PART 03 Field Guide

Stop doing manually
what AI can do
in minutes

Youth workers spend an average of 40% of their working hours on administration. That's time not spent with young people. This section is about reclaiming it — not through shortcuts that compromise quality, but through AI that handles the mechanical parts of knowledge work.

📄
Grant application
3–5 days
📊
Funder report
1–2 days
📅
Programme planning
4–6 hours
✉️
Partner communications
2–3 hours/week
🔍
Research & evidence
Half a day
📱
Content creation
2–3 hours per post

Typical time investment. With AI assistance, most of these drop by 60–80%.

AI doesn't replace your expertise, your relationships, or your judgment. It handles the mechanical parts — first drafts, structure, research starting points, formatting. You review, verify, adapt, and make it yours. Everything in this section assumes that workflow.

Grant writing & funding

From months
to weeks.

Use AI at each stage where human thinking is least required — freeing you for the parts that only you can do.

01
Research the funder
Understand priorities before writing
02
Build your evidence base
Current statistics and policy context
03
Draft section by section
Chain prompts, full context, one section at a time
04
Review and strengthen
AI as critical reader before human eyes

Grant writing is where AI delivers some of its clearest value — not by writing applications for you, but by compressing the weeks of research, drafting, and revision that a strong application requires. The knowledge of your project, your organisation, and your participants remains entirely yours. What changes is how long the mechanical parts take.

The key principle: never ask AI to write your whole application. Ask it to write sections, one at a time, with your specific context provided for each. The quality difference between "write my Erasmus+ application" and a series of focused, context-rich section prompts is enormous.

01 — Research the funder

Before writing a word, use Perplexity to research the funder: their strategic priorities, previous grants, stated language and values, what they explicitly say they will and won't fund. Paste the funder's guidance notes into Claude and ask it to extract the most important criteria and flag common applicant mistakes.

Prompt — funder research
You are an experienced grant writer preparing a funding application. I'm going to paste the funder's guidance document. Please:
1. List the 5 most important assessment criteria in order of emphasis
2. Identify any language or framing they use repeatedly that I should mirror
3. Flag any common mistakes or disqualifying factors they explicitly mention
4. Suggest 3 questions I should be able to answer clearly before I start writing

[PASTE FUNDER GUIDANCE HERE]
02 — Build your evidence base

The evidence section of any application is where most youth workers lose time. Use Perplexity's Deep Research to gather current statistics, research findings, and policy context. Ask specifically for EU-level data if applying to European funders, local data if applying domestically. Always note which figures need independent verification before submission.

Prompt — evidence base (use in Perplexity)
I am writing a funding application addressing [ISSUE] among young people in [REGION]. Using current sources, provide:
- 3 European-level statistics (with dates and sources)
- 2 national/regional statistics for [COUNTRY]
- 1 key research finding from the last 3 years
- The main EU policy framework this connects to

Flag any statistics that may need verification before use in a formal submission.
03 — Draft section by section

Use Claude for actual drafting — it handles long documents and maintains consistency across sections better than any other model. Feed it your context once at the start, then work through sections in order. Start by establishing context:

Prompt — context establishment (send this first)
I am writing an Erasmus+ KA1 application for a youth worker training project. Here is the key information:
- Organisation: [NAME], based in [COUNTRY]
- Project title: [TITLE]
- Partners: [LIST PARTNERS AND COUNTRIES]
- Target group: [WHO]
- Main activity: [WHAT, WHERE, WHEN]
- Core aims: [3 BULLET POINTS]

I will now ask you to help me draft sections one at a time. Keep this context for the whole conversation.

Then request each section individually. Claude will maintain consistency because the full context is in the conversation.

04 — Review and strengthen

Once you have a draft, use AI as a critical reviewer before human eyes see it. Two prompts that consistently improve applications:

Prompt — critical review
You are a critical reader evaluating this grant application section. Identify:
1. The three weakest claims or unsupported statements
2. Any jargon a non-specialist funder might not understand
3. Places where the link between activities and outcomes is unclear
4. Any repetition that should be cut

Be direct. This is a working document, not a final draft.

[PASTE SECTION]
Prompt — rewrite after review
Now rewrite the section addressing the weaknesses you identified. Keep all factual content. Strengthen the narrative logic. Same length or shorter.

Important: AI cannot write a grant application for you. A successful application requires your deep knowledge of the project, your relationships with partners, your understanding of participants' needs, and your credibility as an organisation. What AI can do is assist with the mechanical parts — researching the funder's priorities, building an evidence base, drafting section structure, and strengthening language. You remain the author. The expertise, the vision, and the accountability are entirely yours.

Reports & evaluation

The data is yours.
The structure
is AI's.

You have participant data, session notes, feedback forms, and your memory of what happened. AI doesn't create this evidence — it helps you transform it into coherent, well-structured narrative efficiently.

