See the work before you buy it

Sample Decision Brief

An illustrative robotics case: the recommendation, evidence, access constraints and decision logic a real engagement produces.

An illustrative robotics case: the recommendation, evidence, access constraints and decision logic a real engagement produces.

RecommendationPROCEED TO ACCESS VERIFICATION

Start with Beijing E-Town (the Yizhuang development area). Do not commit to incorporation or a physical validation visit yet.

The published capabilities justify a scoped access and feasibility verification. Proceed to a physical trial only after a host confirms technical fit, an eligible access route, engineering support, data terms, aligned availability and an acceptable budget.

Decision factor	Assessment
Fit with the engineering bottleneck	Promising, based on published capabilities
Yizhuang priority	First route to verify
Access readiness	Unconfirmed for this foreign team
Immediate commitment	Scoped access and feasibility verification
China entity decision	Defer until access options and recurring operating needs are established

This recommendation approves the next verification step. It does not yet approve incorporation, travel or a long-term China operating commitment.

The bottleneck

A Belgian robotics company headquartered in Leuven has a 14-person engineering team building mobile-manipulation prototypes and running early customer trials outside China. Its robot navigates to a workstation, identifies an object and moves it into position. Controlled demonstrations go well. Performance drops when lighting changes, objects arrive at unfamiliar angles, people interrupt the route, or surfaces differ from the training environment.

The next customer deployment needs dependable operation across that variation.

The team's question arrived as a location question: “Should we establish a China company to access the robotics ecosystem?” The immediate engineering requirement is narrower: the robot needs to fail, get diagnosed and get modified faster, across more physical variation than the team can currently produce at home.

That is an engineering-access problem before it is a company-formation problem.

From a location question to an engineering brief

Before searching for a Chinese counterpart, the location question becomes a specification a host can answer. The brief starts with a non-confidential task description, representative failure categories, current intervention frequency and typical modification lead times. That separates perception errors from grasping, navigation, integration and mechanical problems.

The project is screened for technical relevance, operating fit, credible public evidence and practical access for a small team without local staff. Routes that cannot reproduce the failure conditions, support modification and re-test, or offer a workable enquiry path fall away.

For this bottleneck, Yizhuang is the first Beijing route worth verifying. A broader city-to-city comparison would require a separate evidence set.

The evidence for Yizhuang

Beijing E-Town's account of the Beijing Humanoid Robot Innovation Center pilot verification platform (30 January 2026) describes prototype production, performance and process verification, assembly, testing and small-batch trial services. [1] For this team, that creates a concrete prospect: investigate an assembly or component failure, modify the relevant part and test the integrated machine again.

E-Town's August 2026 update describes E-Bot Hub's shared laboratories, robot hardware, technical support, laboratory booking and supply-chain prototyping services, alongside the Jiyin Gongchang platform for component selection, prototyping and production coordination. [2] For this project, a drawing-specific quotation would establish turnaround time for a required mechanical change.

Yizhuang's real-world embodied-intelligence trial programme describes public and operational test settings including shops, hotels, pharmacies, hospitals and logistics warehouses. [3] A suitable setting could expose the robot to clutter, movement and object variation beyond a controlled lab.

These sources establish relevant capabilities in Yizhuang. They do not establish confirmed suitability or availability for this robot. Verification must establish whether the relevant providers accept an externally developed mobile manipulator, whether foreign engineers can participate under the chosen access arrangement, whether the required modifications are supported, and whether the lab, modification provider and test setting can be aligned within one working schedule.

What direct verification established — and what it did not

Directly confirmed access condition — 14 September 2026: the Beijing E-Town robotics taskforce advised that the booking applicant for the E-Bot Hub shared testing platform must be a China-registered enterprise. This finding came from a separate operational enquiry, not outreach for this illustrative company or project. Confirmation record: date — 14 September 2026; counterpart function — Beijing E-Town robotics taskforce; channel — not specified in the published record; scope — the booking applicant must be a China-registered enterprise. The response does not settle whether a local partner may book for a foreign team.

