Last week the client sent over screenshots of three declaration forms and asked which category to report under.
Benchmark, star, fist. He thought he had to choose one of the three, and even asked which one had a higher pass rate.
I turned to Article 6 of the application guidelines and showed it to him: graded assessment, with separate selection according to the standards for three categories of products; the same product can be recognized multiple times, and can simultaneously be a benchmark, star, and flagship product.
He froze for two seconds, then asked whether that meant writing three sets of materials. Yes. And the three sets of materials could not just be the same document with a different title.
First, look at the nature of this document
The Guangzhou Municipal Bureau of Industry and Information Technology's 'Notice on Building an Intelligent Product Reserve Pool' was issued on November 3, 2026. The application system opened at 10:00 on November 4 and will remain open long-term, with the entire process online and no paper materials required. On November 10, the Baiyun District Bureau of Science, Industry and Information Technology forwarded the municipal notice, and district-level mobilization began.
The background is in the first paragraph of the notice: under the guidance of the Municipal Bureau of Industry and Information Technology, Guangzhou City Investment Group, together with the Fifth Electronics Research Institute of the Ministry of Industry and Information Technology, built the country's first all-space intelligent experience center in Haixinsha Asian Games Park. The reserve pool is for dynamically reserving exhibits for this center—"high-quality intelligent products with leading technology, active market performance, outstanding output value, and strong demonstration and driving effects."
The four adjectives in this sentence correspond to the definitions of the three categories of products that follow. "Technology-leading" corresponds to "benchmark," "market-active" to "star," and "outstanding output value" to "fist." Serving as a demonstration and driving force is a shared requirement of all three.
Let's make one thing clear first, so expectations don't mismatch: Being included in the database is not a subsidy. Article 7 of the guidelines lists five services for inclusion in the database—access to a booth at the experience center, priority recommendation for applying for first-of-its-kind equipment and famous-quality products, priority inclusion in government scenario procurement and demonstration application lists, participation in policy briefings and investment and financing matchmaking, and inclusion in the tiered cultivation system for follow-up. None of the five provides funding.
Its value lies elsewhere: a free city-level exhibition booth, plus the endorsement of the Municipal Bureau of Industry and Information Technology. For companies in intelligent equipment, robotics, and industrial intelligent terminals, these two things normally cost money to buy—and cannot be bought.
The three categories don't test the same thing
Article 3 of the guidelines sets out the standards in great detail, but what really matters is the sentence on the "positioning" of each category:
Benchmark products are positioned as "representing the industry's highest level of innovation, with demonstration and leadership, independent and controllable, filling gaps, and standards-driving value." Among the four conditions, the hardest is the second—being the industry's first-of-its-kind (set), first batch, or first edition innovative product, or being included in a provincial-level or above AI terminal or innovative equipment catalogue. This item can be directly proven with certificates, while the other three (independent and controllable core intellectual property, driving the industrial chain, and completed scenario demonstration) all require supporting materials. For enterprises making industrial equipment, if they have first-of-its-kind (set) certification, this category is relatively easier.
Star products are positioned as “high in market popularity, strong in exposure, and highly recognizable; they are the core calling card for industry external promotion, exhibition display, and brand building.” Among the four conditions, one is “frequent appearances at industry exhibitions, major events, and authoritative media.” This is the only one among the three categories that assesses communication—what it requires is not technical parameters but exposure records. If enterprises do not usually accumulate media reports and exhibition photos, they cannot put them together at the last minute.
The positioning of flagship products is "the main force of the enterprise's core revenue, the foundation of industrial output value, with mature mass production, stable scale, and outstanding economic benefits." The four conditions are mass production, market share, output value, tax revenue and employment, and controllable costs. This category examines operating data and also requires that "the market share in the segment ranks among the top in the industry"—no specific percentage is given; during review, third-party data or industry certification provided by the enterprise itself is examined.
Looking at the three categories together makes it clear: benchmarks are assessed on technical evidence, stars on communication evidence, and blockbusters on business evidence. A single product can indeed meet all three, but that involves three different sets of material logic.
There is a client that makes industrial vision inspection equipment. The product has two invention patents and one software copyright, so the technology is passable; but it has never had media coverage, and has only attended exhibitions twice with an association delegation. Applying for the benchmark category has a chance; applying for the star category is basically hopeless—not because the product is bad, but because the type of evidence is wrong.
The two most troublesome items on the materials list
Article 5 of the Guidelines lists five categories of materials. The first three (application form, business license and intellectual property certificates, test reports and qualification certifications) are routine items that most companies have. What really trips people up is the last two:
Item four: demonstration application cases, media coverage, honors and awards — the guidelines note that "benchmark and star products should be supplemented on a priority basis," meaning that when reporting these two categories, don't leave them out.
Item 5, 5 actual product photos and 5 application scenario photos each, required to show the product from all angles, in PDF format, each no larger than 200M.
