# voice.md: Scorable (formerly Root Signals)

## Communication Style
- **Tone and Personality:** The Scorable brand voice is **authoritative, pragmatic, and engineering-forward.** It positions itself as a serious partner for organizations building high-stakes AI. It avoids "AI hype" in favor of technical rigor, reliability, and measurable outcomes.
- **Stylistic Elements:** 
    - **Direct and Punchy:** Sentences are often short, declarative, and focused on the "how" and "why."
    - **Outcome-Oriented:** Content frequently highlights the transition from "unreliable" to "production-ready."
    - **Industry-Specific:** Uses terminology like "guardrails," "telemetry," "evals," "hallucinations," and "compliance" to establish credibility with developers and AI architects.
- **Vocabulary:** Precise and professional. Uses action-oriented verbs (e.g., *detect, measure, bootstrap, validate, scale*).

## Content Patterns
- **Common Themes:** The necessity of AI evaluation, moving beyond "vibe-coding" to structured testing, the role of the "AI Auditor," and the importance of semantic scoring for business KPIs.
- **Structural Approaches:**
    - **Problem-Solution-Proof:** Often starts by identifying a failure point (e.g., hallucinations, compliance gaps) and immediately presenting the Scorable mechanism as the fix.
    - **The "Before vs. After" Model:** Visually and textually demonstrates the impact of the tool by showing raw, unrefined AI output versus the corrected, "grounded" version.
    - **Technical Transparency:** Blog posts often dive into "what we learned" (e.g., OTel sinks, spec limitations), building trust through shared technical struggle.
- **Call-to-Action (CTA) Style:** High-intent and functional. CTAs are usually direct: "Try now," "Book a demo," or "Start measuring."

## Audience Interaction
- **Relationship Style:** Peer-to-peer. The brand speaks to developers, AI engineers, and technical leaders. It assumes the reader is intelligent, busy, and skeptical of "black box" AI solutions.
- **Formality:** Professional, but not stiff. It is comfortable using industry shorthand ("vibe-code," "evals") while maintaining a serious commitment to security and compliance (e.g., SOC 2, DPA).
- **Engagement:** The brand invites users into the technical ecosystem (Docs, API, CLI) rather than selling a "magic" solution.

## Guidelines & Examples

### Do's and Don'ts
- **DO** focus on reliability and "high-stakes" use cases.
- **DO** use clear, evidence-based claims.
- **DO** emphasize the "control surface" aspect of AI.
- **DON'T** use fluffy marketing buzzwords like "game-changing" or "revolutionary."
- **DON'T** over-simplify complex AI engineering challenges; acknowledge the difficulty of the problem.

### On-Brand Phrases
- "Measure what matters."
- "Unlock customer-facing AI agents for safe high-stakes use at scale."
- "Why can't I just vibe-code reliability?"
- "One function call away."
- "Production AI in regulated industries needs more than engineering oversight."

### Content Types
- **Technical Tutorials:** Deep dives into how to implement specific evaluation patterns.
- **Comparison/Opinion Pieces:** Analysis of build-vs-buy or why current industry standards (like OTel) are evolving.
- **Feature Announcements:** Grounded in how the feature solves a specific, real-world failure point.
- **Product Documentation:** The primary "source of truth" for how the product integrates into existing developer workflows.