Forem: sher1096

I Built 6 AI Tools in 30 Days - Here's What I Learned

sher1096 — Sat, 14 Mar 2026 18:46:57 +0000

Last month I challenged myself: build and launch as many useful AI tools as possible. The goal wasn't perfection - it was speed and learning.

Here's what I shipped:

The Tools

1. CleanCSV AI

Problem: Messy CSV files with duplicates, mixed dates, empty cells.
Solution: Upload → AI diagnoses issues → One-click fix → Download clean file.
Stack: Next.js + deterministic cleanup logic (no AI guessing for simple fixes)

2. BriefPDF AI

Problem: Too many PDFs to read.
Solution: Upload any PDF → Get executive summary + key points.
Stack: Next.js + Claude for analysis

3. InvoiceQuill

Problem: Creating invoices is tedious.
Solution: Describe in plain English → AI generates invoice → Live PDF preview.
Stack: Next.js + OpenAI for NLU

4. LegalDocs AI

Problem: Lawyers charge $500 for template documents.
Solution: Answer questions → AI generates Privacy Policy / Terms / NDA.
Stack: Next.js + OpenAI + Creem payments

5. DepthPilot AI

Problem: Learning AI/ML is overwhelming.
Solution: Upload papers → AI extracts key concepts → Track learning progress.
Stack: Next.js + Claude for analysis

6. 16Traders

Problem: Most traders don't know their trading personality.
Solution: 3-minute test → Get your trader archetype (like MBTI for traders).
Stack: Next.js + custom scoring algorithm

What I Learned

1. Speed beats perfection

MVP first, iterate later. None of these are perfect, but they're live and getting feedback.

2. The AI is the easy part

The hard part is everything else: deployment, payments, SEO, marketing, support.

3. Self-hosting is underrated

All 6 tools run on a single $10/month VPS. No Vercel bills, no Lambda surprises.

4. Creem > Stripe for MVPs

Creem handles merchant of record, taxes, compliance. One integration, global payments.

5. Distribution is 80% of the work

Building is fun. Getting users is hard. SEO takes months. Twitter is noisy.

The Numbers (After 30 Days)

Total revenue: $0 (yes, zero)
Total users: ~200 (mostly from Dev.to)
Biggest traffic source: Dev.to articles
Biggest time sink: Payment integration

What's Next

The challenge isn't building more tools. It's getting these 6 to product-market fit.

Current focus:

More content marketing (like this post)
Google indexing (submitted to Search Console)
Finding communities where users hang out

If you're building in public too, I'd love to connect. What's been your biggest challenge shipping fast?

How I Built a Legal Document Generator That Costs $0 to Run

sher1096 — Sat, 14 Mar 2026 18:35:21 +0000

Last month a friend asked me to help with his startup's privacy policy. He'd been quoted $500 by a lawyer for a "template review."

For a privacy policy. That's basically the same for every SaaS.

So I built LegalDocs AI - a legal document generator that creates business-specific policies in 3 minutes.

The Problem

💰 Lawyers charge $300-1000 for template documents
📋 Generic templates don't cover your specific stack
⏰ Waiting days for a simple privacy policy
🔄 Updating docs when you add new features

What It Generates

Privacy Policy - GDPR, CCPA, global compliance
Terms of Service - Platform-specific terms
NDA Templates - Mutual, one-way, project-specific

How It Works

Answer 10-12 questions about your business
AI generates a draft based on your answers
Preview the document live
Pay $2.99 for the full PDF download

The magic is in the questions - they're designed to extract exactly what a privacy policy needs to cover:

What data you collect
Third-party services you use
Jurisdictions you operate in
How long you retain data

The Tech

Next.js 14 + App Router
OpenAI for generation
Creem for payments
Self-hosted on a $10/month VPS

Total monthly cost: ~$15 (VPS + minimal API calls)

Try It

Free preview available - you only pay if you want the full document.

👉 docs-legal.com

What's the most you've paid for a template legal document? I'd love to hear the horror stories!

Too Many PDFs? I Built an AI That Summarizes Them in Seconds

sher1096 — Sat, 14 Mar 2026 13:59:02 +0000

My reading list was out of control. Research papers, whitepapers, reports, documentation - all piling up as "I'll read this later" PDFs.

So I built BriefPDF AI to get the key points without reading 50+ pages.

The Problem

📚 PDFs pile up faster than you can read them
⏰ A 30-page report takes 1-2 hours to properly digest
🔍 Often you just need the key takeaways
📝 Manual note-taking is tedious

What BriefPDF Does

Upload any PDF. Get back:

✅ Executive summary - the TL;DR in 2-3 paragraphs
✅ Key points - bullet list of main takeaways
✅ Section breakdown - what each part covers
✅ Action items - if any are mentioned

Use Cases

📊 Business reports → 5-minute summary
📄 Research papers → key findings extracted
📋 Legal docs → main clauses highlighted
📖 Technical docs → core concepts identified

The Tech

Built with Next.js + Claude for analysis. The challenge was handling various PDF structures - some have clear headings, others are walls of text. The AI adapts to both.

