DEV Community: Manickavasagan

Types of Prompting: Complete Guide to Prompting Techniques

Manickavasagan — Wed, 27 May 2026 14:41:27 +0000

I spent two weeks blaming Claude for being useless. Turns out Claude wasn't the problem—I just had no idea how to ask it anything.

That's when I realized prompting isn't mystical. It's asking better questions. Different ways work for different jobs.

What Is Prompting?

It's how you ask AI to do something. But doing it well.

Asking someone to "make food" gets you whatever's in the fridge. Asking for "that pasta carbonara like you made last month" gets you exactly what you want. Same person. Different ask.

Here are the 7 techniques that actually matter:

1. Zero-Shot: Just Ask. No Examples.

You don't show examples. You just ask and see what you get.

These models learned patterns from massive amounts of text. So when you ask something new, it pattern-matches to what it knows.

Example:

"Write a haiku about AI"

Minds made of numbers
Learning patterns in the dark  
Future takes its shape

When to use: Quick translations, sentiment checks, summarizing articles.

How to get better results: Stop being vague.

❌ "Write about AI"
✅ "Write a 300-word intro for business managers about how AI reduces customer support costs—use actual examples"

2. Few-Shot: Show Examples, Then Ask

Give it 2-5 examples of what you want, then ask it to do the same with new data.

Why it works: When you show examples, the AI learns the pattern right there. The accuracy jump is ridiculous—I've seen 70% accuracy jump to 94% just by adding three examples.

Example:

Classify sentiment:

Review: "Amazing product! Love it!"
Sentiment: positive

Review: "Terrible quality, fell apart immediately"
Sentiment: negative

Review: "It's okay, nothing special"
Sentiment: neutral

Now this:
Review: "Best purchase ever!"
Sentiment: ?

AI: "positive"

Sweet spot: 3-5 good examples. Quality over quantity. One great example beats five mediocre ones.

3. Chain-of-Thought: Make It Show Its Work

Tell the AI to think through something step-by-step instead of just guessing.

Why it matters: Ask an AI to think step by step? It reasons better. You get better logic, fewer mistakes, and you can follow what it's doing.

Example:

Step by step:

Step 1: Books cost = 3 × $15 = $45
Step 2: Total cost = books + pen = $45 + $5 = $50
Step 3: Change = $100 - $50 = $50

Now solve: Alice buys...

When to use: Math problems, logic, analysis, debugging, planning. Anywhere you want to see how it got there.

4. Role/Persona Prompting: Tell It Who to Be

Assign the AI a role and it adopts that perspective and knowledge base.

Examples:

❌ "What should I consider for a startup?"
✅ "You're a serial entrepreneur who's built 3 startups. What should I consider when launching?"
❌ "Explain machine learning"
✅ "You're a professor explaining machine learning to undergrads. Explain it."

The response sounds like someone who knows what they're doing instead of generic AI voice.

5. System Prompting: Set the Baseline Rules

System prompting applies to everything the user sends. It's like briefing someone before they start work.

Example:

System: "You are a helpful customer support specialist.
- Be friendly and empathetic
- Solve problems quickly
- Keep responses under 200 words"

Every response follows those rules.

6. JSON Prompting: Get Data You Can Actually Use

Want structured output your code can parse? Tell the AI to return only JSON with a specific structure.

Example:

Extract customer info. Return only JSON:
{
  "name": "",
  "email": "",
  "issue": "",
  "priority": "low|medium|high"
}

You get parseable JSON instead of messy prose.

7. Context Prompting: Give It Source Material

Want accurate answers? Give the AI actual sources to reference.

Why it works: AI hallucinates. But give it a specific source? It mostly sticks to what's there. Accuracy jumps dramatically.

Example:

Based on this earnings report, what was Q3 revenue?

[paste report]

Return as JSON: {"quarter": "Q3", "revenue": ""}

Actual numbers from the document. Not made up.

Combining Techniques: When It Gets Powerful

Stack them:

Few-shot + role:

You're a fashion expert.

Classify style:

Item: "Oversized blazer, minimal jewelry"
Style: minimalist

Item: "Bright colors, bold patterns"
Style: maximalist

Item: "Neutral colors, classic cuts"
Style: ?

Chain-of-thought + few-shot:

Solve step by step:

Problem: Alice buys 3 books at $15 each, 10% discount. Cost?
Step 1: Before discount = 3 × $15 = $45
Step 2: 10% off = $45 × 0.10 = $4.50
Step 3: Final = $45 - $4.50 = $39.50

Now solve: Bob buys 2 shirts...

