01 · Validate
Paid Consulting
$70–100 / hr
45–60 min three-way call (us + you + your ops lead). 10 questions in, you'll know if it's automatable, roughly how much, how long.
Sydney · AI Automation Studio
Booking May/Jun
Enterprise-grade AI workflows,
from $150K → $3K.
The same AI automation big consultancies charge $150K+ for, we deliver from $3,000. Not because we do less — because we reuse proven templates and only swap config per client.
// runtime: edge agent v2.4 · trace_id: 7f3a..b21
● all systems operational
01 · Ingest
Gmail · WeCom · Lark · Webhooks
events/min: 142 · p99 12ms→
02 · Reason
GPT-5.4 · Tool-calling · RAG
context: 200K tok · tools: 48→
03 · Act
Lark Base · Telegram · Procore
✓ 3,210 actions / 24h$3K+
Per automation, starting
2wks
MVP to working v1
7+
Live / in-progress projects
50:1
Cost vs. big consultancies
// Built on
OpenAI GPT-5.4
Anthropic Claude Sonnet 4.5
Lark Base / 飞书
Telegram API
WeCom · 企业微信
PostgreSQL
FastAPI
Procore API
// Cost calculator
See your number, in 5 seconds
Drag the sliders. Compare big-consultancy quote vs. ours. The savings number updates live.
Your situation
Move the sliders to match your business.
Net advantage
$191,600 / 12 mo
// Services
Pick the right entry point for your stage
Not open-ended consulting. Each tier has a fixed price, fixed scope, fixed timeline — so you can calculate ROI before signing.
02 · Most picked
Automation Project
from $3,000
One concrete workflow end-to-end: invoice follow-up, customer reminders, email field extraction, daily reports… working v1 in 2 weeks.
03 · Replace Excel
System Build
from $5,000
A lightweight ERP / CRM on Lark Base or similar — orders, customers, inventory, team collab in one place. Same template reuse, 4–6 weeks.
04 · Don't disappear
Ongoing Support
Monthly
Hourly or monthly retainer — version upgrades, bug fixes, small new requests. Make the system better the longer you use it.
// Why us
Big consultancies charge 50× more, ship 6× slower
// 01
Big-tech engineering, SME pricing
Our engineers come from big tech and we have a proprietary agent framework. We're not building from zero, so you don't pay the 'exploration' tax.
$150K+
→
$3K
Equivalent AI Agent project
// 02
Working v1 in 2 weeks
No 6-month roadmaps. Week 1 we get it running, week 2 you and your staff use it. If it doesn't work we refund the deposit.
3-6 mo
→
2 wk
vs. big consultancy timeline
// 03
Only what you actually need
No 'AI strategy decks' you can't action. We find your most expensive manual workflow, automate just that one, then reinvest the savings into the next.
50+ features
→
3-5 features
Only what you'll actually use
// Cases
Real projects already running
Six different industries, one underlying template. The featured one rotates — click any tab to see another.
CUSTOM FURNITURE
Lark Base ERP + AI Copilot
13 Lark Base tables + 1 chat bot + Gmail auto-sync, stitching quote-to-after-sales into one chain. Replaces three spreadsheets, two WeChat groups, and a notebook the boss carried for 8 years.
"Quotes used to take half a day. Now my apprentice does it in 20 minutes — and they're cleaner than what I used to send." — Owner, Sydney custom furniture (anonymised)
70%
Quote turnaround
12h/wk
Saved per owner
3,410
Lines of code
CUSTOM FURNITURE
Lark Base ERP + AI Copilot
13 Lark Base tables + 1 chat bot + Gmail auto-sync.
70%
Quote turnaround
12h/wk
Saved per owner
PLUMBER · ON-SITE
Telegram Bot + Quoting Brain
Runs on the plumber's Mac mini: AI queries an Obsidian KB to draft quotes, archives photos, logs worklogs.
95%
Faster quote drafting
100%
Photos auto-archived
CONSTRUCTION PM
Construction PM Workbench
WeCom site messages auto-structured + email contract review + change management + Procore.
60%
Faster contract review
8h/wk
PM admin saved
MULTI-STORE E-COM
Podium Lead Attribution & Routing
Sydney/Melbourne/Perth Podium leads — attribution + auto-routing in seconds.
95%
Attribution accuracy
24h→sec
Routing latency
USED CAR DEALER
Used-Car Sourcing AI Assistant
Desktop assistant: organises listings, runs AI valuation + risk flags, daily bilingual brief.
3h→5min
Daily screening
60h/mo
Saved per owner
MOD SHOP CHAIN
WeChat → AI → Multi-platform
Photos in WeChat → AI multilingual copy → IG/FB/TikTok/Xiaohongshu Overseas with one approval.
