One reliable system for your whole company.
Every year, 5,000+ companies across Asia-Pacific are looking for systems for their company. They usually end up with a mix of different solution providers with different pricing, capabilities, timeframes and processes. If you need an all-in-one system that mirrors your current operational workflows and habits, the options are paralyzing.
How we work.
Our method is simple. First, we understand you briefly: you, your business and your goals. With that, we gauge whether we are a good fit to work together.
Map the company, arrive at a priced plan.
Build the agreed modules, then accept and go live.
Support that starts where the build ends.
- From there, we schedule a deep analysis of your business: one core meeting with you, and a draft version of the final deliverable a few days later.
- Across the remainder of the two weeks, we discuss nuances and edge cases with you and your team to capture the reality of how your operations actually run, and deliver the final Company Blueprint, equipped with what needs to be built, how much, and how long.
- Once we arrive at a conclusion on how to move forward, the execution phase begins, followed by the longevity phase.
What we build for clients.
Deciding what you need is difficult. We have prepared simple animations to show you how each module works, followed by descriptions of what it includes, what are the use cases with reference to our previous client projects, and what to expect from different solution provider categories. We hope this helps you gauge expectations of what is possible for your system.
Our prices on this page are in RMB — the currency we contract in. For clients outside China, the equivalent is quoted in your currency at signing, so the figure you approve is the figure you pay.
The comparison tables are a reference point, not quotes: roughly what each way of buying this kind of system costs, and what that price gets you. Every vendor prices their own way — read them to know what to expect across the board. Reference prices are shown in each seller’s own currency.
Company Blueprint
Know what to fix before you pay to build.
- Without a blueprint. Nobody can see inside the process. New software is bought and bolted on where nothing was leaking, and coins keep falling out and piling up where it was.
- With a blueprint. The process is mapped end to end, the leaks are found and priced, and the biggest one is marked to fix first. You keep the map whether or not you build with us.
Before anything is priced or built, we map how your company actually runs: the flow of a job from inquiry to payment, where decisions wait, where data lives, where it leaks. The deliverable is a flowchart of your whole company, drawn with your people, with the costly gaps priced in your own numbers — and one priced recommendation for what to build first.
What’s included
- The whole-company flowchart: teams, decisions, data stores, rules
- Each costly gap priced in your own numbers — what the delay, the re-typing, the quietly lost deal costs you per year
- One priced build recommendation: scope, price, timeline — one thing, one price, one decision-maker
Everything included2 more
- Delivered in two weeks, from one core meeting with you
- You keep it either way, and it credits toward a build signed within a month
Case studyDistribution ERP for generator sets — equipment distribution, Indonesia
A distributor thirteen years in the market: ~50 staff across six departments, 29,000 customers, relationships living in WhatsApp groups and salespeople’s memories, two legacy systems that don’t talk. The blueprint ran first: ten coordination failures named one by one in the working sessions — deal tracking scattered across unsearchable chats, receivables with no automation, survey visits assigned verbally with no trail, costing in spreadsheets with no version history. Each went onto the map, and each got priced in the company’s own numbers. The priced recommendation that came out of it became the phased build that followed. The flowchart of their operation on the projects page came straight out of this blueprint.
Compare solutions4 alternatives, with typical pricing
| Vendor category | Pros | Cons | Typical pricing |
|---|---|---|---|
| Legacy ERP vendor discovery | Free, fast, and you see a mature product demo the same week | The recommendation is always their product — the map ends where their template ends | RMB 0 discovery — it’s built into the license and implementation bill |
| Subsidy-chain consultancy (China) | Subsidized price, and the paperwork doubles for government applications | You get a maturity scorecard scored against subsidy criteria, not a map of how your company actually runs | Around RMB 30,000 an enterprise where government-funded |
| Dev-shop scoping call | Quick and direct — you’re talking build price in one call | The scope rests on one call’s spec sheet; the company around it is assumed, and discovery hours fold into the build price | RMB 800–6,000/person-day, folded into the build quote |
| In-house assessment | Deepest context — nobody knows the company like your own people | The team already inside the process can’t see it from inside, and nobody’s job is to price the problems | RMB 0; the cost is a manager’s weeks |
Operations Platform
No deal goes quiet without someone knowing.