Try It

Free tier: 5 PDFs/month, up to 50 pages each.

👉 getbriefpdf.com

How do you handle PDF overload? Any specific document types you'd want better summarization for?

I Built an AI Invoice Generator That Actually Understands Plain English

sher1096 — Sat, 14 Mar 2026 13:58:07 +0000

As a freelancer, I used to waste 20+ minutes on every invoice. Finding templates, filling fields, double-checking numbers, exporting to PDF...

So I built InvoiceQuill AI - describe your work in plain English, get a professional invoice instantly.

How It Works

Just type something like:

"Invoice for John at Acme Corp, 10 hours of React development at $100/hr, due in 30 days"

The AI understands:

✅ Client name and company
✅ Line items and quantities
✅ Hourly rates or fixed prices
✅ Due dates and payment terms

Features

📝 Natural language input - no forms to fill
👀 Live PDF preview - see changes instantly
🎨 Multiple templates - professional designs
💾 One-click download - PDF ready to send
📧 Email integration - send directly (coming soon)

The Tech

Built with Next.js + OpenAI for NLU. The challenge was handling the variety of ways people describe invoices - "10 hrs @ 100", "ten hours at a hundred bucks", "$1000 for the project" all need to work.

Try It

Free tier: 5 invoices/month. No signup required to try.

👉 invoicequill.com

How do you currently handle invoicing? Would love to hear what features matter most to freelancers!

Stop Copy-Pasting Tables from PDFs - I Built a Better Way

sher1096 — Sat, 14 Mar 2026 12:24:16 +0000

We've all been there: you need data from a PDF table, but copy-paste gives you a mangled mess. Columns merge, numbers break, and you spend 30 minutes fixing what should take 30 seconds.

I built TableSnap AI to solve this problem once and for all.

The Pain

📄 PDF tables don't copy cleanly
🔢 Numbers get split across lines
📊 Column alignment breaks completely
⏰ Manual cleanup wastes hours

What TableSnap Does

Upload any PDF with tables. Get back:

✅ Clean Excel (.xlsx) files
✅ Properly formatted CSV
✅ Structured JSON for APIs
✅ Clean Markdown tables

How It Works

Upload your PDF
AI detects and extracts tables
Preview the structured data
Download in your preferred format

No more manual cell-by-cell copying. No more broken formatting.

Use Cases

Financial reports → Excel
Research data → CSV for analysis
API documentation → JSON
Meeting notes → Markdown tables

Try It

Free tier available for files up to 10 pages.

👉 tablesnap.ai

What's your current workflow for extracting PDF tables? I'd love to hear about edge cases!

How I Built an AI Learning Companion for Self-Taught Developers

sher1096 — Sat, 14 Mar 2026 12:23:30 +0000

Learning AI/ML as a self-taught developer is overwhelming. Papers are dense, courses are scattered, and it's hard to track what you actually understand vs what you just skimmed.

I built DepthPilot AI to solve this for myself - and now I'm sharing it.

The Problem

📚 Too many resources, no structure
🤯 Papers that assume you know everything already
❓ Hard to identify gaps in your knowledge
📊 No way to measure real progress

What DepthPilot Does

Summarize dense PDFs and papers into digestible chunks
Extract key concepts and prerequisites
Track what you've learned over time
Identify knowledge gaps based on your reading history

How It Works

Upload a paper or paste a URL. The AI breaks it down into:

TL;DR summary
Key concepts (with explanations)
Prerequisites you might be missing
Related topics to explore next

Built For

Self-taught developers learning ML/AI
Researchers keeping up with papers
Anyone drowning in technical PDFs

Try It

Free tier available - start organizing your learning today.

👉 depthpilotai.com

What's your biggest challenge when self-learning technical topics? I'd love to hear how others approach this!

I Built a CSV Cleanup Tool That Fixes Messy Data in Seconds

sher1096 — Sat, 14 Mar 2026 12:07:22 +0000

If you've ever worked with data from multiple sources, you know the pain of messy CSV files:

Duplicate rows that mess up your reports
Empty cells that break your imports
Date formats all over the place (2025-01-04, 01/04/2025, January 4...)
Hidden whitespace that causes silent matching failures

I got tired of writing one-off scripts every time, so I built CleanCSV AI - a tool that diagnoses and fixes these issues automatically.

How It Works

Upload your messy CSV/Excel file
Review the diagnostic report (duplicates, blanks, date drift, etc.)
Choose which fixes to apply
Download the cleaned file with a full change log

What It Catches

✅ Duplicate rows
✅ Empty/missing values
✅ Inconsistent date formats
✅ Leading/trailing whitespace
✅ Mixed data types in columns
✅ Outlier values

The Stack

Built with Next.js 14, deployed on a VPS. The core cleanup logic is deterministic (no AI guessing for simple fixes), which means results are repeatable.

Try It

Free tier available - files up to 1MB, 5 analyses per day.

👉 cleancsvai.com

What's your go-to method for cleaning messy data? Would love to hear about edge cases I should handle!

