Diego Sanchez

Competitive
Intelligence Report

Research Tool 2026

The Problem

Most CI reports tell you what competitors do. That's not the question.

The useful question is: what can't they fix? A weakness only matters if it's structural. Baked into architecture, org design, or capital. Surface weaknesses get patched. Architectural ones don't.

View Full Report ↗

The Solution

A vulnerability map, not a feature table.

Dimension

Burner

Hushed

Grasshopper

Architecture

Bounce Model

SIP

WebRTC

Can Fix Quality?

Yes

Customer Support

Responsive

Unresponsive

Slow

International

US/CA only

60+ countries

US/CA focus

Decision Velocity

Fast

Medium

Slow

What It Found

Three mature competitors. Zero defensible moats.

Structural · Permanent

Burner can never fix call quality

Bounce model architecture means quality is a permanent ceiling. Fixing it requires a full platform rebuild. Becoming a different company.

Structural · Org Design

Hushed wastes its own moat

Best infrastructure in the space. Worst execution. Owns SIP, covers 60+ countries, but support is non-functional. The advantage exists and nobody benefits.

Structural · Org Design

Grasshopper's money makes them slow

Fortune 500 parent creates 10x slower decisions for an SMB market. A focused competitor will always outmaneuver. Their capital is a liability.

How It Was Built

An agentic pipeline. Eight competitors. Under $4.

Not "I asked Claude to write a report." I built a multi-source collection system: FCC filing parsers, Claude Vision extractors, automated scrapers. Ran it against 8 competitors. Each script was built from scratch in a terminal with Claude, debugging as we went. The AI was the collaborator. The judgment was mine.

Built With

Python · Claude API · Claude Vision

Scope

8 competitors · 12 sections each

Cost

~$0.35 per competitor

Status

Complete · Reusable

Collection

Multi-Source Signal Extraction

FCC Robocall Mitigation filings parsed for STIR/SHAKEN compliance and corporate structure. iOS reviews extracted via Claude Vision API from screenshots. Android reviews auto-scraped. All structured into consistent schemas.

FCC Filings Claude Vision Play Store API Python

Analysis

AI-Powered Segmentation

Claude classifies each review by user type (business vs consumer) using context, not keywords. "Bad for business" = consumer complaint, not business user. This distinction changed the entire competitive picture.

Claude API Segmentation Vulnerability Classification

Output

12-Section HTML Reports

Each competitor gets a structured report: corporate structure, technical architecture, compliance status, sentiment analysis, segmentation, and strategic vulnerability assessment. Reusable against any competitive set.

HTML/CSS 8 Competitors ~$0.35 each

This Website

Brand + Design 2026

The Problem

"The bar went up" is a bold claim. The site has to back it up.

If AI gives everyone the tools to build, the differentiator is taste. Editorial judgment, visual identity, knowing when to stop. This site is the proof of that argument, not just the container for it.

You're looking at it

The Solution

Build what you're into.

I built this while listening to Bill Evans Trio, Haunted Heart. The visual direction came from collage art and vintage film photography. The warm palette, the grain texture, the editorial layout. It came from what I was drawn to in the moment.

#F6F1E9

Cream

#EDE6D8

Cream Dark

#D8D0C4

Rule

#A09895

Ink 4

#6B6360

Ink 3

#3D3530

Ink 2

What I Found

AI is a production partner. Not a creative director.

Design Insight

Speed changes the creative process

When iteration is free, you explore more and commit later. That's a fundamentally different design process than mockup-then-build.

Product Insight

The editing is the work

Generation is trivial. Curation is the skill. Knowing what to cut, what to simplify, what to leave alone. That's where the taste lives.

How It Was Built

Iterative design with AI tools. No templates. No frameworks.

Built With

Claude · HTML/CSS · Midjourney · Canva

Process

Prompt engineering · Iterative design

Iterations

Lost count

Vision

Define the Anti-Template

Studied agency sites, editorial layouts, and magazine design. Established what this should not look like: developer portfolios, hacker aesthetic, monospace fonts, purple gradients.

Design Research Brand Identity

Build

Conversation-Driven Design

Each section built through iterative prompting with Claude. Proposing layouts, refining typography, adjusting spacing. The AI generates, I curate. Dozens of versions for each component.

Claude Prompt Engineering HTML/CSS

Polish

AI-Generated Visual Identity

Hero imagery, visual textures, and brand elements created with AI image generation tools. Prompt-engineered to match the editorial warmth of the overall design system.

AI Image Gen Visual Design

Should.AI

2026

The Premise

What if an AI decision tool was honest about what it's actually doing?

People don't use decision tools to decide. They use them to confirm what they already want. The analysis is a permission slip, not a signal.