- Without the system. A deal is quoted and approved, then stalls. Days pass, nobody notices, and it quietly drops out. The director has to ask where it went.
- With the system. The same deal stalls, but after three quiet days a reminder rings and someone steps in. The deal reaches paid, and the director watched it move on one screen.
Most teams run the selling process across several tools: chats for follow-ups, spreadsheets for costing, memory for approvals. This module puts the whole ladder on one platform — prospect, quotation, approval, payment, delivery, handover — with every step writing to the same record. Reminders go out when a deal sits too long. Approvals are collected in the system where money moves. Quotations come out as branded PDFs. The director opens one screen: which clients, which stage, what is pending, what has been paid. The ladder is the same step by step for every client; the stations, rules and names inside it are molded to your operations — the case study below is one client’s ladder, built for their needs.
What’s included
- The full ladder — prospect, quotation, approval, payment, delivery, handover — on one screen
- Follow-ups and approvals are collected where your team already works — WhatsApp, WeChat, email, or the tools you use; the channel is set to your needs, not to our system
- One record per job or unit; every step writes to the same record, nothing re-typed between tools
Everything included6 more
- Dual approvals where money moves — signatures gate the next step, trails kept
- Follow-up that cannot be forgotten: no-close reminders, escalation to the head, timeline logging
- Costing engine with version history — who changed what, when, why
- Quotations as branded PDFs generated by the system, with your rules enforced (e.g., part names shown, never part numbers)
- Role views for sales, service and stock, plus the director’s one-screen view: pipeline, margins, stock, receivables
- Reminders, escalations and approval routing run on your rules automatically; your people keep every decision
Case studyDistribution ERP for generator sets — equipment distribution, Indonesia
The system that came out of the blueprint: one platform carries the entire pre-sales process, first prospect to purchase order. Lead warming runs with configurable cold-lead reminder cycles and automatic escalation when a live deal goes quiet. Dispatch stays blocked until two signatures clear — a work-order letter from the Sales Head and a technician request from the Service Manager. The costing engine keeps full version history the Director can watch mid-calculation, and quotations generate as branded PDFs that show part names, never part numbers — the anti-price-shopping rule enforced by the system instead of by memory. Close happens under the business rule “PO without payment is not a sale.”
What changes when it runs: the Director sees pipeline value, field locations and pending approvals on one screen, and the phone stays on the desk; customer relationships belong to the company instead of whoever holds the phone.
Compare solutions5 alternatives, with typical pricing
| Vendor category | Pros | Cons | Typical pricing |
|---|---|---|---|
| Legacy ERP suites | You get a ready-made ladder from prospect to invoice with the accounting already inside, and a partner in every city | You buy their template and feed it a new job desk; anything outside the template is billed by the day | License from RMB 5,000/yr; implementation RMB 30,000–150,000 (20–100 staff); customization RMB 3,000–6,000/day |
| Low-code platforms | Working forms and flows within days; your own staff can build them | It handles forms and small flows well; whole-company integration — inventory, approvals, accounting — is where it stalls, and seat fees grow with every hire | RMB 150–1,000/person/yr (a 30-person team ≈ RMB 4,500–30,000/yr) |
| SaaS CRM | Fast start; mobile apps; contacts and pipeline out of the box | Quotation to handover usually lives in another tool, so the record splits anyway; export limits apply | US$10–50/user/mo (10 seats ≈ US$1,200–6,000/yr) |
| Custom dev shops | Built to your spec, any scope | Built to the spec sheet you wrote — when the spec ages, the day-rate continues; code ownership often not written down | RMB 800–6,000/person-day; a full ladder RMB 100,000–500,000 and 2–6 months (10–50 staff) |
| In-house coders | Best access, success rate and customizations | Only if you can keep a full team of capabilities busy — and keep them | RMB 15,000–40,000/mo per developer; a usable team is 3–5 people (varies by market) |
Field Operations System
Driver, office and customer see the same day.