Funder reports are often the most dreaded part of a project cycle — and one of the best use cases for AI assistance, because the raw material already exists. The most common scenario: you have scattered notes, feedback forms, and data, and you need to turn them into 500 words of coherent impact narrative. This is where AI removes the most friction.

Transforming raw notes into narrative
Prompt — impact narrative from raw notes
You are an experienced grant writer producing a funder report. I'm going to give you raw notes, participant feedback, and data. Turn this into a 400-word impact narrative with the following structure:
- Context (what we set out to do) — 50 words
- What happened (key activities) — 100 words
- What changed (outcomes and evidence) — 200 words
- What comes next — 50 words

Write in past tense, active voice. Use 1–2 specific participant quotes if I provide them. No generic claims without evidence. Professional but warm tone.

RAW MATERIAL: [PASTE YOUR NOTES, DATA, AND FEEDBACK HERE]
Youthpass and learning outcomes

For Erasmus+ projects specifically, describing participants' learning outcomes in Youthpass terms is a task that AI handles remarkably well. It knows the eight key competences framework and can translate descriptions of activities into competence language — saving significant time on what is otherwise painstaking work.

Prompt — Youthpass learning outcomes
You are familiar with the Erasmus+ Youthpass framework and the eight key competences for lifelong learning. Based on the following activity description, suggest appropriate learning outcomes in Youthpass language for each relevant competence. Keep each outcome specific, measurable, and realistic for the activity described.

Activity description: [DESCRIBE THE ACTIVITY IN PLAIN LANGUAGE]
Evaluation questions and analysis

Use AI to generate well-structured evaluation questions before your project, then use it again to analyse the responses and identify themes after. For the analysis, paste anonymised responses and ask for thematic coding.

Prompt — evaluation questions (before the project)
Generate 8 evaluation questions for participants at the end of [TYPE OF TRAINING/EXCHANGE]. Cover: knowledge gained, skills developed, confidence change, intended application, and overall experience. Mix quantitative (scale questions) and qualitative (open text) formats. Avoid leading questions. Accessible language — assume mixed English levels.
Prompt — thematic analysis (after the project)
You are a qualitative researcher. I'm going to paste anonymised responses to an open evaluation question from [NUMBER] participants. Please:
1. Identify the 4–5 most common themes
2. Note any outlier responses worth highlighting
3. Select 3 quotes that best represent the range of experience (copy them exactly)
4. Write a 150-word summary suitable for a funder report

[PASTE ANONYMISED RESPONSES]

The data is yours. The structure is AI's. The voice is yours again.

Planning & programme design

From a single session
to a full programme.

AI's ability to hold large amounts of context and generate structured content simultaneously becomes most visible in planning. The principles are the same whether you're designing 45 minutes or 5 days.

Card 01
Session design

Describe your participants, your learning objective, your time, and your constraints. Ask for a structured session plan. Then iterate — ask it to make the main activity more participatory, to add a backup for low energy, to create an online version. Each refinement takes seconds.

Starting prompt
You are a non-formal education facilitator. Design a [LENGTH]-minute session for [NUMBER] participants aged [AGE] on [TOPIC]. Learning objective: by the end, participants will be able to [SPECIFIC OUTCOME]. Methods: participatory only, no lecturing. Materials available: [LIST]. Include facilitator notes throughout.
Card 02
Multi-day programme

Give Claude the full arc first: the overall learning journey, the daily themes, the balance between content and reflection. Then develop each day individually. Claude will maintain thematic consistency and flag logical gaps or pacing issues if you ask it to review the whole before you finalise.

Programme review prompt
Review this [NUMBER]-day programme outline and identify:
1. Thematic gaps or jumps that need bridging
2. Days that feel overloaded vs underloaded
3. Reflection time — is it sufficient and well-spaced?
4. Energy management — enough movement and variation?

Give specific suggestions for each issue.

[PASTE PROGRAMME OUTLINE]
Card 03
Adapting existing resources

One of the most underused AI applications in youth work: taking existing activities, toolkits, or programmes and adapting them for a new context. Different age group, cultural context, language level, time available, or delivery format. Describe the original and the new context — AI handles the adaptation while preserving the learning logic.

Adaptation prompt
I have the following activity designed for [ORIGINAL CONTEXT]. I need to adapt it for [NEW CONTEXT]. Keep the core learning objective identical. Adapt: language in instructions, time required, materials, group size management, and any cultural references. Flag anything needing sensitivity in the new context.

ORIGINAL ACTIVITY: [DESCRIBE OR PASTE IT]
A full planning chain — how it looks in practice

Four connected prompts to build a complete training day — each step feeds directly into the next. The whole sequence takes around 20 minutes instead of a full afternoon.

