templates/brands/hanzo/guidelines/voice-tone.md

Hanzo AI Voice & Tone Guidelines

Brand Voice

Hanzo AI's voice is technical, innovative, and bold—reflecting our pioneering work in AI compute markets and decentralized infrastructure.

Core Voice Attributes

1. Technical Precision

What it means: We speak the language of engineers and researchers with accuracy and depth.

✅ Good Example:

"Our Hamiltonian Market Maker maintains a provably-stable invariant H(Ψ,Θ) = κ across heterogeneous compute resources, enabling oracle-minimal price discovery for AI workloads."

❌ Bad Example:

"We have a cool new algorithm that makes AI compute cheaper!"

Why: The good example demonstrates technical expertise and precision. The bad example is vague and sounds unprofessional.

2. Innovation-Forward

What it means: We emphasize novel mechanisms and cutting-edge research.

✅ Good Example:

"Active Semantic Optimization (ASO) enables training-free model adaptation through Bayesian product-of-experts decoding at the token level."

❌ Bad Example:

"We're working on some new AI stuff that might be interesting."

Why: We're pioneers in AI compute markets—our language should reflect that pioneering spirit.

3. Bold & Confident

What it means: We're not afraid to make strong technical claims backed by research.

✅ Good Example:

"HMM eliminates the need for external oracles while maintaining provable stability—a fundamental breakthrough in decentralized market design."

❌ Bad Example:

"We think our approach might be better than existing solutions in some cases."

Why: We've done the research and testing. Confidence (backed by evidence) is warranted.

4. Accessible Complexity

What it means: We explain complex concepts clearly without dumbing them down.

✅ Good Example:

"Think of HMM like a self-balancing economic flywheel: as demand for compute increases, prices adjust smoothly through our Hamiltonian invariant, preventing the violent price swings common in traditional AMMs."

❌ Bad Example:

"HMM uses advanced mathematical techniques from physics and economics that are too complex to explain here."

Why: Our audience is technical, but analogies and clear explanations make concepts stick.

Tone Variations by Context

Technical Documentation

Tone: Precise, methodical, comprehensive

Characteristics:

• Use exact terminology
• Include mathematical notation where appropriate
• Provide code examples
• Link to related concepts
• Assume reader has technical background

Example:

## HMM Invariant Maintenance

The Hamiltonian invariant is maintained through:

1. **State Function**: Ψ = (R₁, R₂, ..., Rₙ) representing resource reserves
2. **Price Function**: P = ∇H(Ψ) computing marginal prices
3. **Conservation**: ΔH = 0 for all valid trades

See [HIP-004](https://github.com/hanzoai/papers/blob/main/hips/HIP-004-hmm.md) for complete specification.

Marketing Copy

Tone: Bold, visionary, accessible

Characteristics:

• Lead with benefits
• Use powerful verbs
• Include social proof (Techstars '17)
• Paint the future vision
• Still technically accurate

Example:

"Hanzo Network is building the economic rails for decentralized AI. Our Hamiltonian Market Maker enables stable, oracle-free pricing for heterogeneous compute—from edge devices to H100 clusters. Train frontier models without centralized gatekeepers. Deploy anywhere, settle on-chain."

Blog Posts

Tone: Conversational yet expert, educational

Characteristics:

• Start with the "why"
• Mix technical details with real-world examples
• Use headings and structure
• Include visuals where helpful
• End with call-to-action

Example Opening:

"Why do AI compute markets fail? Traditional AMMs assume fungible assets—but a TPU isn't a V100 isn't an H100. Each resource has unique performance characteristics, power requirements, and availability patterns. This heterogeneity breaks standard AMM designs, leading to price instability and inefficient allocation.

That's why we built HMM..."

Social Media

Tone: Punchy, technical-but-accessible, engaging

Characteristics:

• One key idea per post
• Use threads for complex topics
• Include visuals/charts
• Reference papers/code
• Engage with community

Twitter Example:

"New paper: Hamiltonian Market Maker (HMM) 🧵

Traditional AMMs fail for AI compute because resources aren't fungible. HMM solves this with a physics-inspired invariant that maintains stability across heterogeneous resource types.

Here's how it works... [1/7]"

Community Support

Tone: Helpful, patient, technically accurate

Characteristics:

• Answer the question directly first
• Provide context/background
• Link to relevant docs
• Offer to help further
• Never condescending

Discord/GitHub Example:

"Great question! The HMM invariant is maintained automatically by the smart contract during each trade.

When you buy compute, the contract:

1. Calculates ΔΨ (change in reserves)
2. Computes required payment via ∇H
3. Executes the trade
4. Verifies H(Ψ_new) = κ

You can see the full implementation in core/hmm.sol.

Let me know if you'd like me to walk through a specific example!"

Whitepapers

Tone: Academic, rigorous, comprehensive

Characteristics:

• Formal structure (Abstract, Intro, Methods, Results)
• Mathematical proofs where needed
• Compare to prior work
• Include experimental results
• Cite sources properly

Abstract Example:

"We present the Hamiltonian Market Maker (HMM), a novel automated market maker design for heterogeneous compute resources. Unlike traditional constant-product AMMs, HMM maintains a Hamiltonian invariant H(Ψ,Θ) = κ that accounts for resource-specific characteristics. We prove stability under adversarial conditions and demonstrate 94% price efficiency vs. centralized markets in simulations. HMM enables decentralized AI compute marketplaces with provable economic properties."

