WeExpand

WeExpand Market Intelligence

v0.3

๐ŸŽฏ Step 1: Industry & Location

For targeted discovery & research context

๐Ÿ“‹ Step 2: Companies

CSV, JSON, or exported results_*.json
Upload results_C.json or results_D.json to skip phases you already completed. Required when start phase โ‰ฅ C.

๐Ÿ“Œ Step 3: Phases

๐Ÿ“„ Step 4: Solution Docs

โ–ถ๏ธ Step 5: Launch

Loading...
Loading...

๐Ÿ”‘ API Keys

DeepSeek
OpenAI

๐Ÿ”ง Custom Models (OpenAI-compatible)

Add your own OpenAI-compatible API endpoints. Each model needs a name, API key, base URL, and model name.

๐Ÿง  LLM Task Routing

Choose which provider handles each task. Deterministic tasks default to local (Ollama), reasoning tasks to cloud.

TaskDescriptionProvider

๐Ÿค– Multi-Model Pain Generation

Select which models to use for Phase C pain point generation. At least 1 required.

โš™๏ธ Pipeline Params

Crawl Workers
Parallel page crawlers
Crawl Timeout (s)
Per-page max time
Max Pages/Company
Pages crawled per target
Model selection: use ๐Ÿง  LLM Task Routing above to assign each task to a provider.

โšก Performance

Parallel Companies
Process N companies simultaneously
Parallel Models (Phase C)
Run 3 AI models concurrently
Batch Delay (seconds)
Pause between company batches
Batch Evidence (Phase C)
Validate all pains in one LLM call
Multi-Model Pain Gen
Use 3 AI models for diverse pain points (Phase C1)
Batch Doc Ingestion (Phase A)
Process all solution docs in one LLM call (faster, less detailed)
Batch Narrative (Phase D)
Generate all pain narratives in one LLM call (faster, more coherent)
Prompt Caching (OpenRouter)
Cache system prompts across calls (-70% input tokens)
Hybrid Critic (Phase C)
Batch first, iterate only low-confidence pains
Critic Threshold
Re-iterate pains below this confidence
Parallel Crawl (Phase B)
Fetch pages simultaneously via async HTTP
Crawl Concurrency
Max simultaneous page fetches
Parallel Report Gen
Render pages + LLM narratives in parallel
LLM Max Concurrent
Max parallel LLM calls during report gen

๐Ÿ”ฌ Pain Analysis

Confidence Threshold
Min confidence to accept pain point
Min Evidence Sources
Sources required per pain point
Min Evidence Categories
Unique source types required
Critic Max Iterations
Max review rounds per pain point (1-10)
Batch Critic
Evaluate all pains in one LLM call (faster, -30% cost)

๐Ÿ“Š Sell Scoring Weights

Weights determine how each dimension contributes to the final sell score. Must sum to 1.0.

Pain Fit
Solution alignment with identified problems
Budget Probability
Likelihood of allocated budget
Urgency
How soon they need to act
Strategic Alignment
Fit with company direction
Digital Readiness
Technical capability to adopt
Survey Validation
Stakeholder survey agreement boost

๐Ÿ“ง Email Discovery

Enabled
Auto-discover emails during crawl (Phase B6.5)
Max Key Persons
Top executives to discover per company

๐Ÿ“‹ Survey

Max Questions
AI-generated survey questions (Phase E)
Min Questions
Below this, skip survey generation

๐Ÿ”ง Advanced Settings

Confidence Weights
Embeddings
Vector Store
Manual Gates
Link Classification
Audit Logs
System

๐Ÿ‘ค My Personal Configuration

Override shared defaults with your own API keys, model routing, and performance settings. Changes apply to your jobs only.

๐Ÿญ Industry Configs

Manage pain categories, segments and discovery prompts.

๐Ÿ“ Phase E: Survey Questions

Generate AI-suggested survey questions based on completed analysis. Select a completed job first.

๐Ÿ“ฅ Phase F: Import Survey Responses

Upload CSV with columns: respondent_id, company_name, question, answer

๐Ÿ“ฌ View Pipeline Email Results

Pipeline Phase B6.5 automatically discovers emails during crawling. Select a completed job to view them instantly โ€” no new search needed.

๐Ÿ”Ž Run New Email Search

Search for emails by company name. Uses LLM + web search โ€” slower but works without a pipeline job. Good for quick lookups.

Columns: Company Name, Website

๐Ÿš€ Quick Start

1. Pick industry + region โ†’ 2. Add companies (paste JSON or use Phase 0: Auto-Discover) โ†’ 3. Upload solution docs (optional) โ†’ 4. Select phases (A-D for full research) โ†’ 5. Click โ–ถ Start Pipeline.

๐Ÿ” Research (Phase A-D)

A=Solution Understanding, B=Web Crawling + Email Discovery, C=Pain Point Analysis, D=Sell Scoring + Report. After completion, click ๐Ÿ“Š v1/v2 to view the report. All phases support resume from previous results.

๐Ÿ“ง Email Discovery

Phase B6.5 automatically finds emails during crawling (key persons + general emails). View results in the ๐Ÿ“ง Email tab โ€” just select a completed job and click ๐Ÿ“ฌ View Results. Or run a new search for any company list using LLM + web search.

๐Ÿ“Š Reports

Click ๐Ÿ“Š v1/v2 on a completed job. Reports include: Dashboard, Company detail pages (pain points, charts, evidence, sell scores), Summary, Industry overview, Methodology. v2 adds survey results page. All pages support dark mode ๐ŸŒ“.

๐Ÿ“‹ Survey (Phase E-F)

Switch to ๐Ÿ“‹ Survey tab. Select a completed job โ†’ Generate Questions for AI survey design. Upload CSV responses โ†’ Recalibrate to update sell scores with real stakeholder data. View survey_results.html in v2 report.

โšก Phase 0: Auto-Discover

Finds companies using AI + web search. Select model (Cloud/OpenAI/Local/Custom). Enable LinkedIn search for extra sources. Use ๐Ÿค– Gen+Discover to generate industry config and discover companies in one step.

๐Ÿ”„ Resume & Import

Click ๐Ÿ”„ Resume on any job to restore its full configuration and continue. Upload previous results_C.json or results_D.json to jump-start Phase C or D with pre-computed data.

โš™๏ธ Settings

Configure API keys (Cloud + OpenAI), custom models, LLM task routing, and industry configs. ๐Ÿค– AI Generate creates industry configs via LLM. ๐Ÿ’พ Save custom configs. Use ๐Ÿ‘ Show Hidden to restore hidden defaults.

๐Ÿ”ง Admin Panel

Manage user accounts, passwords, and roles.

Loading...

๐Ÿ‘ค User Limits

Set company and job limits per user. Admin is unlimited.

Loading...
WeExpand

Market Intelligence

Sign in to continue
Invalid username or password
WeExpand Limited | All Rights Reserved