Step 01 — Overview Design a 1-day training on digital citizenship for 20 youth workers. Give me: learning arc (3 stages), 4 session titles with 1-line objectives, suggested timing. No content yet — just the structure.
Output Morning: Building awareness (3h). Afternoon: Practical skills (3h). Close: Reflection & transfer (1h). Sessions: 1. What is digital citizenship? 2. Navigating misinformation 3. Privacy & safety online 4. Teaching it to young people...
feeds in
Step 02 — Session 1 Now develop Session 1: "What is digital citizenship?" — 45 minutes, participatory, mixed language levels. Include: energiser (10 min), main activity (25 min), debrief (10 min). Facilitator notes throughout.
Output Energiser: "Digital identity cards" — participants draw or write 5 things about their online life (10 min). Main: "Digital citizenship spectrum" — large visual spectrum, pairs place scenarios... Debrief: 3 questions, whole group...
builds on
Step 03 — Session 2 Now develop Session 2: "Navigating misinformation" — 60 minutes. It follows Session 1 so participants already have a shared vocabulary. Main activity should build on the spectrum exercise. No repeat of concepts already covered.
Output Building on spectrum from Session 1: "Fact-checking relay" — 5 teams, each receives a viral claim to investigate using 3 verification tools. Teams present findings. Bridges to Session 3 on privacy...
then review
Step 04 — Full review Review the full day programme so far. Check: thematic coherence, energy flow, reflection balance, time totals. Identify any gaps or improvements before I finalise.
Output Strong thematic arc. Energy concern: Sessions 2 and 3 both cognitively heavy — suggest a movement break between them. Reflection time is well-distributed. Total time is 7h — suggest cutting energiser in Session 3 by 5 min to allow buffer...
Communication & admin

Small tasks.
Big drain.

Individually, each communication task feels too small to warrant a tool. Collectively, they represent hours of every working week. AI handles this category particularly well — the tasks are well-defined, the quality bar is clear, and iteration is fast.

Writing & Communication
✉️ Difficult emails

The emails that sit in your drafts folder for days — declining a partner, chasing an overdue deliverable for the second time, communicating a change of plan. AI helps with tone calibration: firm without being cold, direct without being abrupt.

Prompt
Write a professional email to [RECIPIENT TYPE] about [SITUATION]. Key points: [POINT 1], [POINT 2]. Tone: [e.g. "firm but warm"]. Relationship context: [DESCRIBE]. Maximum [NUMBER] words. End with a clear next step.
🌍 Multilingual accessibility

Participant information packs, programme schedules, house rules, safety guidelines — documents that need to be understood by people with varying English levels. Specify the target reading level and what to change.

Prompt
Rewrite the following text for someone reading in their second language at approximately B1 English level. Use: short sentences (maximum 20 words), no idioms or figurative language, define any specialist terms on first use. Keep all factual content identical.

[PASTE TEXT]
📋 Meeting preparation

Before a significant meeting: generate an agenda, draft discussion questions per item, anticipate concerns from different stakeholder perspectives, and prepare a follow-up email template — all in one prompt.

Prompt
I have a meeting with [WHO] about [TOPIC]. Main points to cover: [LIST]. Please:
1. Draft a 30-minute agenda
2. Suggest 2 discussion questions per agenda item
3. Anticipate 2–3 concerns [RECIPIENT] might raise and suggest responses
4. Draft a follow-up email template summarising decisions and next steps
Research & Admin
🔍 Evidence gathering

Before writing any document that makes claims about youth issues, use Perplexity to build your evidence base. Every claim Perplexity makes is linked to a verifiable source — you can build an evidence base you can actually stand behind.

Prompt — use in Perplexity
I am preparing materials on [TOPIC] for a professional youth work audience in [REGION]. Provide: 3 current statistics with sources and dates, 2 key research findings from the last 3 years, the main EU or national policy framework this connects to. Note any figures that need verification before formal use.
📄 Document summarisation

Policy documents, research reports, funding guidance, partnership agreements — paste any document into Claude and ask for what you actually need: plain-language summary, action points, key deadlines, the parts most relevant to your specific context.

Prompt
You are a professional summariser. Read the following document and provide:
1. A 150-word plain-language summary of the main points
2. A bullet list of any action points or requirements for our organisation
3. Any deadlines or important dates
4. The 2–3 sections most relevant to [YOUR SPECIFIC CONTEXT]

[PASTE DOCUMENT]
📝 Reference & admin documents

Reference letters, Terms of Reference for partnerships, volunteer agreements, privacy notices — documents that follow predictable structures but take time to draft from scratch. Give AI the specific details and ask for a full draft, then review and adapt.

