🐕 Caro
⭐ Features 🎬 Watch Demo
Compare
📊 All Comparisons ⚡ vs Warp 🤖 vs GitHub Copilot 👻 vs Kiro CLI
Resources
📝 Blog 🔍 Explore ❓ FAQ 🗺️ Roadmap 📖 Glossary 🐙 GitHub
💖 Support Caro
Appearance
🌓 Dark Mode
Get Started
Jobs To Be Done Framework

Understanding Why
People Hire Caro

People don't buy products—they hire them to get jobs done. This page documents the specific jobs users hire Caro to perform, using the canonical JTBD framework.

Based on Anthony Ulwick's Outcome-Driven Innovation and Clayton Christensen's JTBD theory

The JTBD Lens

Jobs, Not Features

Users don't care about Caro's 50 safety patterns or bundled Qwen model. They care about not breaking production and getting work done even when offline.

Hiring and Firing

When users "hire" Caro, they're "firing" something else—manual review, Google searches, or cloud-dependent AI tools. Understanding what gets fired reveals the real job.

Three Job Types

Functional: Tasks to accomplish. Emotional: How users want to feel. Social: How users want to be perceived.

Core Jobs (Functional)

The practical tasks users hire Caro to accomplish

Critical Daily Gap: High
When I'm about to run a command in production, I want to validate it won't cause damage, so I can prevent outages and keep my systems running.
Critical Daily Gap: Very High
When I'm working in an air-gapped or restricted environment, I want to get AI assistance without network access, so I can be productive even when isolated from the internet.
High Daily Gap: Medium
When I know what I want to do but not the exact command syntax, I want to describe my intent in plain English, so I can get a working command without searching documentation.
High Weekly Gap: High
When I'm writing commands that need to work on Mac, Linux, and CI, I want to generate platform-aware commands automatically, so I can stop debugging BSD vs GNU flag differences.
Critical Weekly Gap: High
When I'm troubleshooting an incident at 3 AM, I want to get commands quickly without making mistakes, so I can resolve incidents faster without causing additional damage.
High Monthly Gap: Very High
When I'm responsible for my team's command-line safety, I want to deploy safety standards without micromanaging, so I can protect my team from dangerous commands automatically.

Emotional Jobs

How users want to feel when doing their work

Feel confident when running production commands
Pain Point: Anxiety about accidentally breaking things
How Caro Helps: Deterministic safety validation provides certainty before execution
Feel in control during high-pressure incidents
Pain Point: Stress leads to mistakes, mistakes make incidents worse
How Caro Helps: Fast, offline-capable assistance when you need it most
Feel competent despite not memorizing every flag
Pain Point: Imposter syndrome when forgetting basic commands
How Caro Helps: Natural language interface lets you focus on intent, not syntax
Feel secure working in restricted environments
Pain Point: Other tools require internet, leaving you stranded
How Caro Helps: Bundled model works where nothing else can

Social Jobs

How users want to be perceived by others

Job Triggers

Events that cause users to seek a solution

Near-miss incident Immediate

Almost ran a dangerous command, caught it just in time

Actual incident caused by command error Immediate

Production outage from typo or wrong flags

New air-gapped deployment High

Moved to restricted environment, existing tools don't work

Cross-platform migration Medium

Moving from Mac to Linux or vice versa, commands breaking

Team growth Medium

New engineers joining who might make dangerous mistakes

Security audit High

Need to document and enforce command-line safety standards

Desired Outcomes

What users measure success by (Outcome-Driven Innovation format)

Outcome Statement Metric Without Caro With Caro
Minimize the time it takes to validate a command is safe Seconds from command to safety verdict Manual review: 30-60 seconds Instant: <100ms
Minimize the likelihood of running a destructive command Dangerous commands executed per month Varies, depends on vigilance Zero (blocked before execution)
Minimize the time it takes to get a working command Seconds from intent to executable command Google + trial/error: 2-5 minutes <2 seconds inference
Minimize dependency on network connectivity Percentage of functionality available offline 0% (cloud AI tools) 100% (bundled model)
Minimize cross-platform debugging time Time debugging BSD vs GNU differences 15-30 minutes per issue Zero (platform-aware generation)

Hiring and Firing

What solutions users fire when they hire Caro

Command validation
Fires:
  • Manual review
  • Hope and prayer
  • Bash aliases with safety checks
Hires: Caro's deterministic pattern matching
Because: Can't hallucinate, doesn't require vigilance
AI shell assistance
Fires:
  • ChatGPT (requires network)
  • Copilot CLI (requires network)
  • Stack Overflow searches
Hires: Caro's offline-first AI
Because: Works in restricted environments, no dependency on cloud
Command learning
Fires:
  • Man pages
  • TLDR pages
  • Google searches
Hires: Caro's natural language interface
Because: Describe intent, get working commands instantly
Cross-platform scripting
Fires:
  • Trial and error
  • Maintaining separate scripts
  • Conditional shell logic
Hires: Caro's platform detection
Because: Generates correct flags for current platform

Job Map: Safe Command Execution

Step-by-step breakdown of the core job

1
Define
Decide what action to take
Example: "I need to delete old log files"
Caro's Role: Understands natural language intent
2
Locate
Identify what to operate on
Example: "Files older than 30 days in /var/log"
Caro's Role: Generates correct find/date syntax
3
Prepare
Formulate the command
Example: "find /var/log -mtime +30 -delete"
Caro's Role: Creates platform-correct command
4
Validate
Check command is safe
Example: "Is this going to delete something important?"
Caro's Role: Matches against 52+ dangerous patterns
5
Confirm
Make go/no-go decision
Example: "Safe verdict gives confidence to proceed"
Caro's Role: Clear safety status with explanations
6
Execute
Run the command
Example: "Press enter with confidence"
Caro's Role: Optional: direct execution with safety gate
7
Verify
Confirm expected result
Example: "Check files were deleted correctly"
Caro's Role: Can generate verification commands

The Core Insight

Users don't hire Caro because it has a bundled AI model or 50 safety patterns. They hire Caro because they want to feel confident running commands, get work done in any environment, and protect their systems from their own mistakes.

The functional job is command safety. The emotional job is peace of mind. The social job is being seen as careful and professional.

Try Caro in 30 Seconds

No account. No API key. No data collection. Just safer shell commands.

bash <(curl --proto '=https' --tlsv1.2 -sSfL https://setup.caro.sh)

Then run:

caro "find files modified in the last 7 days"
✓ Installs to ~/.cargo/bin
✓ Single binary, no dependencies
✓ Uninstall anytime: cargo uninstall caro

Prefer to build from source? See all installation options →