Build AI agents that do real work

HiroAI is the platform to describe, ground, connect and launch customer-facing agents, with your knowledge, your tools and business controls built in.

Start building Book a demo

100 CREDITS INCLUDED · NO CREDIT CARD · OPENAI-COMPATIBLE

01Website widgetEmbedded chat

Ask for it, ground it, ship it.

Four steps take an agent from a sentence to a live channel. Open each one on the bench to see what Hiro does behind it.

HIRO / WORKSPACE / DESCRIBEREADY

STEPS

1 / 4 OPENED

WORKSPACEDRAFT · LIVE

Ground first. An answer without a source is a guess somebody has to check.

Cap spend before the first run.

CHAT

Where is my order?

It ships in 2 to 4 business days. Want me to check today's status?

Source · shipping-policy.md

❯try: open ground

FIG. 001 — FOUR STEPS, ONE WORKSPACEDESCRIBE, GROUND, CONNECT, LAUNCH — OPEN ONE ON THE LEFT

Built around the work.

Start with the customer or workflow that matters most, then give Hiro the context and access it needs.

TRACK 01

Customer support

Grounded answers · Clean handoff

An agent that knows your products, cites the page it answered from and hands the whole conversation to a person when it should.

1 / 3Build a support agent

Credits, by the numbers.

Every plan runs the same core. Pick the capacity; credits follow real AI usage, never seats you do not fill.

Compare the plans

Free

credits, one-time grant

Starter

credits / month

Pro

credits / month

Business

credits / month

Everything an agent needs after the demo.

The pieces that become expensive to build and hard to govern once an agent meets real customers, already in one place.

A1
Visual agent builder
Build
Instructions, teams, response behaviour and workflows in one guided workspace.
A2
Knowledge & citations
Ground
Hybrid retrieval over approved documents, with the source page on every answer.
A3
Persistent memory
Ground
The right customer context carried across sessions and runs.
A4
Tools & integrations
Connect
Business apps, HTTP tools and MCP servers so agents can take useful action.
A5
Governance controls
Connect
Spend, rate, model, tool and customer policies enforced before every model call.
A6
Delivery surfaces
Launch
A web chat widget, the agent API, signed webhooks and connected apps from one core.
A7
Tracing & evaluation
Operate
Inspect each run, grade results and track quality, speed, cost and feedback.
A8
Usage & billing
Operate
Metering, quotas and credits per workspace, visible before the invoice.
A9
Developer-ready delivery
Launch
Scoped keys, signed webhooks and an OpenAI-compatible endpoint, with typed clients in preview.
A10
Isolated by design
Operate
Workspaces, products and modules isolated on one platform, operated from one console.