That establishes an applicant requirement. It does not yet establish that:

a China-registered partner can book on behalf of this foreign engineering team;
the foreign engineers and their externally developed robot can participate under that arrangement; or
the platform is technically suitable and available for the proposed test window.

The first execution task is therefore to verify the route into the capability rather than assume one.

Open questions

E-Bot Hub access: can a suitable China-registered partner or host provide a workable booking route for this team, or would a dedicated entity be required?
Project admissibility: will the platform accept this externally developed mobile manipulator and the planned hands-on engineering work?
Coordinated availability: who coordinates the shared lab, modification provider and real-world test setting, and can their schedules be aligned within the proposed visit?
Engineering support: what technical support is available during diagnosis, modification and re-test?
Data and IP: what terms apply to confidential drawings, failure data, test outputs and ownership or permitted use of modifications?
Modification route: what quotation and turnaround time can Jiyin Gongchang provide for the anticipated mechanical change?
Alternative facility: if E-Bot Hub itself is not workable, is there a different eligible facility that can support the required test–modify–retest loop?

A different booking route into E-Bot Hub and a different facility altogether are separate fallback options and should be tested separately.

Why this route, and why now

Yizhuang is the first route to verify because the public evidence exposes the parts of a potentially useful loop: physical validation, engineering support, component or hardware modification, and real-world test settings. The value of the route depends on whether those parts can actually be connected for this team.

The timing is also practical. The relevant E-Town services are publicly described as operating, and China's current visa-free policy reduces short-visit travel friction for eligible Belgian ordinary-passport holders through 31 December 2026. [2] [4] Neither point proves availability; both make a scoped verification exercise timely.

The table below sequences complementary workstreams. It is not a ranking of equivalent locations or providers.

Sequence the work by the problem it solves

Workstream	Sequence	Purpose
Beijing E-Town / Yizhuang access	First verification	Confirm technical fit, applicant route, participation conditions and aligned availability before a physical visit.
Wider Beijing research ecosystem	Targeted parallel enquiry	If failure analysis points to a research problem in perception, control or manipulation, identify a relevant specialist and test whether the China–Belgium S&T channel provides a useful institutional route.
Jing-Jin-Ji manufacturing network	Prepare after validation	Compare fabrication, assembly and quality-control options once the design change and first batch are defined.
General incubator / operating support	Later	Consider workspace and operating support if the project develops recurring local activity.
Incentive-led location search	After technical fit	Assess financial support against an already-defined work programme and real operating commitments.

Pre-travel execution gate

A physical validation visit should be booked only after the following items are confirmed:

an eligible booking applicant and a workable participation route for the foreign engineers and robot;
technical suitability of the host and the required engineering support;
aligned dates for the shared lab, modification provider and real-world test setting;
equipment shipping and import arrangements, including battery transport where relevant;
site safety, insurance and permitted visitor activities;
written terms for confidential drawings, test data, ownership and permitted use of modifications;
an itemised trial budget and a named coordination owner.

Visa-free entry reduces short-visit administrative friction for eligible Belgian ordinary-passport holders, but the planned hands-on activities should still be confirmed against the chosen access arrangement before travel is booked. [4]

Illustrative trial budget structure

Cost line	What should be priced before travel
Host facilities	Lab days, test-space access and equipment charges
Engineering support	Technical support hours and specialist assistance
Modification / fabrication	Parts, machining, prototyping and integration work
Real-world test setting	Site access, supervision and operating support
Robot logistics	Freight, temporary import handling and battery transport where applicable
Team travel	Transport and accommodation
Coordination and documentation	Scheduling, counterpart coordination and decision record
Contingency	Rework, replacement parts and schedule slippage

No travel decision should rely on a budget line that remains material but unpriced.

The first Beijing experiment, and the decision it produces

Once the pre-travel gate is cleared, the following indicative ten-working-day plan can be finalised.