Prompt
Draft a [DOCUMENT TYPE] for [CONTEXT]. Include: [KEY ELEMENTS TO COVER]. Tone: [DESCRIBE]. This will be used for [PURPOSE]. Draft in full — I will review and adapt to our organisation's specific style and requirements.
Working with young people using AI

AI isn't only
for the background.

Used thoughtfully, AI tools become powerful resources to use with young people — as a subject for critical discussion, as a creative collaborator, and as a means of removing barriers that prevent some young people from participating fully.

The crucial distinction: using AI on young people — generating content about them, making decisions affecting them — requires careful ethical consideration. Using AI with young people — as a transparent, explorable tool they can question and interrogate — is one of the most valuable things a youth worker can do right now.
Mode 01
AI as a subject

Explore it together — don't lecture about it. The most powerful sessions on AI with young people aren't presentations, they're explorations. Give them access to a tool and a question: "Ask it something you care about and see what happens. Then ask it the same thing differently." Let them discover hallucinations, biases, and capabilities themselves. Your role is to facilitate the reflection.

Questions worth exploring together
  • "Is this answer true? How would we check?"
  • "Why do you think it answered like that?"
  • "What did it leave out? What assumptions did it make?"
  • "Who might be harmed if someone trusted this answer?"
Mode 02
AI as a creative tool

Lower the barrier to expression. For young people who struggle with writing, language barriers, or confidence, AI can be a genuine equaliser — it can take a rough idea and help shape it into something they're proud of. The key facilitation principle: the young person's idea always comes first. AI refines and expands, it doesn't replace the original voice. Always ask: "Does this still sound like you?"

Practical uses
  • CV and covering letter drafting with young people who lack writing confidence
  • Translating their ideas into a different format (spoken idea → written pitch)
  • Generating options for creative projects they then select from and develop
  • Creating accessible versions of information they need
Mode 03
AI for accessibility

Remove barriers, not bypass them. AI accessibility tools — text-to-speech, captioning, translation, simplified language conversion — can meaningfully reduce participation barriers for young people with learning differences, hearing impairments, language barriers, or low literacy. The youth worker's role is to know these tools exist and frame them positively as tools everyone can use.

Tools worth knowing
  • Otter.ai — live transcription and meeting summaries
  • Microsoft Immersive Reader — adjustable text, read-aloud, translation
  • Gemini in Google Docs — instant translation and plain-language rewriting
  • ElevenLabs — high-quality text-to-speech for audio versions of written materials
Common mistakes

Five mistakes
worth avoiding.

These appear in almost every organisation's early AI adoption. They're all preventable — and each one has a reliable fix.

01
✗ Mistake
Trusting output without review

Copying AI-generated text directly into a grant application or funder report without reading it carefully. Statistics may be hallucinated. Voice may not match your organisation. Claims may not be verifiable.

✓ Fix
Treat all output as a first draft

Review for factual accuracy (especially statistics and dates), verify the voice sounds like your organisation, and check that any claims are ones you can substantiate. The efficiency gain comes from starting further ahead — not from skipping the review.

02
✗ Mistake
Sharing sensitive information

Pasting real participant names, case notes, or personal details into AI tools to get help writing reports. Depending on tool and account settings, that data may be logged.

✓ Fix
Anonymise before pasting

Replace names with roles ("Participant A"), remove identifying details, describe situations in general terms. The AI doesn't need the real names to help you — and protecting participant data is a non-negotiable organisational responsibility.

03
✗ Mistake
One-shot prompting for complex tasks

Asking for a full grant application, full programme design, or full evaluation report in one prompt. The model handles depth much better than breadth — breadth produces shallow results across all items.

✓ Fix
Break into a chain of connected prompts

Start with context and structure. Develop sections individually. Review and strengthen in separate passes. Quality increases dramatically with each additional step, and you maintain control throughout the process.

04
✗ Mistake
Not iterating

Getting a mediocre response and giving up, or manually editing the output without asking AI to fix it. Manual editing is slower and produces worse results than iterating with the model.

✓ Fix
Every conversation is iterative

"Make this shorter." "The tone is too formal." "Add a specific example." "Rewrite the opening." These follow-up instructions take seconds and often produce the output you needed from the start. Three iterations is faster than one round of manual editing.

05
✗ Mistake
Using AI for things that should stay human

Using AI to draft personalised responses to young people in crisis, or to inform safeguarding decisions where the specific person and context are everything.

✓ Fix
Know the line

AI is for the mechanical parts of knowledge work. Anything that requires knowing the specific person, their history, their emotional state, or the nuance of a live situation — that stays with you. Test: would you be comfortable telling the young person that AI helped write your response to them?

"

AI doesn't make youth workers less necessary. It makes the ones who use it well significantly more effective — with more time for the parts of the job that actually require a human.

Youthwork.AI · AI For Your Work · Part 03
Up next in this toolkit
Part 04 — NotebookLM & Learning
Continue reading