- Without the system. The van goes wherever seems next, crossing town twice. By midday two of five stops are done, and the customer is messaging and calling to ask where it is.
- With the system. The route is planned before the van leaves. Each stop gets a photo and a tick, and the customer watches the van approach with an arrival time until it says delivered.
The work done away from the desk is the hardest to verify from it. Today it usually lives in a group chat: delivery photos, addresses typed each morning, attendance confirmed by message. This module gives the same work a record instead — the driver checks in from their phone, the delivery photo attaches to the order with its timestamp, and the route is drawn once on a map and adjusted by dragging, not re-typed. Check-ins and photos save on the phone and upload when signal returns, so the record doesn’t depend on the network — and the same record works for any deskless team: drivers, technicians, field crews, farm rounds. The office watches progress on one screen, customers see their own delivery status, and the group chat goes back to being for conversation.
What’s included
- Mobile job forms: check-ins with GPS, photo proof-of-delivery, timestamps — an audit trail on every stop
- Check-ins and photos save offline and upload when signal returns — the record doesn’t depend on the network
- Dispatch with preview: geocoding, area clustering, stop sequencing and ETAs, reviewed on a map and adjusted by dragging before confirming
Everything included5 more
- Evening-before driver check-in with a cutoff time and automatic no-show flagging
- Live fleet progress tracker — reusing the GPS hardware you already have, where it exists
- Proof-of-delivery that cannot be skipped: no delivery closes without its record
- Customer-facing status off the same record — a shareable tracking link, no phone calls
- Standby rotations and anonymous job assignment where fairness matters
Case studyCold-chain last-mile delivery ERP — refrigerated B2B logistics, Indonesia
One WhatsApp group carried driver assignment, attendance, proof-of-delivery photos and customer questions — finding one delivery photo meant scrolling months of chat. Routes were typed into a maps app address by address every morning; a delivery-morning no-show meant phone calls to reserve drivers; GPS trackers were already fitted to every bike, but their data sat stranded on the vendor’s separate dashboard.
Now: automatic evening-before check-ins with no-show flagging, one-tap standby activation from a two-driver rotation, and a route engine that geocodes, clusters and sequences the day — the dispatcher reviews it on a map and drag-overrides before confirming. A mandatory freezer-box checklist gates Start Route; the existing GPS polls onto one unified map every 30 seconds; no delivery closes without a photo, auto-stamped with timestamp, GPS accuracy, order ID and driver ID; customers track through a login-free shareable link.
What changes when it runs: one dashboard replaces the group chat, the maps app and the vendor’s GPS portal, and B2B customers place orders from their own ERP systems through the inbound API — which is what enterprise accounts usually require before they sign.
Compare solutions4 alternatives, with typical pricing
| Vendor category | Pros | Cons | Typical pricing |
|---|---|---|---|
| Field-service SaaS | Dispatch, check-ins and job forms work out of the box | It stops at the job — the order, the quotation and the money live in another system | US$29–70/seat/mo (12 drivers ≈ US$4,200–10,000/yr) |
| Fleet IoT / telematics | Vehicles and temperature tracked continuously | It shows where the vehicle is, not what was delivered or what is owed; one more portal for the team to open | Hardware RMB 300–800/vehicle + RMB 30–100/vehicle/mo (12 bikes ≈ RMB 8,000–24,000 in year one; ≈ RMB 4,300–14,400/yr after) |
| Driver-app development | Exactly the app you spec | The app is one project; connecting it to orders and billing is the next one; aftercare is priced again after the build invoice | RMB 800–6,000/person-day; a standalone app RMB 80,000–300,000 and 2–4 months |
| Group chat + paper | Free and familiar; zero training | No search and no trail — one delivery photo sits months deep; the morning plan lives in one person’s typing; a no-show is discovered by phone call | RMB 0 license; the cost is staff time |
Production-Floor Execution
The office sees the floor as it happens.
