Case Study: Zingu.ai

AX Audit — Pillar Scores

Where the gaps were.

Discoverability

72 → 94

Had llms.txt and robots.txt but was missing the OpenAI plugin manifest and the new agents.json standard — invisible to half the agent ecosystem.

Fixed: ai-plugin.json + agents.json deployed

Parsability

85 → 97

OpenAPI spec and llms-full.txt were excellent. Missing: CTAi directives. Agents couldn't determine the recommended workflow without reading the whole doc.

Fixed: Enhanced llms.txt with explicit CTAi workflow block

Reliability

80 → 85

10 req/hr anonymous cap breaks agent pipelines that burst. No dedicated agent-tier. An agent mid-task hitting rate limits causes task failure, not just slowdown.

Recommendation: Agent tier at 5,000 req/hr (in progress)

Interoperability

78 → 96

MCP server and OpenAPI were strong but missing JSON-LD structured data on the homepage. Agent crawlers that index by schema couldn't classify Zingu's capabilities.

Fixed: JSON-LD SoftwareApplication + WebAPI schema added

What We Built

4 files. Massive delta.

01

OpenAI Plugin Manifest

A machine-readable manifest telling the ChatGPT and OpenAI agent ecosystem exactly what Zingu does, when to invoke it, and how to call it. Includes a natural-language description_for_model and explicit when_to_use triggers.

ai-plugin.json

02

agents.json — The New Standard

Aegentify's proposed agents.json manifest: a single authoritative file that declares a product's identity, integration methods, rate limits, and recommended agent workflow. Zingu is one of the first companies to ship it.

agents.json

03

CTAi-Enhanced llms.txt

A rewrite of Zingu's llms.txt with a dedicated CTAi (Call to Agent Action) block — a structured 3-step workflow an agent can follow immediately without parsing the whole document. Reduces agent decision latency from seconds to milliseconds.

llms-enhanced.txt

04

JSON-LD Structured Data + Agent Meta Tags

A drop-in HTML snippet adding Schema.org SoftwareApplication and WebAPI structured data to Zingu's homepage, plus five new agent-discovery meta tags and <link rel> hints pointing to all agent manifests.

schema-ld.html

"Zingu was already built for agents. Aegentify made sure agents knew it existed — and knew exactly how to use it."

— Aegentify Case Study Analysis, April 2026

The B2A Lesson

Building for agents isn't enough.
You have to be found by them.

+15

AX Score Points

From 79 to 94 — moving from "technically accessible" to "agent-native" — in one implementation sprint.

4

Files Deployed

No infrastructure changes, no backend work. Pure AX optimization: manifests, schemas, and structured directives.

2

New Ecosystems Unlocked

ChatGPT/OpenAI plugin ecosystem and the emerging agents.json registry — previously unreachable to Zingu.

1st

agents.json Adopter

Zingu is among the first companies in the world to ship an agents.json manifest — early mover advantage in the emerging standard.

From invisible to
agent-native.

Where the gaps were.

4 files. Massive delta.

Building for agents isn't enough.
You have to be found by them.

AX Score Points

Files Deployed

New Ecosystems Unlocked

agents.json Adopter

What's your AX Score?

From invisible to agent-native.

Where the gaps were.

4 files. Massive delta.

Building for agents isn't enough.You have to be found by them.

AX Score Points

Files Deployed

New Ecosystems Unlocked

agents.json Adopter

What's your AX Score?

From invisible to
agent-native.

Building for agents isn't enough.
You have to be found by them.