Ten images sounds simple, but in practice it is the easiest to redo. Most companies' product photos are promotional images shot by the marketing department, emphasizing atmosphere and composition, making the structure unclear; the review needs to see what the product looks like and in what scenarios it is used. The two purposes are different.
When we take photos at the customer site, our approach is: for product photos, take one each from five angles—front view, side view, top view, detail, and interface—with a clean background; for scene photos, capture the product's actual installed state in the customer's workshop or on the production line, including some environmental context. Export all of them at the same resolution to avoid large differences in size among the ten images.
The application form requires more detailed information: basic company information, product introduction, technical parameters, market situation, awards and qualifications, and application scenarios. The technical parameters section is the dividing line—filling in "high precision, high stability" is meaningless; you need to fill in specific measurement range, accuracy grade, repeat positioning accuracy, cycle time, and yield rate. These figures are available in the product's technical specification sheet, but if the company's product master data is not maintained, R&D has to be asked for them again every time an application is submitted.
In Odoo, this can be very lightweight: store technical parameters on the product template as attributes, or create a custom field to store the spec sheet link, and attach product photos and scene images under the product as attachments. This way, when quoting any product category, the parameters and images are ready to use, with no need to compile a document on the fly. Customers doing foreign trade have one extra step—the English and Chinese spec sheets must match; otherwise, once the technical parameters are entered, a review comparison will reveal that the two versions are inconsistent.
Selected quarterly, not a one-time pass
Article 6 of the Guidelines sets the pace: applications are accepted year-round; in principle, a centralized selection is organized once at the end of each quarter; assessments are conducted separately according to three categories of standards; products that pass are included in the 'Guangzhou Key Intelligent Products Dynamic List,' which is updated quarterly, iteratively expanded, and has ineffective products removed.
Article 8 is the exit mechanism: products already in the catalog may apply quarterly to update iteration information; a comprehensive review is conducted once a year to verify technological advancement and market activity; if production or sales are discontinued, major quality and safety issues arise, there is dishonesty or illegality, serious negative public opinion, or technology becomes outdated and is eliminated, they will be directly removed from the catalog.
These two points make it clear: after being entered into the database, it is not left unattended. Product iterations require updated information, and annual reviews require new evidence. We have seen market share figures filled in at the time of entry still filled with the same number in a review two years later.
This guide does not set quantitative thresholds: how much revenue, how many patents, what market share—all are described qualitatively. This is not an oversight—the scale standards for industrial intelligent terminals and smart home AIoT cannot be the same. The cost is that reviewers have broad discretion; what companies can do is make their evidence specific and not replace numbers with adjectives.
Three things you can do this week
Before November 20, go through the product line and select one or two. Criteria: whether the product falls within the nine categories listed in the guidelines (low-altitude terminals, intelligent connected vehicles and in-vehicle terminals, intelligent ships, XR and wearables, ultra-high-definition display and interaction, robots and embodied intelligence, industrial intelligent terminals, communications and industry-specific intelligent terminals, smart home AIoT), and is a complete machine, core component, or intelligent system that is independently developed, independently produced, and independently sold. Action: make a candidate list, annotate for each product which categories it can apply for, and for those that cannot, clearly state which type of evidence is missing.
By the end of November, have the technical parameters and ten images ready. Decision condition: whether key parameters in the product specification—such as range, accuracy, cycle time, yield, etc.—can be disclosed externally. Actions: take physical product photos from five angles and scene photos in the actual installation state, and export them uniformly as PDF; enter the parameters into the product master data and verify them against the specification for consistency.
By December 10, fill the evidence gaps. Assessment condition: among the three types of evidence (technical, communications, and business operations), which one is blank. The action is: for the technical category, find first-of-its-kind equipment, test reports, and patent certificates; for the communications category, compile exhibition records, media report links, and industry awards from the past two years; for the business operations category, prepare explanations of market share in segmented markets and output value and tax data. For whichever category is missing, either fill it in, or do not submit in this round.
The application channel remains open year-round, but selection is conducted quarterly. Miss this quarter, and you'll have to wait for the next one.
—— Policy basis of this article: the Guangzhou Municipal Bureau of Industry and Information Technology's "Notice on the Construction of an Intelligent Product Reserve Pool" (2026-11-03) and Attachment 1 "Guangzhou Intelligent Product Reserve Pool Entry Application Guidelines" on the same page; the terms, classification standards, material requirements, review and exit mechanisms all come from the original text on the portal website of the Guangzhou Municipal Bureau of Industry and Information Technology; the document forwarded by the Baiyun District Science, Technology, Industry and Information Technology Bureau on 2026-11-10 is available on the portal website of the Baiyun District People's Government. The Guidelines do not set quantitative thresholds, and this article does not set any numerical standards on their behalf; inclusion in the reserve pool does not include funding subsidies and falls within the scope of the service items listed in Article 7 of the Guidelines. Operational data such as the number of exhibits in the experience center are not found in this document and are not cited in this article.
