Get your business found, recommended, and booked by AI assistants.

We help small businesses win the new front page: the answer an AI assistant gives when a buyer asks it for a recommendation. WARBAND is an operator console — build a team of AI agents in plain English, measure your AI citation share, and fix what's holding you back.

Private beta — from $0 platform fee. Bring your own keys. No terminal, no code.

Join the waitlist

Already have access? Sign in  ·  Questions? Book a demo

AI recommends businesses now — does it recommend yours?

Buyers are increasingly asking ChatGPT, Claude, Perplexity, and Google AI Overviews for recommendations — often before they reach your website. If those assistants can't read, trust, and act on your offer, they recommend someone else. WARBAND measures whether you're being cited and shows you how to fix it.

The Traffic Loop

Measure → Diagnose → Prepare

Three tools that work together to grow the traffic AI assistants send you. Every step is read-only or dry-run — nothing is published or sent without your explicit approval, and every tool fails closed if a required key is missing.

Step 1 · Measure

ORACLE — Citation Audit

ORACLE mystery-shops AI answer engines with your buyers' real questions and reports your Citation Share — how often you're cited versus competitors — with a per-engine breakdown and movement since your last run.

Read-only. Nothing is published.
Step 2 · Diagnose

Beacon — Agent-Readability Audit

Beacon crawls your site against the Agent-Readability Rubric (ARR v1) and scores it 0–100 on whether AI agents can find, trust, and act on your offers — with per-category findings for structured data, pricing clarity, trust signals, and action affordances.

Read-only. Respects robots.txt.
Step 3 · Prepare

Minuteman — Intent-Velocity Prospector

Minuteman detects public buying signals in your niche, matches them to your ideal customer profile, and stages prospects in a review queue with the signal attached as evidence — so your outreach is ready the moment you approve it.

Dry-run only. Nothing is sent.

What you can build

A whole agent team, run from one console

WARBAND turns a plain-English description into a working AI agent — then gives you the surfaces to run, watch, and scale it. No terminal sessions, no repo checkout, no code.

Agents

Describe what you want done — "when a quote request comes in, check my service area and draft a reply with a rough estimate" — and the NL Builder returns a named agent draft with a system prompt, capabilities, and triggers. Activate it, deploy it, done.

Build → Activate → Deploy

Video Studio

Turn a topic into broadcast-ready short video: the Video Loop plans a storyboard, drafts the script, and assembles clips so you can publish consistent content without a production crew.

Topic → Storyboard → Publish

Marketplace

Skip the blank page. The Marketplace ships pre-built agents — Inbox Triage, RFP Closer, Content Engine, Spend Watch and more — each with an honest status. If an agent isn't wired yet, it's labelled "activating," never faked.

Browse → Install → Run

Pricing & access

How much does WARBAND cost, and how do I get in?

$0 platform fee

WARBAND is in invite-only private beta and charges no per-run platform fee. You bring your own LLM keys (Anthropic, OpenAI, DeepSeek, or Kimi) and pay your provider directly for the tokens your agents use — typically a few hundred to a few thousand tokens per run. Track every cost under Spend Watch inside the console.

Access is by beta code or waitlist

Have a beta code? Register and enter it to create your account. No code yet? Join the waitlist from the same page and we'll send an invitation. Prefer to talk first? Book a demo and we'll walk you through the Traffic Loop.

Join the waitlist

Self-audit · Proven before launch

We ran Beacon on this page before publishing it

Before this page went live, we ran it through the same Beacon Agent-Readability Rubric (ARR v1) we sell. The 97 / 100 score below is a real result from a real audit — not a placeholder. Read the full methodology.

Beacon Agent-Readability Score (ARR v1)
97 / 100
Pre-launch self-audit of warband.dev — run before this page published.
What Beacon scores
4 rubric
categories
Structured data · Pricing clarity · Trust signals · Action affordances. Full rubric →

FAQ

Frequently asked questions

What does WARBAND do?
WARBAND is an operator console for small businesses. You build a team of AI agents by describing what you want in plain English, and you run the Traffic Loop to get your business found, recommended, and booked by AI assistants. It runs on your own LLM keys (BYOK) — WARBAND charges no per-run fee.
Does AI recommend my business?
Buyers increasingly turn to ChatGPT, Claude, Perplexity, and Google AI Overviews for recommendations. WARBAND's ORACLE tool mystery-shops those engines with your buyers' real questions and measures your Citation Share — how often they cite you versus competitors. Beacon then diagnoses why, scoring how readable your site is to AI agents from 0 to 100.
What is the Traffic Loop?
Three tools in sequence. Measure: ORACLE finds your Citation Share across AI answer engines. Diagnose: Beacon scores your website 0–100 against the Agent-Readability Rubric. Prepare: Minuteman stages a dry-run review queue of prospects from public buying signals. Nothing is published or sent without your explicit approval.
What is BYOK (bring your own key)?
BYOK means Bring Your Own Key. WARBAND never uses a platform LLM key for your runs. You add your own Anthropic, OpenAI, DeepSeek, or Kimi key, and every token an agent uses is billed directly to your provider account. Your keys are encrypted and isolated per tenant — WARBAND never sells or routes them.
How much does WARBAND cost?
WARBAND is in invite-only private beta with no per-run platform fee — from $0. You pay your own LLM provider directly for the tokens your agents use (typically a few hundred to a few thousand tokens per run). Monitor usage inside the console under Spend Watch.
Is anything sent or published without my approval?
No. Beacon and ORACLE are read-only audits — they never modify or publish anything. Minuteman is dry-run only: it stages prospects in a review queue but sends no email or message. The dry-run guard is a hard code constraint, not a toggle. Every tool fails closed if a required key is missing.
How do I get access to WARBAND?
WARBAND is invite-only during private beta. If you have a beta code, register at /register. If you don't, join the waitlist from the same page and we'll send an invitation. You can also book a demo by emailing beta@warband.dev.

Ready to see if AI recommends you?

WARBAND is in invite-only private beta. Join the waitlist and we'll send your access.

Join the waitlist