocate
When: you need to stress-test a plan or decision
"Here's our Q4 expansion plan. Argue against it as strongly as possible. What would a skeptical investor say? What could go wrong?"
Why it works: Claude is excellent at steelmanning opposing views. Use it to find holes before your critics do.
08
Structured output
When: the output needs to slot into a doc, system, or template
"Extract the key information from this email thread and return it as a table with 4 columns: Action item | Owner | Deadline | Priority."
Why it works: defining the exact output structure makes Claude's response immediately usable.
💡 Pro tip, combine techniques
The best prompts stack multiple styles. Role + chain of thought + constraint = a very powerful combination.
Step 4 of 13, dlivrd policy
dlivrd Claude Usage Policy
Read the full policy. Scroll to the bottom, open the document, confirm all checkboxes, then pass a 3-question policy check before continuing.
dlivrd Technology Inc.
Artificial Intelligence Usage Policy
Effective: July 2025 · Applies to: All dlivrd employees, contractors, and program participants
1. Why This Policy Exists
dlivrd is adopting AI deliberately and responsibly. Claude gives our teams a powerful tool to work faster, think more clearly, and deliver better results. But that power comes with responsibility. This policy protects our clients, protects dlivrd, and protects you.
2. Scope & Who This Applies To
This policy applies to every person who accesses Claude through dlivrd's AI Access Program, including full-time employees, part-time staff, contractors, vendors, and interns. Access is personal, role-specific, and non-transferable.
3. What You Can Use Claude For
- Written communications, drafting, editing, and improving emails, Slack messages, client updates, partner communications, and internal memos
- Document work, summarizing reports, meeting notes, proposals, contracts (non-confidential), and operational documents
- Operational analysis, analyzing non-confidential route data, SLA performance, delivery metrics, and process inefficiencies
- Engineering support, writing, reviewing, and explaining code; drafting PR descriptions; generating test cases
- Project management, turning meeting notes into action items, drafting project briefs, building timelines
- Research & synthesis, gathering market intelligence, summarizing industry trends, comparing vendor options
- Content creation, writing internal documentation, training materials, templates, and marketing copy
- Strategic thinking, brainstorming, stress-testing plans, generating alternatives
4. What You Must Never Do
- Input confidential client data, no client names, addresses, delivery volumes, pricing agreements, or SLA terms without written approval from your manager and IT Security
- Input employee or HR data, salaries, performance reviews, disciplinary records, or any PII about dlivrd staff
- Input system credentials or security data, passwords, API keys, access tokens, database connection strings
- Submit output without review, never send, publish, or act on Claude's output without reading and verifying it yourself. No exceptions, even under deadline pressure
- Share your login or access, your Claude seat is assigned to you personally. Zero exceptions
- Use Claude for personal commercial work, dlivrd's subscription exists for dlivrd business only
- Attempt to bypass Claude's safety guidelines, no prompt injection, jailbreaking, or role-play exploits
- Present AI output as original human work, in legal filings, regulatory submissions, investor materials, or certifications, you must disclose AI assistance
5. Data Privacy & Confidentiality
dlivrd accesses Claude through Anthropic's enterprise-tier API. Prompts are not used to train Anthropic's models. However, enterprise protection does not change your obligation to handle data responsibly. Treat every prompt as a professional communication that could be reviewed by your manager or Legal. When in doubt, leave it out, contact sam.hart@dlivrd.io before proceeding.
6. Human Accountability, No Exceptions
Claude is a tool. Every piece of content, every analysis, every communication that flows from your use of Claude is your professional responsibility, not Anthropic's, not dlivrd's AI program's, and not Claude's.
"Claude generated it" is not a defense. If you send a client an incorrect SLA summary that Claude wrote and you didn't verify, that error belongs to you.
7. Understanding Claude's Limitations
- Hallucination, Claude can generate factually incorrect information with complete confidence and fluency. Never treat its output as ground truth
- Knowledge cutoff, Claude's training has a cutoff date. It cannot access current events, live delivery data, or current regulations
- No memory between sessions, Unless you are using a configured Project, Claude starts each conversation fresh
- No judgment, Claude cannot assess whether an action is right for dlivrd's relationships or strategic context
- No legal or financial authority, Claude's output on legal, tax, or financial matters is informational only
8. Quality Standards for Claude Output
- Read the entire output, not just the first paragraph
- Verify any specific facts, numbers, dates, names, or statistics against a primary source
- Confirm the tone, framing, and content are appropriate for the audience
- For client-facing content: have a second team member review before sending
- For legally sensitive content: route through Legal before use
9. Access Management
- Steps 1,5, Complete the foundation training: videos, policy, learning module, and foundation quiz
- Steps 6,9, Register, watch the dlivrd briefing, pass the scenario quiz, and submit a use case
- Access granted, Upon approval, your Claude seat is activated within 1,2 business days
Access may be suspended or revoked for policy violations, extended inactivity, or a role change. A significant change in responsibilities requires resubmission of your use case.
10. Reporting Violations
Reports go to sam.hart@dlivrd.io. All reports are treated confidentially. If you are unsure whether something is permitted, ask before acting.
dlivrd Technology Inc. · AI Usage Policy · Version 1.0 · July 2025 · Owner: Samantha Hart
Scroll to read
REQUIRED, open the policy document before continuing
Open full policy document ↗
Official dlivrd Claude Usage Policy, opens in a new tab
Confirm you have read and understood dlivrd's Claude Usage Policy.
