Tech Edit LAB

AI Tool Directory

Tier 01

Anthropic, Claude logo

Anthropic, Claude

LLM Labs · Frontier (US, closed-weight)

Anthropic's Claude family (Opus, Sonnet, Haiku) is one of two or three frontier LLMs that hospitality vendors most often build on top of. Strong on long-document reasoning, coding, and "policy-faithful" outputs, i.e. it follows the system prompt closely. Used directly via Claude.ai, the API, AWS Bedrock, and Google Vertex AI.

Identity

Frontier closed-weight LLM lab · founded 2021 · ~$60B+ valuation (early 2026)

When it fits

Vendors and in-house teams that prioritise instruction-following discipline (refund policies, allergen rules, brand-voice constraints) and long-context work like reading 60 reviews to summarise three themes.

Pricing

Quote at this tier — direct API consumption (per-token, varies by model) and AWS Bedrock / Google Vertex resale with enterprise contracting. Consumer Claude.ai pricing is covered separately under the Tier 02 chat entry.

Vendor HQ

US

Why hospitality teams notice Claude

Two things show up consistently when hotel teams compare frontier chat models. The first is instruction-following discipline. Claude tends to stay inside the rules of a system prompt, refund tiers, allergen disclaimers, brand-voice constraints, even when the conversation drifts. That makes it a forgiving choice for vendors building guest-messaging features where a single wrong sentence can become a complaint.

The second is long-document handling. Reading a 60-review chunk and pulling out the three recurring themes is a Claude-shaped task. The model holds the context, cites the relevant snippets, and resists the urge to invent a fourth theme that wasn't in the data.

How it shows up in the directory

Anthropic Claude appears three times in the directory because it sits at three different layers:

  • Tier 01 · Anthropic Claude (lab): the model family itself (Opus, Sonnet, Haiku).
  • Tier 02 · Anthropic Claude (chat): Claude.ai, the consumer chat product.
  • Tier 02 · Anthropic Claude Code: the agentic coding CLI.

When a hospitality vendor says "we use Claude," they almost always mean the lab, the API or one of its cloud resellers (AWS Bedrock, Google Vertex). The chat and Code products are what your own team opens directly.

What to ask in a procurement conversation

If a vendor's product is built on Claude, the questions worth asking are:

  1. Which model size? Opus, Sonnet, and Haiku have very different cost and latency profiles.
  2. Which deployment? Direct API, AWS Bedrock, or Google Vertex changes data-residency.
  3. What's in the system prompt? This is the actual product, the model is the engine.
  4. What's the fallback? What happens if the API has an outage during peak check-in?