Quick Comparison

Type	Speed	Accuracy	Best For
Zero-shot	⚡⚡⚡	⭐⭐	Quick tasks
Few-shot	⚡⚡	⭐⭐⭐⭐	Production
Chain-of-thought	⚡⚡	⭐⭐⭐⭐⭐	Complex reasoning
Role/Persona	⚡⚡	⭐⭐⭐	Specific voice
System prompting	⚡⚡	⭐⭐⭐⭐	Production apps
JSON prompting	⚡⚡	⭐⭐⭐⭐	Automation
Context prompting	⚡	⭐⭐⭐⭐⭐	Accuracy critical

The Strategy

Start simple. Add examples if needed. Add structure for automation. Add context for accuracy. Combine what works.

Want the full guide with code examples, real-world projects, and common mistakes to avoid?

Read the complete interactive guide on TechBasics

It includes practical examples, code snippets, and how professionals actually use these techniques in production.

You Don’t Need a Backend: How I Built an AI Document Generator in the Browser

Manickavasagan — Fri, 15 May 2026 10:36:11 +0000

Why Most Apps Start With a Backend

When building web apps, most developers default to:

Frontend (React / JS)
Backend API
Database
Cloud hosting

This is often unnecessary for simpler use cases.

👉 In many cases, the browser is powerful enough to handle everything.

The Problem I Wanted to Solve

I wanted to build a tool that could:

Take a plain-English prompt
Generate structured content using AI
Export it as a fully formatted .docx file

Initially, I assumed I needed:

A backend server
File storage

But I challenged that assumption.

Frontend-Only Architecture

Instead of using a backend, I built a client-side workflow:

The browser sends a request to an AI API
The AI returns structured content
A JavaScript library converts it into a .docx file
The file is downloaded directly

Architecture Flow

User Input → AI API → Structured Data → Docx Generator → Download

Benefits of a No-Backend Approach

Zero Infrastructure Cost-No servers, no hosting, no maintenance.
Better User Privacy-No data stored on servers Everything processed locally

3.Simpler Deployment-Just deploy a static site No backend configuration

4.Faster Development-Fewer moving parts No API layer to maintain

When You Still Need a Backend

A frontend-only approach doesn’t work if you need:

User authentication
Persistent data storage
Complex business logic
Rate limiting or access control

Key Insight

Many applications are over-engineered by default.

Before building a backend, ask:

“Can this be done entirely in the browser?”

Modern browsers are more powerful than we think.

Real Example

I built a working version of this idea as a browser-based tool that generates formatted Word documents from plain-English prompts:

👉 docreplacer.online

Why I Stopped Building a Backend (And What Happened Next)

Manickavasagan — Sun, 10 May 2026 08:17:41 +0000

docreplacer.online is a tool that converts a plain-English prompt into a formatted .docx file — entirely inside your browser. No server, nothing uploaded or stored remotely. You type a prompt, a Word document downloads. It's currently in MVP — free to use, no account required at this stage.
I built it. And the most interesting part of building it wasn't the document generation itself — it was the decision to use no backend at all, and what that choice forced me to think through.
I'm a developer who builds small tools to solve workflow problems. This one started as a minor utility for my own use and turned into a more considered architectural exercise than I expected. This post is about that process — what the problem was, how the client-side approach works in practice, and where it genuinely falls short.

Most side projects start with a problem. This one started with a boring observation: generating a .docx file from structured text is something browsers can already do — and almost nobody does it that way.
I wasn't trying to build a product. I was trying to avoid spinning up a server for something that didn't need one.

The Actual Problem

I was generating documents repeatedly — contracts, briefs, structured reports — where the content changed but the format didn't. The standard workflow: keep a Word template, open it, fill in the blanks, save-as, send. Manageable for one document. Tedious for twenty.
The obvious modern fix is AI. Feed a prompt, get content back. But AI gives you text. Turning that text into a properly formatted .docx — with correct heading levels, paragraph spacing, and section structure — still requires a human hand in Word.
That gap is small, but it compounds. And the tooling built to close it almost universally requires an account, a subscription, or a server that holds your files.
I didn't want any of that. So I looked at whether it was actually necessary.
The tools that exist to bridge this gap — AI writing assistants, document editors with template engines, cloud-based DOCX converters — all solve it server-side. Your prompt goes up, the file comes down, and somewhere in the middle your content passes through infrastructure you don't control. For most use cases that's a reasonable trade-off. For documents containing personal information, legal agreements, or internal business data, it's worth at least pausing on.

What the Browser Can Do

The docx JavaScript library constructs valid Office Open XML documents entirely in memory. No server. No conversion endpoint. The file is assembled in the browser and downloaded directly — the same way a client-side CSV export works, just more structurally complex.
This meant the architecture could be:
• Browser receives prompt
• AI API returns structured content
• docx library assembles the file client-side
• File downloads
No file ever touches a server I control. No authentication layer needed. No storage costs. No database.
The result is docreplacer.online — a tool that converts a plain-English prompt into a .docx file, runs entirely in the browser, and is free to use.