5×
Content output
2h→15min
Per-post manual time
// Testimonials
What owners say after 30 days of running it
Names anonymised at request. Industries and quotes are real, edited for clarity.
They didn't try to sell me an 'AI strategy'. They asked which thing on my desk was eating my time, automated that one piece in 9 days, and only billed me when it ran.
DT
D. T.Custom furniture · Sydney
My phone used to buzz 80 times a day with site photos. Now they get filed, captioned, and tied to the job number before I even open the chat.
RH
R. H.Construction PM · Brisbane
I quoted three other studios. The cheapest was $58K and 14 weeks. ManifoldX delivered v1 in 11 days for under $4K. I'm still suspicious of how cheap it was.
JL
J. L.Used-car dealer · Melbourne
// Industries
The kinds of businesses we work with
Clients in the same domain share most of our process templates, so onboarding the next one has near-zero marginal cost.
Construction & Trades
Site records, contracts & variations, job tracking, takeoff & quoting
Manufacturing
End-to-end ERP from quote to after-sales — orders, production, inventory, install
Property & Real Estate
Client CRM, viewing & maintenance follow-up, document triage, multi-branch coord
Professional Services
Auto document classification, reminders, archiving, file collection
Retail & E-commerce
Multi-store lead routing, content production, inventory alerts, social publishing
Wholesale & Trade
Supplier reconciliation, document archiving, multilingual comms, stock & logistics
// Process
3 steps to live — pay only for things that ship
-
01
Free 30-min assessment
WeChat or video call. 10 questions about your business. By the end we'll tell you if it's automatable, which tier fits, and rough cost.
-
02
2-week working v1
Adapt from existing templates, not scratch. Week 1 it runs. Week 2 you and your staff use it and we tune. You only pay a 30% deposit.
-
03
Deploy on your machine, monthly maintenance
Runs on your own machine / server — your data never leaves. Monthly retainer covers updates, bug fixes, small new requests. Cancel anytime, no lock-in.
// About us
A small team that ships, not a deck factory
Four people: one product/customer lead, two engineers, one for marketing. Small on purpose — no account managers between you and the people writing the code.
B
// Product / Customer Lead
Banruo
Spends every week with SME owners — on showroom floors, in workshops, on plumbers’ vans — figuring out which manual tasks are quietly killing margins. Translates that into product specs the engineers can build against.
Product Research
Customer Success
Bilingual
T
// Engineer
Tony
Builds the agent and ML side — model selection, tool wiring, fallback paths, monitoring. Cares about how a workflow runs in production, not how it looks in a slide.
AI Agents
ML Infra
Agentic Ops
D
// Engineer
Dylan
Builds the agent layer + LLM integration + data ingestion. Specialises in turning messy real-world inputs (chat history, emails, photos) into clean structured data.
Agent Dev
LLM
Data Pipelines
Y
// Marketing Lead
Yannis
Handles positioning, channels, content, and the AU / CN bridge — so the work the engineers ship actually reaches the owners who need it.
GTM Strategy
Growth
AU · CN
4
People on the team
2 wks
To working v1
0
Layers between you & code
AU · CN
Bilingual delivery
// FAQ
The 6 questions owners ask most
Our processes are messy and we have no data — can AI even help?
That's the perfect starting point. Phase 1 is auto-collecting 'information scattered across WeChat, email, Excel' into structured data. Phase 2 automates off it.
We're a 5-person team — is it too early for AI?
5 people is exactly when it makes sense. The owner is IT, sales, support and accounting — automation buys you back the most leverage. We have a 3-person client who saved the cost of the next hire.
How is data safety handled?
Default deploy on your own machine / server — data never leaves your company. AI calls are per-request, no full DB upload. For higher compliance we can run locally-hosted open-source models.
What if I want to maintain it myself or switch studios later?
No. Code, config, and DB stay with you. All docs in plain Chinese for non-technical readers. Monthly support is month-to-month. No lock-in.
Will the '$3K starting' project balloon to $30K?
Fixed-scope fixed-price. After the first call you get a written quote with scope spelled out (in/out). Once signed, that price is locked. New asks get quoted as follow-on projects.
We're not in Sydney — can you still work with us?
Yes. Most delivery is remote. Melbourne / Brisbane / Mainland China are fine. Far time zones get a small comm-cost adjustment in the quote.
// Contact
30 min, and you'll know if it's worth doing
No consultation fee. If it's not a fit I'll tell you why — could save you tens of thousands of being upsold elsewhere.
Location
Sydney, AU
Remote-first across AU & 中国大陆
Booking May / Jun 2026 · 2 slots left