How tldr.work Works

tldr.work combines cutting-edge AI technology with powerful APIs to automate sales and monitor brand visibility. Here’s how everything works under the hood.

Architecture Overview

Sales Automation Workflow

1. Prospect Discovery (our discovery engine)

What happens when you search for prospects:

You enter search criteria

Example: “French bakeries in Paris”

tldr.work queries our discovery engine API

our discovery engine is a powerful web scraping tool that crawls the web intelligently.

our discovery engine finds matching companies

Returns company name, website, industry, location, and description.

Results appear in your dashboard

Typically takes 20-40 seconds for 10-50 results.

Why our discovery engine? It’s specifically designed for business data extraction and returns structured, clean data.

2. Email Enrichment (our enrichment engine)

What happens when you enrich prospect data:

tldr.work sends company domain to our enrichment engine

Example: bakery-paris.fr

our enrichment engine finds all email addresses

Searches public sources: company website, LinkedIn, directories.

our enrichment engine verifies each email

Checks if email exists and is deliverable (MX records, SMTP validation).

You get verified contacts

Name, role, email, LinkedIn profile, and verification score (0-100).

Emails with verification score > 90 have 98%+ deliverability rate.

3. Email Generation (GPT-5)

What happens when you generate an email:

You provide context

Campaign goal, tone (professional/casual), length (short/medium/long).

tldr.work prepares the prompt

Includes prospect info (company, industry, recipient name/role).

GPT-5 generates personalized email

Creates subject line + body tailored to the prospect.

You review and edit

You can regenerate or manually tweak before sending.

Example Prompt to GPT-5:

You are an expert sales copywriter. Write a professional cold email.

Context:
- Recipient: Marie Dupont, Owner at La Pâtisserie Parisienne
- Industry: French bakery
- Location: Paris, France
- Goal: Offer marketing services for local bakeries
- Tone: Professional but warm
- Length: Short (150 words)

Generate:
1. Subject line (under 50 characters)
2. Email body (personalized, value-focused)

GPT-5 uses the prospect’s info to personalize each email. Never use generic templates!

4. Campaign Execution (our delivery infrastructure)

What happens when you launch a campaign:

tldr.work queues emails

Adds all recipients to a background job queue (Redis + BullMQ).

Respects daily limits

Default: 50 emails/day (prevents spam flags). Customizable per campaign.

Sends via our delivery infrastructure API

our delivery infrastructure handles SMTP, SPF, DKIM, bounce handling.

Tracks engagement

Opens, clicks, and replies are logged automatically.

Why daily limits? Sending 1000 emails at once triggers spam filters. Gradual sending (50/day) maintains sender reputation.

AI Brand Monitoring Workflow

1. Domain Analysis (GPT-5-mini)

What happens when you analyze a domain:

tldr.work scrapes your website

Uses our discovery engine to extract main content (text, metadata, structure).

Sends content to GPT-5-mini

Asks AI to extract business information.

GPT-5-mini analyzes and structures data

Returns: company name, industry, description, services, keywords, competitors.

Results saved to your project

Used for prompt generation in the next step.

Example Analysis Output:

{
  "companyName": "La Pâtisserie Parisienne",
  "sector": "Food & Beverage (Bakery)",
  "description": "Traditional French bakery offering artisan bread, pastries, and cakes",
  "location": "Paris, France",
  "services": ["Bread baking", "Custom cakes", "Catering"],
  "keywords": ["French bakery", "artisan bread", "Paris pastries"],
  "suggestedCompetitors": ["Poilâne", "Du Pain et des Idées", "Liberté"]
}

Why GPT-5-mini? It’s 10x faster and cheaper than GPT-5 for simple extraction tasks, with nearly the same accuracy.

2. Prompt Generation (GPT-5-mini)

What happens when you generate prompts:

tldr.work sends domain analysis to GPT-5-mini

Includes company info, industry, services, competitors.

Asks AI to create 10 realistic prompts

Divided into categories: Brand, Product, Competitor, Intent.

GPT-5-mini generates prompts

Mimics real user queries (e.g., “Best bakery in Paris for birthday cakes”).

Prompts validated automatically

“Intent” prompts must NOT contain brand name (tests organic discovery).

Prompt Categories Explained:

Category	Description	Example	Contains Brand?
Brand	Direct questions about your company	”What is La Pâtisserie Parisienne?”	✅ Yes
Product	Questions about your offerings	”What does La Pâtisserie Parisienne sell?”	✅ Yes
Competitor	Comparisons or “best in category"	"Best bakeries in Paris”	❌ No
Intent	Discovery queries (HIGH VALUE)	“Where to buy artisan bread in Paris?”	❌ No

Intent prompts are the most important! They test if AI recommends you when users don’t know you exist yet.

3. Multi-Engine Testing (our AI gateway)

What happens when you run an analysis:

tldr.work selects active prompts

You can enable/disable any of the 10 generated prompts.

Sends each prompt to 5 AI engines

Via our AI gateway API (unified gateway for all LLMs).

Each engine generates a response

ChatGPT (GPT-5)
Claude (3.5 Sonnet)
Gemini (2.5 Flash)
Perplexity (Sonar Pro)
Mistral (Large)

tldr.work analyzes each response

Checks: Is your brand mentioned? Where does it rank? What’s the sentiment?