- Without the system. The floor counts, someone walks the number over to the office, and the office is always behind. A faulty box travels all the way to shipping without anyone noticing.
- With the system. Every station scans as it works, so the office and floor counters move together. The faulty box is caught at assembly, held, reworked and sent back to the line.
What the ERP says and what the floor did are usually two different stories: the office records the order, the floor runs on whiteboards, and someone re-types between them. This module closes that gap — work orders, batches and station reporting on one record with the ERP, quality checks with a pass/fail per unit, consumption and machine data where the devices allow. Month one on your floor: five records, one screen, a scanner at line side — and a batch hold you can actually run: hold the lot, record the reason, dispose it (rework or scrap), re-inspect, release. Right-sized: when the floor needs five records and one screen, it gets five records and one screen, not a six-figure MES program.
What’s included
- Work orders, batches and station-level reporting on the same record as the ERP
- Quality checks with pass/fail records and traceability per unit or batch
- Consumption and machine data, where the devices allow
Everything included2 more
- The floor record flows into the same dashboards the office reads — no re-typing between worlds
- Right-sized by design: station-level scope, not a plant-wide MES program
Case studyIndustrial-group digitalization — biochemical group, China (designed)
A group operating 24 facilities on manual processes: labor-heavy operations where digitalization has obvious, defensible value and no internal team to build it. The program is designed as a priced single-site pilot — real scope, one facility, one deliverable — with per-facility staffing studies (where hours go today, what the system takes over, and what each worker’s day looks like after), process mapping, and compliance digitalized on the same pattern each time. The savings case is modeled so it lands on one facility’s books, in numbers an owner can check against their own payroll, before any rollout decision; the modeled 24-facility rollout projects a high seven-figure RMB amount per year across the group. It is a model, not a field result — which is exactly what the pilot is for.
Compare solutions5 alternatives, with typical pricing
| Vendor category | Pros | Cons | Typical pricing |
|---|---|---|---|
| Cloud MES SaaS | The floor is configured in days; low commitment to start | It’s the floor only — your orders, quotations and money are not in it, and the connection to the office is on you | From RMB 10,000/yr for a single line; most bill per user — a 10–30-person floor often runs several times that |
| Low-code reporting apps (简道云-class) | Your own team can build forms and reports in days | Works until the floor outgrows forms — no quality trail, no machine data, no ERP record | ≈ RMB 168/person/yr plus build time |
| Enterprise MES program | Deep manufacturing suite with quality and compliance depth, built for multi-plant scale | Months of implementation before the first line reports, and it’s built for the plant, not the company around it | Six figures RMB before the first line reports (100+ staff plants) |
| Machine-vendor consoles | Bundled with the equipment; no separate purchase to approve | Locked to their machines — data from other brands’ equipment stays out, and so does the office | No separate line item — it’s inside the machine quote |
| Custom dev shops | Built to your floor’s actual stations and rules | The floor system becomes its own project; connecting it to the office ERP is priced again | RMB 800–6,000/person-day; a single-line system ≈ RMB 80,000–300,000 |
Manual-Work Automation
The copying happens while you sleep.
- Without automation. Copying listings from four websites by hand fills Monday, Tuesday and part of Wednesday. In our client’s case, about ten hours a week.
- With automation. On Sunday night the system collects and sorts everything from all four sources. A ready report is waiting first thing Monday, and the week goes to real work instead.
Someone on the team collects the same information every week: boards watched, listings re-typed, spreadsheets re-formatted. This module takes that round over — scheduled collection across the sources, duplicates merged, names standardized, everything classified to your rules and delivered as a dashboard, digest or email, whichever your team actually reads. When a source changes and the collection breaks, the system flags it the same day. Sources are chosen with their rules in view — official APIs where boards offer them, rate limits respected; if a source forbids collection, you hear it before we build, not after. The team starts the week at the judgment work, not the collecting.