The thresholds below are illustrative planning assumptions, not validated acceptance criteria for this fictional robot. Final thresholds should be agreed against the robot's actual baseline, the customer's deployment requirement, cycle time and the available trial budget.

Define the measurement before collecting the baseline

Before baseline testing, agree what counts as:

a successful completion;
an intervention;
a safety event;
a failed attempt;
a material hardware or integration change.

Record the home baseline using those definitions. On arrival, establish a matched on-site baseline before making changes.

Indicative ten-working-day sequence

Days 1–2 — matched baseline and failure reproduction

Reproduce the main failures under agreed conditions, confirm the measurement definitions and identify the highest-impact failure modes.

Days 3–5 — controlled variation

Introduce unfamiliar layouts, object positions, lighting and surface conditions while keeping a repeatable task set. Separate perception, manipulation, navigation, integration and mechanical failure modes.

Days 6–8 — diagnose, modify and re-test

Select the highest-value changes, complete mechanical or integration modifications, and run documented modification-and-retest cycles.

Days 9–10 — held-out validation and next-stage economics

Test the modified system against held-out conditions not used for tuning. Record remaining interventions, safety events, turnaround time and the costed route for the next batch or engineering iteration.

Proposed continuation measures

Reliability: target an improvement of at least 10 percentage points over the matched baseline across an agreed pilot evidence set. A planning target of at least 100 documented matched attempts across the evaluation set can be used if robot cycle time and test conditions make it practical. This is pilot evidence, not proof of deployment reliability.
Iteration speed: complete two documented modification-and-retest cycles, with at least one hardware change ready for re-test within three working days.
Distinctive access: identify test conditions or engineering support that materially extend what the team can reproduce at its current base.
Next-stage economics: obtain a named engineering lead and an itemised next-stage route with lead times and costs that can be compared with the team's existing alternative.

Decision rule

Access gate: the team has a workable and repeatable access route with acceptable data, IP and operating terms.

Safety gate: there is no unresolved safety regression or breach of agreed operating limits. Any such event pauses testing pending review.

Engineering gate: the trial demonstrates a material engineering benefit through reliability improvement, faster iteration or a clearly actionable failure diagnosis that the team could not obtain as efficiently at home.

Delivery gate: the next stage has a named owner, defined scope, acceptable lead time and a costed delivery route.

Continue only if the access and safety gates are met, the engineering gate shows a material benefit, and the delivery gate produces an acceptable next-stage route.

Distinctive access is supporting evidence, not a substitute for engineering value.

If the technical fit looks promising but the E-Bot Hub access route fails, test a different eligible host before abandoning the wider Beijing route. If access works but the engineering gate fails, do not deepen the Beijing commitment. If the results are inconclusive, define the additional evidence required rather than treating the visit as a pass.

The entity decision comes after this evidence. If partner-based or host-based access supports the required recurring work, use it where operationally sensible. If sustained engineering, contracting, platform access or repeated activity gives a dedicated China entity a clear job to do, incorporation can then become part of the execution plan.

If the work continues beyond a first visit

The Belgian-specific legal and institutional layer becomes relevant as the operating model deepens. These frameworks do not independently justify the initial trial or establish eligibility for a particular benefit.

When each Belgian-specific layer enters the decision

Trigger	Review layer	Why it matters
Short validation visit	Visa-free business / exchange entry	Reduces travel friction for eligible Belgian ordinary-passport holders through 31 December 2026. [4]
Repeated engineer presence or recurring China activity	China–Belgium tax treaty	Review permanent-establishment exposure and the operating structure. [5]
IP licensing or intercompany payments	China–Belgium tax treaty	Review royalty and payment treatment before the structure is fixed. [5]
China subsidiary and profit repatriation	China–Belgium tax treaty + MLI	Review dividend treatment and the conditions for any reduced treaty rate. [5] [9]
Qualifying long-term investment	China–BLEU investment treaty	Review the investment-protection framework around the Belgian parent's China investment. [6] [7]
Research problem requiring specialist collaboration	China–Belgium S&T cooperation	Investigate a partner-specific research route if the engineering evidence warrants it. [8]