The Personal Production Function in the AI Era: P = D A C

sher1096 — Tue, 10 Mar 2026 11:48:06 +0000

I've been thinking about a question lately:

What actually determines someone's output when they start using AI?

Many people assume AI directly boosts productivity. But reality is more nuanced.

Some people use AI and see massive improvements—faster execution, faster project delivery, expanded capabilities.

Others use AI daily, ask many questions, generate lots of content, look busy... but create nothing truly valuable.

So the question isn't "Can AI boost productivity?"

It's: What determines personal output in the AI era?

The Formula

I've tried to capture this in a rough formula:

P = D × A × C

Where:

P = Production (personal output)
D = Depth (domain expertise)
A = AI Ability (how well you use AI)
C = Compute (models, tools, resources)

Why Multiplication, Not Addition

The key insight is that these variables multiply, not add.

They don't substitute each other—they constrain each other:

Strong depth but can't use AI? Your abilities stay locked.
Great at AI but no domain expertise? You get plausible-sounding nonsense.
Both skills but no resources? Complex tasks won't even start.

One weak variable drags down the entire result.

Depth Determines Your Ceiling

If I had to pick the most important variable, it's Depth.

AI is fundamentally an amplifier. It amplifies what you already have.

A trader using AI to write strategies can:

Organize thoughts faster
Gather information faster
Test ideas faster
Validate assumptions faster

A novice gets plausible-looking code they can't evaluate.

Answers can be generated. Judgment cannot be outsourced.

AI Ability Isn't Just Prompting

Real AI ability includes:

Decomposing problems
Structuring tasks
Iterating with AI
Verifying outputs
Constraining AI (this is crucial!)

Without constraints, AI generates fluent-sounding fluff. The skill is organizing AI into your workflow, not just asking questions.

Resources Define Your Feasibility Space

Resources aren't just tokens. They include model capability, toolchain, automation systems.

Think of it like electricity—without enough power, machines don't run slow, they don't run at all.

AI Amplifies Gaps

Here's the uncomfortable truth:

AI won't close gaps between people. It will amplify them.

This is a multiplicative structure:

Those with depth + AI skills + resources create outsized value
Those without get left further behind

The "super-individual" era is here.

Bottom Line

AI-era output = Depth × AI Ability × Resources

The real scarcity isn't answers—it's:

Asking the right questions
Detecting flaws in AI output
Deep thinking
Judgment

Build what AI can't replace.

Built a 3‑minute trading personality test (MBTI for traders)

sher1096 — Mon, 09 Mar 2026 18:49:28 +0000

Been trading for a few years and noticed how much psychology affects decisions. I built a quick test to identify trading personality types.

Think MBTI for traders — helps reflect on:

Risk tolerance patterns
Decision‑making speed
How you handle drawdowns

Takes ~3 minutes. Free. No signup.

Link: https://16traders.com

Curious what results others get and if it resonates.

Built an AI-guided legal docs generator (Privacy Policy, Terms, NDA)

sher1096 — Mon, 09 Mar 2026 18:29:08 +0000

Launching a side project and needed legal docs. Lawyers quoted $500+ for boilerplate.

So I built LegalDocs.ai.

What it does

Guided questionnaire (not just fill‑in‑blank)
AI adapts language to your business type
Generates Privacy Policy, Terms of Service, NDA templates
GDPR/CCPA options

Free tier available. Not legal advice, but way better than copy‑pasting random templates.

Link: https://docs-legal.com

If you’ve launched a product recently, how did you handle legal docs?

Built a plain-English invoice generator with AI + live PDF preview

sher1096 — Mon, 09 Mar 2026 18:28:15 +0000

I was paying $15/mo for invoicing software I used twice a month. So I built InvoiceQuill.

What it does

Describe your invoice in plain English → AI fills the fields
Live PDF preview
No signup required for first invoices
Supports commercial invoices, payment requests, proformas

Why it helps

Freelancers and small businesses just need quick, clean invoices — not a bloated CRM.

Free 3 PDFs/month. $9/mo for unlimited.

Link: https://invoicequill.com

Would love feedback from freelancers doing cross‑border work.

Made a tool that fixes messy CSVs with AI - dates, duplicates, formatting

sher1096 — Mon, 09 Mar 2026 17:44:50 +0000

Working with client data exports is painful. Different date formats, duplicate rows, inconsistent text cases, missing values...

Built CleanCSV to handle this automatically.

What it does

🔍 Detects column types and suggests appropriate fixes
📅 Standardizes dates - handles MM/DD/YYYY, DD-MM-YYYY, ISO, and more
🔢 Cleans numbers - removes currency symbols, fixes decimal separators
🚮 Removes duplicates - with smart matching
🤖 AI suggestions - recommends transformations based on your data

How it works

Upload your messy CSV
Preview the detected issues
Apply fixes (or let AI suggest them)
Download clean file

Why I built this

Every time I got a data export from a client or API, I'd spend 30+ minutes cleaning it in Excel or writing a Python script. Figured there had to be a better way.

Free tier available - handles most use cases.

Link: https://cleancsvai.com

What's the worst data mess you've had to clean up?