Sign up, build your agent in the dashboard, and press Publish — it goes live for your customers on your surfaces: a public page, an embed widget on your site, your own domain. No setup files, no command line, no docs to read first.
Three commitments, in plain language. No credits to decode, no seats to buy, no infrastructure to reason about — you build the agent, we run and publish it.
One Publish press puts your agent live everywhere at once — public page, embed widget, API, webhook, cron, email — with a test-before-publish beat inside the minutes budget.
Free to start, no card. One plan, a prepaid dollar wallet, and every run a visibly priced line — with a hard never-overdraw promise. No credits, no seats, no surprise bill.
Your agent carries your brand on your-domain.com from the first publish — the white space the nearest competitor sells as Enterprise plus $199/mo extra.
At publish your agent lands on one of the three jobs SMBs pay for — starting from a template, never a blank canvas. It answers from your content, captures what you would have lost, and hands the customer off to a human. It catches what would slip; it never claims to replace your support.
Trained on your pages and docs, it answers questions in your voice — clone a template and edit, no blank canvas to stare at.
Catch the visitor who would otherwise bounce. Every conversation is a lead you can act on.
Turn the answer into a booking, then hand the customer to a real person — 85% still prefer one.
Live for your customers in minutes — on your page, your widget, your own domain.
Prefer to go under the hood? Bring your own Anthropic, OpenAI, or any provider key. We route through Cloudflare AI Gateway — cache hits, fallbacks, rate limits — and surface every cent in the same dashboard, with REST, webhooks, and MCP on the API surface.
// makent.config.ts export default defineAgent(({ name: 'support-triage', model: 'claude-sonnet-4-6' // ← any provider tools: [search, getCustomer, classify], memory: { vectorize: 'makent.support', topK: 5 }, })
Every step shows which tier ran it. No guessing where the latency came from.
Billing surfaces Think / WfP / Dynamic separately. Per agent. Per environment.
Usage flows through one queue into one Analytics Engine dataset. Filterable per env.