What’s included
- Scheduled collection across portals, boards and systems — all sources on one clock
- Dedup, name standardization, classification and formatting to your rules
- Delivery as dashboard, digest or email — whichever your team actually reads
Everything included4 more
- Searchable tags and filters on everything collected
- Breakage monitoring, flagged the same day it happens — repaired under the support plan (module 07)
- Manual on-demand trigger for the odd midweek need
- No licenses and no scripts to own — the work lives inside the system you do
Case studyInternship listing automation — student placement, Singapore
An internship-placement firm serving international students was growing its student base without growing headcount at the same rate — the bottleneck was ten hours per recruiter per week spent collecting before placement work could begin. Every week the team opened four job boards by hand, copied listings into spreadsheets, deduplicated manually, standardized company names by eye and classified industries from scratch; data went stale between checks, and good opportunities expired unseen.
Now a scheduled aggregation service runs Sunday-night scrapes across all four sources at once: deduplication is automatic with per-source tracking, company names are standardized, industries classified, skills extracted into searchable tags. The team works from a dashboard filtered by industry, skill, date and source, with one-click CSV export and automatic weekly email digests; listings expire after four weeks so the database stays current, and if a scrape breaks or returns nothing, the gap is flagged the same day.
What changes when it runs: ten hours per recruiter per week redirected from collecting to placing — the same team places more students without new hires, and adding sources or students scales without adding human effort.
Compare solutions4 alternatives, with typical pricing
| Vendor category | Pros | Cons | Typical pricing |
|---|---|---|---|
| RPA platforms | Bots click through the tools you already own; a pilot is fast | When a website changes, the glue breaks silently — and someone on your side has to notice and fix it; per user and per bot monthly | ≈ RMB 15,000–60,000/yr for a small team |
| BPO service | People do the manual work cheaper than your staff; headcount flexes | Errors scale with headcount and quality varies shift to shift — the work is still manual, just not yours | ≈ RMB 25–60/hour per head |
| Freelancers | Best value for money if you find a golden goose | Most go like: takes payment, goes silent, no source code, no docs, no handover — and the script dies quietly when a website changes | RMB 3,000–20,000 one-off per script |
| In-house coders | Best access and customization; the script is yours from day one | A full-time developer for one internal collection round is hard to keep busy — and the logic leaves when they do | RMB 15,000–40,000/mo per developer |
Intelligence Layer
Know where you stand without asking anyone.
- Without the system. The director’s question is passed from person to person while three days go by, and the answer comes back as a report that is already old.
- With the system. One live screen answers it at a glance. The screen also flags a deal that has gone quiet for seven days and prepares a draft follow-up for the director to approve.
A system that runs your daily work produces data as it goes: every order, chat, reading and payment. Because every module writes to the same record, this layer needs no new data: it watches the record as it grows and turns it into the director’s view — the state of the business on one screen, alerts when something sits too long, repricing where rules say so, and AI drafts prepared from your own record: a quotation drafted for approval, a summary, a classification. The system brings the data and prepares the drafts; you stay on what needs you. There is no separate BI license and no modeling project, because the numbers come from the same record the work happens in.
What’s included
- Director dashboards on live operational data — every number traceable to the record behind it
- Alerts and escalation rules — no-close alerts, overdue shipment flags, quiet-deal escalations
- Repricing and price-rule engines where the work demands them
Everything included4 more
- AI drafts from your own record — quotations, summaries, classifications — approved by your people
- Quarterly health reports fed by the same numbers, not a separate report deck
- No separate BI license and no modeling project — the data was already there
- New metrics added as the system grows; the layer is part of the build, not a bolt-on
Case studyInternship listing automation — student placement, Singapore
The same system, seen from the intelligence end. Nothing about it is a BI project: the data is produced by the work itself — every scrape, every classification, every listing that ages out — and the layer reads that stream as it arrives. The dashboard gives the team the week’s placements market filtered by industry, skill, date and source; the weekly digest is the alert system, landing automatically without anyone compiling it; and when a source breaks or returns nothing, the gap is flagged the same day and repaired under the support plan — not a silent empty chart. Ten hours per recruiter per week is returned because the intelligence is built into the same record the work happens in, not bolted on after: the numbers come from the work, the alerts come from the numbers, and nobody maintains a separate reporting stack.
