01Early access
Agent Readiness Gate
The pre-deployment gate you install in your cloud: attack pack, evaluation harness, cost baseline, permission audit, Agent BOM — one Evidence Report, on every release.
Product details →The production layer for AI agents
AI agents that pass the gate, stay on budget and leave a record.
Software that tests AI agents before they go live, governs what they cost, verifies the tools they load and keeps the records auditors ask for — installed in your own cloud, at published prices.
- Published pricesevery product, every tier, on one page
- Installs in your clouda CLI, a container, a gateway; nothing leaves your account
- Open-source framethe Readiness Kit ships on GitHub, Apache-2.0
- No lock-inopen formats, your data, cancel at renewal
The gap
Most agents stall between the demo and production. The model is rarely the reason.
40%
of agentic-AI projects will be cancelled by 2027 — cost, unclear value or inadequate risk controls, not model quality.
5%
of companies have every critical control in place to run AI agents, although most are already deploying them.
54
AI-agent incidents a year in the average organisation; 85% of CIOs lack real-time visibility of AI spend.
Products
Four products. One platform for the part after the demo.
02In build
Agent Cost Governor
A budget the agent cannot exceed — per task, per step, per model — enforced in your own cloud before the invoice, not after.
Product details →03Early access
MCPSentry
Continuous verification of the MCP servers and skills your agents load — approved once, checked every time, with an Agent BOM to prove it.
Product details →04Design partners
Compliance Evidence Pack
Tamper-evident records from running agents, mapped to ISO/IEC 42001, the EU AI Act, Colorado SB 26-189 and your auditor's request list.
Product details →Open source · free Readiness Kit — the frame under the Gate, Apache-2.0, v0.1 · 18 October 2026.
How it works
Install today. Evidence on the first run.
$3,600 / yearfounding licence · Gate Team · list $6,000 · first 10 customers until 31 December 2026Book a 30-minute callEvery price →
- 01
Install
A CLI and a container in your own AWS, Azure or GCP account, or on a laptop. Nothing is copied out.
- 02
Run the Gate
The attack pack, an evaluation set seeded from your real failures, the cost meter, a permission audit and the Agent BOM, in one run.
- 03
Ship with evidence
The Evidence Report on every release; the CI gate blocks the ones that regress.
What you receive
One Evidence Report. Ten pages. Every number traceable.
What the Gate generates on every release, published in full on a fictional agent so you can judge the format before you license it.
Solutions by role and industry
Who it is for.
Engineering leads
A held-out evaluation set, a cost-per-task number and a harness that gates every release.
Agent Readiness Gate →Security and AppSec
Attack findings mapped to OWASP ASI01–ASI10 and continuous verification of the tools agents load.
MCPSentry →Risk, compliance and audit
An Agent BOM, a controls statement and tamper-evident records mapped to ISO/IEC 42001 and the EU AI Act.
Compliance Evidence Pack →Finance and platform owners
A budget the agent cannot exceed, enforced in your cloud before the invoice.
Agent Cost Governor →Insurance, banking and healthcare
The records regulators now ask for, from the agents making the decisions.
Industries →Integrators and consultancies
The Gate inside your programmes under a partner licence, with your brand on the report.
Partner licences →What's next
Five dates. Each one is a commitment.
If a date moves, this page says so the same day. We publish a product when it ships, not while we build it.
- 18 Oct 2026
Readiness Kit v0.1
- 20 Nov 2026
Gate early access
- Dec 2026
Agent Cost Governor pilots
- Q1 2027
MCPSentry free scanner
- Q2 2027
Compliance Evidence Pack
Resources
Things you can check without talking to us.
Free tool
The Readiness Score
Eight questions, two minutes: the two areas most likely to stall your agent.
Score your agent →The Lab
Public work anyone can check
One checkable item a week, advisories under our name, benchmarks with their methods.
The Lab →Insight
The $50,000 hour: why agent cost needs a governor, not an alert
Every budget control sold today stops at the API key. The money is spent inside the agent loop, one retry, one escalation and one spawned sub-agent at a time — and the alert arrives a day later.
Read →Tell us the agent. We'll tell you where it would stall.
A 30-minute call, no deck, no charge. You leave with a clear read on production readiness and, if it fits, an early-access date and a licence quote inside the published price.