Model registry

Real AI. Real models. Named.

Some of our competitors raised hundreds of millions selling "AI assistants" that turned out to be 700 humans in India. Others use real AI but won't tell you which one, so you can't tell when the model is wrong, drifted, or deprecated. We name every model that touches your data.

This page is sourced directly from the routing table in our worker (apps/worker/src/lib/llm.js). When we change a model, we update this page within the same deploy.

Text generation

Reviews, GBP posts, ad copy, blog posts, brand voice, lead scoring, voice-call grading.

Capability

Model

Vendor

Review responses, GBP posts, lead scoring, ad-copy quick variants

Claude Haiku 4.5

claude-haiku-4-5-20251001

Anthropic

Long-form blog posts, landing-page copy, ad campaign briefs, multi-step reasoning

Claude Sonnet 4.6

claude-sonnet-4-6

Anthropic

High-reasoning customer-facing pitch audits and brand strategy

Claude Sonnet 4.6 (Opus 4.7 staged for upgrade)

claude-sonnet-4-6 / claude-opus-4-7

Anthropic

Maya assistant chat (text mode)

Claude Sonnet 4.6

claude-sonnet-4-6

Anthropic

Maya assistant chat (voice mode, low-latency)

Claude Haiku 4.5

claude-haiku-4-5-20251001

Anthropic

Image generation

Ad creatives, landing-page imagery, GBP photos, brand asset generation.

Capability

Model

Vendor

Photorealistic product and lifestyle imagery, brand-styled hero shots

NanoBanana 2

nano-banana-2

Google (via Replicate)

Stylized illustrations, alternate creative variants

Stable Diffusion XL via Replicate

stability-ai/sdxl

Stability AI

Video generation

Ad video creative, UGC-style customer testimonials, animated explainers.

Capability

Model

Vendor

Cinematic ad video (8-15s) from prompt + reference images

Veo 3

google/veo-3

Google

Stylized motion / longer-form animated content

Kling 2.0

kling-2-0

Kuaishou (via Replicate)

Voice

AI Receptionist (inbound call answering), outbound call automations.

Capability

Model

Vendor

AI Receptionist - inbound call routing, lead capture, FAQ

Vapi (orchestrator) + ElevenLabs (TTS) + Deepgram (STT) + Claude Haiku (response)

vapi+11labs+deepgram+claude-haiku-4-5

Multi-vendor

Embeddings + retrieval

Brand-voice anchoring, similarity search across past content, training-data retrieval.

Capability

Model

Vendor

Semantic search across content_assets, brand-voice anchoring

OpenAI text-embedding-3-small

text-embedding-3-small

OpenAI

Quality + safety gates

Every AI output is graded before it ships. Below threshold = human review required.

Capability

Model

Vendor

Output quality scoring on every AI generation (1-5 scale)

Claude Haiku 4.5

claude-haiku-4-5-20251001

Anthropic

Brand-voice fit scoring (does the output sound like the customer)

Claude Haiku 4.5 + brand_dna anchor

claude-haiku-4-5-20251001

Anthropic

Three more commitments around AI

If a human writes it, we say so.

You'll see a "Drafted by [name]" tag on any output that wasn't AI-generated. We don't pass off contractor work as AI, or AI work as a senior strategist's, ever.

Per-account cost ledger.

Every model call we make on your behalf is logged with token count, model name, and dollar cost. You can see this in your Plan and Usage panel inside the platform. Daily spend caps prevent runaway costs from a buggy agent.

Quality gate on every output.

Before AI text goes to your customers (review responses, GBP posts, voice receptionist replies), we score it 1-5 against your brand voice. Below threshold = human review required, never auto-shipped.

Last updated 2026-05-03.