EN中
Applied AI & Automation_

Manual workflows,
retired.

Custom AI agents that take the repetitive half of operations off your team, with a person approving every step that matters. Our first build paid for itself in under two months.

Start a projectView work
Scroll to explore
$1M+value created for clients
$50K+/morevenue one deployment brings its client
<2 monthsto break even
Platform in motion

Armature at work

An animated example that cycles through four workflows. Finance: invoices are read, matched against the bank and posted to the ledger. Recruiting: applicants are scored, ranked and contacted, and replies become interviews. Legal: a contract is read, its key terms and deadlines extracted, and an unusual clause sent to a lawyer. Research: overnight sources are cross-checked and written into a morning brief. Each waits for a person to approve before it goes ahead.

Supplier invoice reconciliation

Reads the invoices, matches them against the bank, drafts the vouchers. Nothing posts to the ledger until a person approves.

Work

Three builds, in full

Recruiting · 2025, rebuilt on Armature in 2026

Manhunt

Recruiting and outreach automation: candidate discovery, screening and multi-platform outreach, from sourcing to a formatted submission.

The challenge

Recruiters spent hours scrolling LinkedIn, copying profiles and formatting submissions. Manual work could not scale.

What we built

End-to-end automation from discovery to delivery: candidate summaries, revision, record tracking and company-specific formatting, while client relationships stay human.

Manhunt 3.0 run report: 47 candidates processed, rated A or C with the agent's reasoningManhunt 3.0 run report: 47 candidates processed, rated A or C with the agent's reasoning
Manhunt 3.0 on Armature: one sourcing run's report, content area only. Demo data; no real candidates.
$50K+/mo
client revenue
<2 months
to break even
100+
candidates / minute
70%
fully automated

Now in its third generation with the same users: Manhunt 3.0 runs on Armature, with 19,742 records carried over.

Internal tooling · since 2026 · grew out of Zenith (2024)

AVIS

operatorAVISlive oversightticketed dispatchworkspaceagentworkspaceagentworkspaceagenta2aacross machines

Our own command cockpit on Armature: one operator seat supervising a fleet of AI agents, with live oversight, ticketed dispatch into isolated workspaces and coordination across machines.

Internal. Runs Tirmenal's own engineering.

Consumer product · since 2026

Prism

PrismPrism

A life-admin assistant on the same Armature core: it triages email and calendar, keeps what you save to read, and answers from everything it has kept. One installer, English or Chinese at runtime.

In development. Local-first, and nothing changes without your approval.

See all projects
Platform

One platform under every deployment

Armature is what each client deployment runs on, so a new one starts from a working base instead of from zero, and gets better as the platform does.

a person approves every step that matters

01 / 04

Your operations

What the agents work from. Armature connects to the tools you already run; nothing has to be replaced first.

In practice

JAMOL: the firm's existing document folders and client accounts.

Approach

How an engagement runs

  1. 01

    Diagnosis

    Free · 30 minutes

    We pick the workflow that costs your team the most and agree on the one number that defines success.

  2. 02

    Pilot

    Fixed scope · fixed price

    One workflow, end to end, on Armature, quoted after the diagnosis. It goes further only if the number moves.

  3. 03

    Run

    Ongoing

    We deploy it, maintain it and keep improving it. Your team keeps the approvals; we stay on call.

Your data

  • Every consequential action waits for a person's approval, and every step is logged.
  • Deployed in your environment or hosted by us. Your choice.
  • Your data is never used to train general models.
  • Sensitive work can run on-premise or in a private deployment.
Who we are

The people behind it

Commerce × Cross-borderCommercialisation and enterprise delivery

Alex Xie

Co-Founder

  • MS Finance (STEM), Simon Business School, University of Rochester
  • Equity research VP at China Securities
  • Investment analyst at Fosun Capital
  • Founded and ran a cross-border trading company, taking overseas markets from zero to one
China SecuritiesFosun CapitalUniversity of Rochester
Engineering × AI infrastructureProduct engineering and delivery

Edward Han

Co-Founder

  • Senior engineer, Scotiabank
  • Frontend lead for an AI infrastructure platform at inference.ai — GPU and CPU cluster orchestration, multi-tenant inference
  • That console carried 1,000+ inference instances for 100+ teams at 99.9% uptime
  • Engineer at Huawei
Scotiabankinference.aiHuaweiUniversity of Toronto
Finance × AIPlatform architecture and delivery

Henry Zhang

Co-Founder

  • BCom, Rotman School of Management, University of Toronto — finance and economics
  • MS Finance (STEM), Baruch College, City University of New York
  • Quantitative pricing analyst at S&P Global
  • Founded Tirmenal and built its enterprise agent platform in-house
S&P GlobalUniversity of TorontoBaruch College

Start a project.

Tell us about your operations.A bit of context helps us respond faster.

We reply within one business day.