Selected treaty facts show why the trigger matters. Article 5 of the China–Belgium tax treaty includes services for the same or a connected project exceeding 183 days in any twelve-month period in the permanent-establishment analysis. Article 12 caps qualifying royalties at 7% of gross payments where the treaty conditions are met. Article 10 provides a 5% dividend ceiling for a qualifying corporate beneficial owner directly holding at least 25% of the paying company, and 10% in other treaty cases. Under the synthesised MLI text, the 5% dividend condition requires the ownership threshold to be met throughout a 365-day period including the payment date; the MLI also introduces a principal-purpose test. [5] [9]

A later qualifying Belgian investment would bring in the China–Belgium-Luxembourg Economic Union investment agreement, signed on 6 June 2005 and effective from 1 December 2009. MOFCOM lists it as currently effective; it replaced the 1984 agreement. [6] [7]

China and Belgium have maintained a government S&T cooperation mechanism since their 1979 agreement. The twentieth China–Belgium federal science and technology joint commission met in Beijing on 2 November 2023. If the validation work exposes a research problem, this provides an institutional background to investigate a specific partner route without assuming an open grant or company eligibility. [8]

What happens next

For this illustrative case, the immediate work is to resolve the access decision before travel.

Action	Decision it resolves	Illustrative owner
Confirm whether an eligible China-registered partner or host can book E-Bot Hub for this team	Access route	CyberNomad China
Confirm admission of the foreign engineers, robot and planned hands-on work	Project admissibility	CyberNomad China + host
Obtain lab dates, engineering-support scope and data/IP terms	Execution readiness	Host / taskforce
Obtain a drawing-specific Jiyin Gongchang quotation	Modification speed and cost	CyberNomad China + provider
Identify the real-world test setting and supervision terms	Test relevance	Host / programme contact
Align all providers into one schedule and price the trial	Go / no-go for travel	CyberNomad China + engineering lead
Finalise measurement definitions, thresholds and held-out conditions	Experiment validity	Engineering lead + host

A real engagement attaches named contacts, target dates and costs to this handover. If the access route cannot be made workable on acceptable terms, the project does not proceed to the proposed visit.

The work CyberNomad China does, and the files you receive

The bottleneck framing, evidence, confirmed/open breakdown, access verification, experiment design and decision gates show what paid project work produces when applied to a real case.

The written scope states which questions can be answered through public research and which require direct counterpart verification. A Decision Brief may include limited direct verification needed to resolve the agreed decision question. Experiment booking, delivery, on-site coordination, extensive outreach, physical testing and third-party engineering work are separate execution work unless the written scope explicitly includes them.

Decision Brief: A readable PDF with the recommendation, alternatives, evidence, decision gates and conditions for continuing.
Capability Shortlist: A comparison sheet linking each surviving option to the engineering task, public evidence and next access question.
Counterpart Record: For contacts included in scope: organisation, relevant function, conversation date, confirmed support and next-step owner.
Verification Log: A dated record separating published capability, direct answers, project-specific assumptions and unresolved questions.
Recommended Experiment: A working test plan with inputs, tasks, host responsibilities, budget assumptions, schedule, measurement definitions and continuation criteria.
Next-Action Handover: An owner-and-deadline list the engineering lead can use immediately.

Bring the engineering bottleneck

Bring the task your system must perform, where it currently fails and the milestone you need to reach. That is enough to define the first useful piece of work and the decision it should produce.

Evidence and sources

The public evidence below was reviewed during the September 2026 evidence pass. The E-Bot Hub applicant requirement was directly verified on 14 September 2026 through a separate operational enquiry after the published evidence had been reviewed; it is not stated in the original E-Bot Hub announcement.

Citations establish what is published. The route recommendation, experiment design, thresholds and decision rules are CyberNomad China's analysis of the illustrative project.

Source page: https://www.cybernomadchina.com/sample-decision-brief

Discuss your next experiment →How We Work →