QUESTION 1 OF 3
0 / 0
Step 6 of 13, Learning module 0 / 5 complete
Step 7 of 13, Lessons quiz 0 / 0 correct
QUESTION 1 OF 10
Step 8 of 13, Prompting cheat sheet
8 techniques for better prompts.
Read all 8 techniques. A 10-question quiz follows before you can continue.
01
Role prompting
When: you need expert-level depth or a specific perspective
"Act as a senior logistics operations manager with 15 years of last-mile delivery experience. Review this SLA and flag any clauses that are unrealistic for urban routes."
Why it works: assigning a role shifts vocabulary, assumptions, and depth of analysis immediately.
02
Few-shot (show examples)
When: you need Claude to match a specific format, tone, or style
"Here are two examples of how we write client update emails: [Example 1] [Example 2]. Now write one for this situation: [context]."
Why it works: examples remove ambiguity. Claude reverse-engineers your pattern and replicates it precisely.
03
Chain of thought
When: the task involves logic, decisions, or multi-step analysis
"Think through this step by step before answering: We have 3 delivery routes with different cost and time tradeoffs. Which should we prioritize for peak season?"
Why it works: asking Claude to reason before concluding dramatically reduces errors on complex problems.
04
Constraint-based
When: you need tight, usable output, not a wall of text
"Summarize this partner agreement in exactly 5 bullet points. Each bullet must be under 20 words. Focus only on obligations and penalties."
Why it works: constraints force prioritization. Claude cannot ramble when you have defined the box it must fit into.
05
Audience targeting
When: the same content needs to land differently for different readers
"Explain our new tracking feature. Version A: for a non-technical operations manager. Version B: for an engineer integrating our API."
Why it works: Claude adjusts vocabulary, assumed knowledge, and framing based on who is reading.
06
Iterative refinement
When: the first draft is close but not quite right
"Good start. Now make it 30% shorter, change the tone from formal to direct, and move the risk section to the top."
Why it works: Claude holds context within a conversation. You redirect, not rewrite.
07
Devil's advocate
When: you need to stress-test a plan or decision
"Here's our Q4 expansion plan. Argue against it as strongly as possible. What would a skeptical investor say? What could go wrong?"
Why it works: Claude is excellent at steelmanning opposing views. Use it to find holes before your critics do.
08
Structured output
When: the output needs to slot into a doc, system, or template
"Extract the key information from this email thread and return it as a table with 4 columns: Action item, Owner, Deadline, Priority."
Why it works: defining the exact output structure makes the response immediately usable, no reformatting needed.
Pro tip: combine techniques
The best prompts stack multiple styles. Role + chain of thought + constraint is a very powerful combination: "Act as [role]. Think step by step. Answer in [format] under [limit]."
✓
Foundation complete.
You've watched all modules, read the policy, passed the policy check, completed 5 lessons, and passed the foundation quiz. Steps 1,5 are locked in.
YOUR FOUNDATION RECORD
Videos completed4 / 4
Policy read & acknowledged✓
Policy check passed✓
Lessons completed5 / 5
Foundation quiz score,
Steps 6,9 are next, register, watch the dlivrd briefing, pass the scenario quiz, and submit your use case to earn your Claude seat.
Step 10 of 13, Register for access
Register your application.
Enter your details below. Only approved roles can proceed to Steps 7,9.
10
Register
Your name, email, and role, only approved roles can proceed
11
dlivrd briefing
Timed slides on how we specifically use Claude, must read each one
12
Scenario quiz
10 scenario questions, 90% to pass, randomized every attempt
13
Use case
Describe a real task in your role, vague answers are rejected
Claude isn't available for your role yet.
Wave 1 is open to PMs, Engineers, EAs, and Marketing. Here are free resources to get you started with Google Gemini in the meantime.
Gemini for Beginners, 25 min
Kevin Stratvert · Free · Gmail, Drive, Docs productivity
Free Gemini Course with Certificate
Great Learning · Free · No coding required
Notify me when Wave 2 opens
Questions? sam.hart@dlivrd.io
Step 13 of 13, Use case submission
Prove your intent.
Describe a specific, real task in your current role where Claude would save meaningful time or improve quality. Sam Hart reviews every submission personally.
WHAT GETS REJECTED
Rejected: too vague
"I could use Claude to help with my work and save time on various tasks."
Rejected: no specifics
"Claude would help me write emails and summarize things faster."
WHAT GETS APPROVED
Specific, real, and role-grounded
"Every Monday I spend 2 hours compiling our weekly ops report from 6 Slack threads and 3 spreadsheets. Claude could help me structure and draft the narrative in 20 minutes from my bullet notes. I'd then verify it against raw data before sending to leadership."
YOUR SUBMISSION MUST INCLUDE:
1. What is the specific task you will use Claude for at dlivrd?
2. How will you verify Claude's output before using it?
3. What is the business impact if this works well?
2. How will you verify Claude's output before using it?
3. What is the business impact if this works well?
0 / 50 words minimum
Questions? sam.hart@dlivrd.io
Optional, go deeper while you wait ↗ ↗ ↗ ↗
Keep learning. No gate, no requirement.
Your application is in. These are the best free resources for getting better at Claude before your seat is activated.
Anthropic Academy, official free courses
Claude 101
Introduction to Claude Cowork
AI Fluency: Framework & Foundations
AI Capabilities & Limitations
Videos worth watching
Coursera, free to audit