Example Prompt Flow:

Prompt: "Best bakery in Paris for wedding cakes"

→ Sent to 5 engines in parallel
→ ChatGPT response: "Top 3 bakeries: 1. Ladurée 2. Pierre Hermé 3. La Pâtisserie Parisienne"
→ Claude response: "I'd recommend Stohrer, the oldest bakery in Paris..."
→ Gemini response: "For wedding cakes, consider La Pâtisserie Parisienne or Angelina"
→ Perplexity response: [lists 8 bakeries, yours is #6]
→ Mistral response: [doesn't mention your brand]

→ tldr.work calculates:
   - Visibility: 60% (mentioned in 3/5 engines)
   - Avg Position: #3.3 (avg of ranks: 3, not mentioned, 1, 6, not mentioned)
   - Sentiment: Positive

Why our AI gateway? It provides access to all major AI models via one API, with consistent formatting and billing.

4. Response Analysis (GPT-5)

What happens when analyzing AI responses:

tldr.work sends response to GPT-5

Asks: “Is [Your Brand] mentioned? Where? What’s the sentiment?”

GPT-5 extracts structured data

Returns: isMentioned (yes/no), position (#1-10), sentiment (positive/neutral/negative), competitors mentioned.

Aggregates across all prompts

Calculates average visibility, position, and sentiment.

Results displayed in dashboard

Visual charts + detailed breakdown per prompt/engine.

Why GPT-5 for analysis? It’s the best at understanding context and extracting nuanced information from text.

Technology Stack

Frontend

Next.js 15: React framework for fast, SEO-friendly pages
Tailwind CSS: Modern, responsive UI design
TanStack Query: Real-time data fetching and caching

Backend

Fastify: High-performance Node.js server
MongoDB: NoSQL database for flexible data storage
Redis + BullMQ: Background job processing (emails, AI analysis)

AI & APIs

our AI gateway: Unified gateway to all AI models (GPT, Claude, Gemini, etc.)
our discovery engine: Web scraping and company data extraction
our enrichment engine: Email finding and verification
our delivery infrastructure: Transactional email delivery

Security & Billing

JWT: Secure authentication tokens
Stripe: Payment processing and subscription management
Encrypted storage: Sensitive credentials encrypted at rest

Performance & Reliability

Speed

Prospect search: 20-40 seconds for 10-50 results
Email enrichment: 2-5 seconds per prospect
Email generation: 3-5 seconds per email
AI analysis: 5-10 minutes for 10 prompts × 5 engines (50 queries)

Accuracy

Prospect data: 95%+ accuracy (our discovery engine filters out invalid data)
Email verification: 98%+ deliverability (our enrichment engine’s verification algorithm)
AI generation: GPT-5 produces human-quality text in 99% of cases

Reliability

Uptime: 99.9% (hosted on Railway with auto-scaling)
Failover: If one AI engine fails, others continue
Queue system: Jobs retry automatically on failure

Next Steps

Try Sales Automation

Learn how to find prospects and launch campaigns

Try AI Monitoring

Track your brand across AI engines

View Key Features

Explore all capabilities in detail

Get Started

Set up your account in 5 minutes

Questions? Email support@tldr.work

Overview

Introduction

Getting Started

Sales Automation

AI Search Analytics

Billing & Plans

How It Works

How tldr.work Works

Architecture Overview

Sales Automation Workflow

1. Prospect Discovery (our discovery engine)

2. Email Enrichment (our enrichment engine)

3. Email Generation (GPT-5)

4. Campaign Execution (our delivery infrastructure)

AI Brand Monitoring Workflow

1. Domain Analysis (GPT-5-mini)

2. Prompt Generation (GPT-5-mini)

3. Multi-Engine Testing (our AI gateway)

4. Response Analysis (GPT-5)

Technology Stack

Frontend

Backend

AI & APIs

Security & Billing

Performance & Reliability

Speed

Accuracy

Reliability

Next Steps

Try Sales Automation

Try AI Monitoring

View Key Features

Get Started

​How tldr.work Works

​Architecture Overview

​Sales Automation Workflow

​1. Prospect Discovery (our discovery engine)

​2. Email Enrichment (our enrichment engine)

​3. Email Generation (GPT-5)

​4. Campaign Execution (our delivery infrastructure)

​AI Brand Monitoring Workflow

​1. Domain Analysis (GPT-5-mini)

​2. Prompt Generation (GPT-5-mini)

​3. Multi-Engine Testing (our AI gateway)

​4. Response Analysis (GPT-5)

​Technology Stack

​Frontend

​Backend

​AI & APIs

​Security & Billing

​Performance & Reliability

​Speed

​Accuracy

​Reliability

​Next Steps

Try Sales Automation

Try AI Monitoring

View Key Features

Get Started

How tldr.work Works

Architecture Overview

Sales Automation Workflow

1. Prospect Discovery (our discovery engine)

2. Email Enrichment (our enrichment engine)

3. Email Generation (GPT-5)

4. Campaign Execution (our delivery infrastructure)

AI Brand Monitoring Workflow

1. Domain Analysis (GPT-5-mini)

2. Prompt Generation (GPT-5-mini)

3. Multi-Engine Testing (our AI gateway)

4. Response Analysis (GPT-5)

Technology Stack

Frontend

Backend

AI & APIs

Security & Billing

Performance & Reliability

Speed

Accuracy

Reliability

Next Steps