Features

Operating procedures, runs, evidence. All in one HQ.

Four capabilities that together turn a folder of prompts into a governable operations layer. Each one is useful on its own. Together provide the whole picture.

01 · Define

Procedures in a central UI, not a prompt.

Ordered steps. Defined input and output. Expected outcomes. Defined in a clean editor that a domain expert can use — not a system prompt.

  • Version history with diff view — every edit traceable
  • Authored in the Hque UI, or via our Platform-MCP - perfect for authoring your existing procedures
  • Reusable procedures across all staff members/loactions/branches
Procedure: new patient intake
v 1.4 · draft
01
Capture patient details
name string, dob date
agent
02
Verify medical history
allergies string[], conditions string[]
agent
03
Confirm consent & privacy
consent_signed boolean
human
04
Book first appointment
slot datetime
agent
4 steps · 2 schemas edited 12m ago
Connected runners
10 active
ElevenLabs voice agent
live
Zapier (booking trigger)
live
n8n (Stripe payment)
live
Slack bot
live
Custom agent (Claude SDK)
live
Que (in-product)
live
Same MCP. Same procedures. mcp.hque.ai
02 · Run

Any agent. Any platform. Same procedure.

Hque is deliberately neutral about how agents connect. Anything that can call an MCP server or a REST API can run a Hque procedure — and they all run the same one.

  • MCP-native — the standard agent-tool interface, day one
  • Native integrations for Zapier, n8n, Make, ElevenLabs, Slack, Teams
  • Voice agents load a procedure and run the conversation as the procedure
  • Change your agent layer without changing your system of record
03 · Capture

Structured outputs, not text transcripts.

Every step has a defined output schema. When the agent reports back, Hque captures the data — not a blob of free-form text that the next step has to re-parse.

  • Typed fields: strings, numbers, dates, enums, references, files
  • Validation at capture time — bad data never reaches the next step
  • Queryable across runs — "show me every intake with a penicillin allergy"
  • Evidence attachments — audio, photos, signatures, documents
run #2841 · structured output
14:03:41 AEST
{
"patient": {
"name": "Sarah Mitchell",
"dob": "1987-04-12",
"allergies": ["penicillin"],
"conditions": []
},
"consent_signed": true,
"slot": "2026-05-22T09:30:00+10:00",
"evidence": ["audio/intake.mp3"]
}
Captured by agent schema v 1.4
Audit trail · run #2841
tamper-evident
14:03:12 step 01 captured sha:9f3a…
14:03:19 step 02 captured sha:c14e…
14:03:24 consent signed sha:5b71…
14:03:41 slot booked sha:b7e2…
Azure SQL Ledger · hash chain verified
04 · Prove

Audit-ready by default.

Every procedure run is recorded in a tamper-evident hash chain backed by Azure SQL Ledger. Not because every customer asks for it — but because once you've built deterministic execution this way, it's free.

  • Cryptographic hash of each step output, chained per run
  • Backed by Azure SQL Ledger — tamper-evidence at the database level
  • Export evidence packs for audits, incidents, regulators
  • Versioned procedure definition stored alongside the run

One procedure. Any agent.
See the difference.

The fastest way to see the power of Hque is to try it! Each account comes with a built in Playground - point an agent at it, and see what gets captured.