What "Client-Side Only" Actually Means in Practice
It means the tool has genuine constraints, and being honest about them is more useful than pretending otherwise.
What works well:
Document generation is stateless by nature. You have a prompt, you want a file. There's no meaningful user state to persist server-side, no collaboration requirement, no reason the process needs to leave the browser. Client-side architecture fits the problem well.
localStorage handles recent templates. The browser handles the download. The only external dependency is the AI inference call — which originates from the client and returns content, not a stored file.

Where it gets harder:

Large-scale batch generation isn't the right use case here. If you need to produce 500 documents from a data pipeline, you want a backend. This tool is for individuals generating documents on demand.
Template sharing across a team isn't supported in the current version. That would require either a sync layer or a server — which may be added later, but deliberately isn't there now.
These are real trade-offs, not oversights. The goal was to build the simplest thing that solved the problem without introducing infrastructure the problem doesn't require.

The Broader Point About Architecture Decisions
There's a habit in software development — understandable, given the tooling available — of reaching for a backend before asking whether one is necessary.
Servers solve real problems: persistence, authentication, multi-user access, compute-intensive operations. But they also introduce real costs: hosting, maintenance, security surface area, latency, and for users, the ongoing question of what happens to their data.
For a certain class of tools — document generation, data formatting, local calculations, file conversion — the browser is sufficient. WebAssembly, modern JS libraries, and browser storage APIs have expanded what's possible client-side considerably in the past five years.
The question worth asking before spinning up a server is: what does the server do here that the client can't? Sometimes the answer is obvious. Sometimes it isn't, and the server exists out of habit.
This isn't an argument for never using backends. It's an argument for making the choice deliberately rather than by default.

Where This Is Going

The current version is an MVP. It handles single-document generation from a prompt, with basic formatting and section structure.
Planned additions include template saving, more granular formatting control, and better handling of complex document structures (tables, multi-column layouts). Whether any of those features require server infrastructure will be evaluated when the time comes — not assumed in advance.
If you work in an environment where documents contain sensitive information — legal, medical, HR, financial — the client-side architecture is worth understanding. Nothing is transmitted to a storage layer. The document exists in your browser's memory and then on your local file system. That's a meaningful property for certain workflows, and one that's difficult to replicate credibly with a server-based tool.

Try It

docreplacer.online is currently free to use — a free tier will remain available in the full version.
A few prompts worth testing if you want to see the range:
• A consulting services agreement with payment terms and IP ownership clauses
• A technical specification document for a REST API
• A performance review template for quarterly engineering team reviews
One thing worth noting: the output is a standard .docx file. It opens in Microsoft Word, Google Docs, LibreOffice — anywhere that reads the Office Open XML format. There's no proprietary format, no export step, no compatibility issue. It's just a Word document, generated from a prompt.
Feedback on document quality, formatting gaps, or missing features is genuinely useful at this stage — the tool is early and the edge cases are being found in real use.

The author built docreplacer.online as a client-side document generation tool. It is currently in MVP — free to use, with a free tier planned for the full release.

I was losing sleep over broken Word formatting. So I built a fix.

Manickavasagan — Thu, 30 Apr 2026 13:12:21 +0000

I'm a college student. Every semester = 3-4 major projects.

Not the coding. Not the research. The documentation killed me.

Every time I used AI to draft content, the copy-paste into Word was
a disaster. Fonts breaking mid-sentence. Tables collapsing. Weird
indentation. I was spending hours just fixing margins and headers
before submitting anything.

It got bad enough that I started having actual nightmares about
formatting deadlines. I still have 5 semesters left. I couldn't
keep doing this.

What I originally wanted to build

A simple tool: paste AI content → get a clean .docx without broken layout.

No reformatting. No manual fixes. Just a file that opens correctly in Word.

What it became

While building it, I realized the real problem was earlier in the chain.
Why copy from AI at all? Why not just prompt directly into the .docx?

So I rebuilt it. Now DocReplacer lets you:

Enter a prompt
Get a structured .docx with headers, tables, bullet points
Download it immediately

No login. No server. Runs entirely in your browser.
Close the tab — session gone. Nothing stored anywhere.

The part I care most about

It's free. No paywall. No email required.

I built this to solve my own problem. If it saves one other student
from a 2am formatting spiral, that's enough.

Try it

👉 https://www.docreplacer.online

Honest feedback welcome — especially on formatting edge cases.
Still fixing: PDF export, document length limits, feedback form bug.

What's the most painful part of your documentation workflow?