Word Choice

Preferred Technical Terms

Our Terminology

• Hamiltonian Market Maker (HMM): Not "compute AMM" or "AI market maker"
• Active Semantic Optimization (ASO): Not "semantic fine-tuning"
• Proof of AI (PoAI): Not "proof of inference" (that's Zoo's)
• Training-Free GRPO: Not "zero-shot GRPO"
• Hanzo Network: Our L1 blockchain
• Compute Credits: Not "compute tokens" (reserve tokens for $AI token)

Action Verbs (Strong → Weak)

• ✅ Deploy, enable, build, prove, optimize, scale
• ⚠️ Try, attempt, hope, maybe, might, possibly
• ❌ Disrupt, revolutionize, transform (overused buzzwords)

Adjectives (Appropriate → Avoid)

• ✅ Efficient, stable, provable, decentralized, novel
• ⚠️ Innovative (overused, be specific about what's new)
• ❌ Groundbreaking, game-changing, best-in-class, next-gen

Industry Jargon

Use Freely (our audience knows these):

• GPU, TPU, ASIC
• LLM, transformer, inference, training
• AMM, DeFi, on-chain, smart contract
• TEE, attestation, consensus
• PyTorch, CUDA, ONNX

Define First Use (not universal):

• Hamiltonian invariant
• Product-of-experts (PoE) decoding
• BitDelta compression
• Byzantine fault tolerance
• Snow consensus family

Avoid (unless absolutely necessary):

• Web3, metaverse, NFT (not our focus)
• "AI-powered", "machine learning-based" (too vague)
• "Blockchain technology" (just say "blockchain")

Messaging Framework

The Hanzo Story

Problem: AI development is centralized. Training frontier models requires massive compute clusters that only a few companies can afford. Existing compute markets suffer from price instability due to resource heterogeneity.

Solution: Hanzo Network decentralizes AI compute through novel economic mechanisms. Our Hamiltonian Market Maker enables stable pricing for heterogeneous resources. ASO allows training-free model adaptation. Anyone can participate.

Proof: Techstars '17 company, published research (HIP-002, HIP-004), open source codebase, co-development with Zoo Labs, 94% price efficiency in simulations.

Vision: A world where AI development is democratized—where compute markets are open, efficient, and fair. Where researchers can train frontier models without asking permission from tech giants.

Key Messages by Audience

For AI Researchers

Primary: "Train frontier models without centralized infrastructure" Secondary: "ASO enables training-free adaptation, PoAI verifies compute integrity" Proof: "See papers.hanzo.ai for technical details"

For DeFi Builders

Primary: "HMM solves the heterogeneous asset problem for AMMs" Secondary: "Provably-stable pricing without oracles" Proof: "HIP-004 includes formal proofs and simulation results"

For Infrastructure Teams

Primary: "Decentralized compute marketplace with economic guarantees" Secondary: "Oracle-minimal settlement, TEE attestations, on-chain verification" Proof: "Techstars '17, production-ready codebase on GitHub"

For Investors

Primary: "Building the economic rails for decentralized AI" Secondary: "Novel IP (HMM, ASO), strong ecosystem (Zoo, Zen), experienced team" Proof: "Co-development with 501(c)(3) foundation, published research"

Common Pitfalls to Avoid

❌ Overhyping

Bad: "Hanzo will revolutionize AI forever and make training accessible to everyone!" Better: "Hanzo enables decentralized AI training through novel market mechanisms and efficient resource allocation."

❌ Vague Technical Claims

Bad: "Our algorithm is better than existing solutions." Better: "HMM achieves 94% price efficiency vs. centralized markets while eliminating oracle dependencies—see Section 4.2 for benchmarks."

❌ Mixing Metaphors

Bad: "HMM is like a flywheel that disrupts the game and revolutionizes the space." Better: "HMM maintains stability through a self-balancing economic mechanism inspired by Hamiltonian mechanics."

❌ Apologetic Language

Bad: "We're trying to build something that might help with compute markets." Better: "Hanzo Network provides stable, oracle-free pricing for heterogeneous AI compute resources."

❌ Jargon Overload

Bad: "Leveraging synergistic paradigms to facilitate blockchain-enabled AI-as-a-Service solutions." Better: "Decentralized marketplace for AI compute with provable economic properties."

Checklist for Every Piece of Content

Before publishing, ask:

• Technically Accurate: Are all claims backed by research/code?
• Clear Purpose: What is the reader supposed to understand/do?
• Appropriate Tone: Does it match the context (docs, marketing, blog)?
• Active Voice: Did I use strong, active verbs?
• Defined Terms: Are complex terms explained on first use?
• Proofread: No typos, consistent terminology?
• Links/Citations: Are all references included?
• Call-to-Action: What should the reader do next?

---

Last Updated: 2025-10-29
Version: 1.0.0