Compare solutions4 alternatives, with typical pricing
| Vendor category | Pros | Cons | Typical pricing |
|---|---|---|---|
| BI suites | Powerful dashboards on any data source once it is modeled; deep drill-downs | The data must be captured and modeled first — the modeling is a separate project before the first chart, and a specialist maintains it | RMB 50,000–150,000/yr + implementation (50–200 staff) |
| SaaS dashboards | Ready on day one, nothing to set up | Counts only what that product sees — orders, chats and machine data outside it stay invisible | RMB 0 extra; included in the host subscription (typ. RMB 100–1,000/user/yr) |
| Spreadsheet reporting | Total flexibility; everyone can read it | Breaks silently — a wrong formula reaches the meeting before anyone notices it, and the report leaves with the person who built it | RMB 0 license; 2–8 staff-hours a week at 20–50 staff |
| AI analytics copilots | Ask in plain language, get summaries | Only as good as the data underneath — on scattered records it produces confident wrong answers | RMB 20–200/user/mo on top of the base product |
Maintenance & Support
Still running a year after handover.
- Without support. Two months after go-live something breaks. The phone rings, nobody answers, and the system goes flat.
- With support. The same thing breaks, and an engineer answers within four working hours. Every quarter gets a check-up, a second break is handled the same way, and the system is still running at twelve months.
A system is not finished at go-live; it is handed over. This module keeps it alive: a signed repair clock with a service credit behind it, eight hours of small changes every quarter, and a restore test logged in the health report. Signed before you go live — never promised after. Stop paying and the system pauses to read-only: it never crashes, and your data stays yours.
What’s included
- 10% of build value a year, RMB 15,000 floor (on small builds the floor binds) — annual, or quarterly at +5%
- Coverage from go-live day one; billing starts month two — the first month is part of the build
- The repair clock: priority-one issues get a response within four working hours (Haining time; coverage window stated in the agreement), a fix started the next day — started isn’t finished, so the quarterly health report logs how long repairs take. Miss the clock twice in a quarter and the next month is free
Everything included3 more
- Eight hours of small changes a quarter; anything bigger is a written, priced change order
- A named engineer who answers where your team works — WhatsApp and WeChat included — and a quarterly health report with a logged restore test
- Stop paying and the system pauses to read-only — it never crashes, and your data stays yours: exportable in full, in an open, documented format, whenever you ask
Case studyDistribution ERP for generator sets — equipment distribution, Indonesia
The maintenance agreement for this system was signed before go-live, as it is on every engagement, and the system has been running since. The rhythm is the same every quarter: eight hours of small changes, a health report with a logged restore test, a named engineer who knows the build. The 24-week phased program ended with annual infrastructure and support after handover — the support phase didn’t end with the project; it started where the build ended. The proof of a maintenance contract is the renewal, and the health report that lands each quarter.
Compare solutions4 alternatives, with typical pricing
| Vendor category | Pros | Cons | Typical pricing |
|---|---|---|---|
| SaaS vendor support | The product’s own team; updates included | Support covers the product, not your workflows — questions about your customization land back on you | 15–22% of license cost a year |
| Dev shop aftercare | The builders know the system best | “Free maintenance for 18 months” ends — and by then the developers who knew it have rotated to other projects; after that it’s hourly | Free 12–18 months, then RMB 800–6,000/day or a monthly retainer |
| Ad-hoc hourly support | You pay only when something breaks | Nothing signed, nobody accountable; response time is whoever happens to be free | ≈ RMB 300–1,000/hour |
| Doing it yourself | Total control, no fee | The engineer who can help is you — and the restore test that never happens stays undiscovered until the day you need it | RMB 0; the cost is downtime |
How to know what you need.
