Haynechi
Haynechi Index Understand where you stand in AI Search before competitors own the prompt map.

Docs

System architecture for answer operations.

How Haynechi models prompts, answers, sources, agents, and proof while keeping production readiness boundaries explicit.

Architecture

Answer evidence moves through four product layers.

The model is intentionally explicit: every surface shows where evidence entered, how it was normalized, what work it created, and how movement is reviewed.

Trust boundaries
System LoopProduct model
01 Capture

Prompts, answer snapshots, citations, crawler events, referral context, competitor language

evidence intake
02 Normalize

Entities, engines, regions, source classes, sentiment, freshness windows, and owner fields

answer ledger
03 Operate

Prompt maps, source plans, supervised agent runs, approvals, and campaign backlogs

workflow layer
04 Prove

Answer deltas, citation movement, crawl windows, referral quality, and executive readouts

proof layer
Core ObjectsData contract
Prompt Cluster

intent, buyer stage, modifiers, region, competitor set

prompt graph
Answer Snapshot

engine, prompt, answer text, sentiment, rank, captured at

answer ledger
Source Record

domain, page, source type, freshness, ownership path

citation graph
Agent Run

template, inputs, nodes, artifacts, approvals, owner

runtime
Proof Event

shipped work, answer delta, citation shift, crawl window

readout
Stack RationaleCurrent path
implemented Static marketing layer

Node-generated HTML keeps the public surface fast, inspectable, and Cloudflare Pages-ready with crawler guidance and deployment metadata generated alongside routes.

implemented Structured content collections

Feature, solution, resource, legal, and architecture content is represented as data before being rendered into routes.

implemented Product primitive library

Reusable sections for agents, answer maps, operating consoles, proof boards, article rails, and pilot offers keep detail consistent.

implemented Contact intake function

Cloudflare Pages Function validates submissions, filters honeypot traffic, and can forward to CRM or webhook destinations.

future app Authenticated workspace

The pilot workspace sits behind a separate app boundary for accounts, audit logs, exports, SSO planning, and data retention controls.

scoped Analytics and CRM sync

PostHog or Plausible plus HubSpot, Salesforce, or warehouse sync connect after production tracking policy and buyer-approved destinations are defined.

Stack Decisions

Built static first, with a clear path to a governed workspace.

The public layer stays fast and mostly static. The product model is structured around prompts, answers, sources, agents, and proof events so workspace controls can move into a governed app boundary when customer data enters the system.

implemented

Static marketing layer

Node-generated HTML keeps the public surface fast, inspectable, and Cloudflare Pages-ready with crawler guidance and deployment metadata generated alongside routes.

implemented

Structured content collections

Feature, solution, resource, legal, and architecture content is represented as data before being rendered into routes.

implemented

Product primitive library

Reusable sections for agents, answer maps, operating consoles, proof boards, article rails, and pilot offers keep detail consistent.

implemented

Contact intake function

Cloudflare Pages Function validates submissions, filters honeypot traffic, and can forward to CRM or webhook destinations.

future app

Authenticated workspace

The pilot workspace sits behind a separate app boundary for accounts, audit logs, exports, SSO planning, and data retention controls.

scoped

Analytics and CRM sync

PostHog or Plausible plus HubSpot, Salesforce, or warehouse sync connect after production tracking policy and buyer-approved destinations are defined.

Production Boundary

The public site is edge-ready. The authenticated workspace remains a governed app boundary.

Evaluators and pilot teams can inspect the operating model today without confusing static marketing surfaces for a complete production platform.

implemented Public site

Build, metadata, headers, redirects, sitemap, robots, llms.txt, and security contact

implemented Lead intake

Validation, honeypot, optional webhook token, and local preview fallback

designed Pilot workspace

Scope, answer evidence, action backlog, approval model, and proof readout

future app Production app

Authentication, roles, workspace data model, audit exports, and retention controls

scoped Enterprise procurement

Security questionnaire, subprocessors, DPA, SLA, SSO/SCIM plan, and support paths