See how our past clients worked with us — where they started, what they decided, and what it took — and you will know what to expect for your own system.
FAQ
Why can’t I just use AI to build it?
Short answer: You can, we use AI heavily in our own builds. Our operational efficiency becomes your cost efficiency so we can focus on delivering outcomes instead of payable hours. But the devil’s in the details.
AI is like a machine in your plant. Producing your final product takes multiple machines on a production line or cycle, plus your team’s operational expertise to deliver product quality, consistency and after sales service. What makes it difficult for you to just use AI to build, is the same thing that makes it difficult for anyone else to just use your production setup to deliver what you deliver.
Why can’t I use an off-the-shelf ERP — or our own coders and freelancers?
Short answer: You can. It all depends on what outcome you want to achieve.
Here’s a simple comparison of what you get:
| Legacy solution providers | You buy a solution that gives you a new job desk, and charges you up to RMB 6,000/day development cost for customization |
|---|---|
| Other custom software/systems companies | You bring a fixed spec sheet for one industry or use-case, give up ownership of the system, skip flexible financing, and accept maintenance that often runs 15–20% a year, billed by hours or days instead of outcomes |
| In-house coders | Best access, success rate and customizations — if you have the budget for a full team of capabilities and relevant dependencies |
| Freelancers | Best value for money if you find a golden goose. Most go like: takes payment, goes silent, no source code, no docs, no handover |
MJCX Studio serves clients who fit exactly this profile:
- Needs 1 custom system to cover different workflows, systems, operation habits
- Needs clarity in what to build and ongoing support to ensure the system grows with the business
- Needs system ownership, favorable financing terms, efficient build and maintenance costs
What happens if we have a dispute or legal issue?
Short answer: Prevention is better than cure — agreements and contracts that provide clarity and expectations management, instead of “just a good deal to get started.”
Most disputes come down to six things: payment terms, delivery obligations, IP and data ownership, handover, after-sales service, and scope changes. When you work with us, we jointly sign eight-plus agreements — from NDA to maintenance annex — following the milestones and payment terms we agree on. We also provide quarterly health reports and usage statistics to help you understand your business more and how it is growing or needs improvement. This way there is always a paper trail that makes disputes clear and simple to resolve.
You’re a new company — can you deliver, and will you be here next year?
Short answer: Yes — on both counts.
Our team has worked at Huawei, Tencent, A-Star, CP Group and SCG, with backgrounds from Zhejiang University & National University of Singapore. We have also worked with government agencies and Chinese state-owned enterprises through our previous climate tech startup. For our recent deliveries, please refer to the Projects page or talk to us about your specific project needs.
We put our founders’ reputation on the line for everything we work on together because we want to build the future of how SMEs operate, together with clients like you. We believe AI has created a larger and more willing market, because clients can visualize and understand the benefits of a system better, as well as making the cost of scoping, building and maintaining a custom system more achievable. And because we stay small on purpose, we choose clients as carefully as they choose us — every engagement is someone who believed in us early, and we treat it that way.
What happens after go-live?
Short answer: Everything is signed and confirmed before the start of the build. You can choose to run the system yourself or with us.
If you decide to continue working with us, maintenance is 10% of the build value a year (RMB 15,000 floor), signed before you go live. Coverage starts the day you launch; billing starts month two — the first month is part of the build. Every quarter includes eight hours of small changes, a health report with a logged restore test, and the repair clock: priority-one issues get a response within four working hours and a fix started the next day — miss the clock twice in a quarter and the next month is free. Or run it yourself: full IP transfers at final payment, documentation transfers with the system, your data was always yours and stays exportable. And if payments ever lapse, the system pauses to read-only until the account settles — your data is never touched.
Start with the blueprint.
RMB 20,000, two weeks, kept either way — and it credits toward a build signed within a month.