Frequently asked questions

Starbox AI is a trusted AI marketing service provider. We build proprietary knowledge bases and custom marketing AI agents for enterprises. When adopting AI, marketing offers the lowest risk and fastest time to results. Different business needs correspond to different products, which can be used individually or combined: if you want AI to recommend your brand when buyers search, there is Starbox GEO (AI search optimization); if you lack marketing content, there is Starbox Content Factory (automated social media content generation); if you want company-wide AI adoption, there is Starbox Starbox!; if you need deep customization, there is FDE consulting.

Generic AI tools do not know your business. Starbox first organizes your products, customers, brand guidelines, and domain expertise into a dedicated knowledge base, then customizes marketing AI agents for your business. As a result, generated content and recommendations are grounded in your actual facts and remain consistent across every output.

Choose based on the problem you need to solve right now: if you want AI to recommend your brand when buyers search, choose Starbox GEO and start with a free AI visibility audit; if you are short on social media marketing content, choose Starbox Content Factory; if you want your team to access AI directly through WeChat, choose Starbox Starbox!; if you want a systematic, deeply customized solution, choose FDE consulting. If you are not sure, ask Xiaohe on the homepage or leave your contact details so a consultant can help assess your needs.

Pricing is quote-based, depending on your industry, target market, and the mix of AI agents you need. Please schedule an appointment with a consultant; we will review your situation first before proposing a tailored plan.

Starbox has served over 500 businesses across industries including consulting, education, legal services, apparel, agency operations, home appliances, consumer goods, and manufacturing. Authorized client case studies are published on our Case Studies page.

Starbox AI was founded on January 18, 2021, by Wang Pengyou, evolving from community building, education, and the Starbox Alliance into AI. Backed by 15 years of marketing service experience across more than 500 businesses, the team pivoted fully to AI in 2024, packaging practitioner-level marketing capabilities into AI agents for enterprise delivery. Our slogan is "Starbox AI: Focused on Enterprise Growth."

No, they are two separate systems. Starbox GEO is a fully automated AI search optimization system designed to ensure AI actively recommends your brand when buyers ask Doubao, DeepSeek, or ChatGPT. Starbox Content Factory is an automated social media content generation system designed to solve content shortages and inconsistent output for marketing teams. You can use either one individually or deploy both together.

Marketing departments feel the strongest pain points, produce quantifiable metrics, have low onboarding friction, and run on easily replicable playbooks. Key metrics like content volume, lead generation, and AI recommendation rates can be tracked and compared directly, making marketing the easiest place to deliver visible wins before expanding AI into customer service, channel recruitment, administration, and other departments.

Yes. Starbox handles all deployment, delivery, and post-sales support, so your company does not need in-house developers. On the employee side, all you need is WeChat: say a single wake phrase or enter a prompt in Starbox Starbox! to trigger the agents. GEO and Content Factory run automatically in the background, leaving your team only to review and approve.

Starbox products run on your company's own knowledge base. Starbox Starbox! is deployed on your own servers, with data permissions managed across three tiers: public, private, and sensitive. Starbox GEO and Starbox Content Factory use only brand materials confirmed by your company, drawing facts solely from your knowledge base with zero fabrication.

A report showing exactly how AI currently describes your business. Starbox offers three free audits: Official Website GEO Audit, China AI Visibility Audit, and Global AI Visibility Audit. After you submit a request, a consultant will confirm details with you before scheduling the audit, and will review the report together with you.

Xiaohe is the AI advisor on the Starbox website. It can answer questions about Starbox products, services, case studies, and partnership models, and recommend suitable solutions based on your industry. Xiaohe's answers are AI-generated and for reference only; specific proposals are subject to discussions with our consultants.

Starbox delivers, services and supports every technical product directly from its headquarters in Hangzhou, China. The GEO China edition serves companies selling in mainland China, and the GEO Global edition serves companies selling internationally.

You can reach out in three ways: ask Xiaohe directly on the homepage; apply for a free AI visibility audit; or leave your contact information and requirements on the Book a Demo page. A consultant will contact you to understand your needs before presenting a proposal.

Starbox GEO re-tests every 30 days using the same fixed question set, tracking changes in your AI visibility index and metrics each round. The exact timeline depends on industry competition and existing content foundations; our consultants provide an estimate after the initial audit.

Just review. Audits, topic planning, writing, image generation, publishing, and re-testing all run automatically on schedule. Every draft enters a pending review queue where you can approve, reject for a rewrite, or download it; drafts are approved automatically once the review window expires.

Yes, as a local project. For local projects, author profiles include the city name across every version, ensuring that when users ask "city + question", AI connects the brand directly to that city.

Yes, and that is where most Chinese businesses expanding overseas start. We begin with your official site's AI readability, structured data, and content assets, then gradually expand brand mentions across third-party platforms and industry media.

The China edition targets domestic AI engines such as Doubao, DeepSeek, Tencent Yuanbao, and Tongyi Qianwen, using Doubao web search for diagnostics. The Global edition targets overseas AI engines like ChatGPT Search, Perplexity, and Google AI Overviews, using Gemini plus Google search for diagnostics, with content written in the target market's language.

Brand and product information, target audience, target markets, and key competitors. Starbox uses this to build your brand knowledge base, and all factual claims in generated content are drawn exclusively from this knowledge base.

GEO (Generative Engine Optimization) is the practice of making your brand and content discoverable, trusted, and cited in AI engine answers. As more buyers ask AI directly "which one is best" and "how to choose," GEO determines whether your brand appears in those answers.

Brand voice content is published through the brand's own accounts. Third-party voice content provides an external perspective and is recommended for publishing via third-party accounts. The workspace supports review, downloading, copying, and publishing, and automatically syncs publishing records.

A comprehensive metric that measures whether AI can "see" your brand when answering buyer queries. Based on a fixed set of prompts, the system checks prompt by prompt whether your brand is mentioned, where it ranks, which competitors replace it, and whether facts are accurate. It aggregates this into an overall score and compares each retest with the previous round.

SEO aims to rank webpages in search results; GEO aims to get your brand cited in AI-generated answers. Because AI synthesizes multiple sources to produce conclusions, GEO focuses on whether content can be parsed, trusted, and cited by AI, and whether verifiable brand mentions exist across third-party platforms.

Six stages: automated diagnostics, automated topic selection, automated drafting, automated image matching, automated publishing, and periodic automated retesting. Retest results drive the next round of topics, creating a continuous loop where your team only needs to review and approve.

They are two independent systems. Starbox GEO focuses specifically on AI search optimization, producing content tailored to diagnosed visibility gaps to get AI to recommend your brand. Starbox Content Factory focuses on digital and social media content to solve marketing content shortages. They can be used separately or together.

Leading domestic AI engines, including Doubao, DeepSeek, Tencent Yuanbao, Tongyi Qianwen, and ERNIE Bot.

Yes, by running it as a local project. In local projects, author profiles include the city name across every version, ensuring that when users search "city + query," AI connects your brand with that city.

Yes, for educational content. For sensitive industries, author bios and account profiles focus strictly on education, avoiding claims around diagnosis, treatment outcomes, or financial returns. The platform checks this across content generation, selection, and saving.

It runs a fixed set of diagnostic queries via Doubao with web search enabled. It evaluates whether your brand is mentioned, where it ranks, which competitor took its place, and whether brand facts are accurate, while tracking which sources the AI cited.

Content is prioritized for platforms most frequently cited by AI in the diagnostics, typically including Zhihu, Xiaohongshu (RED), Toutiao, Baijiahao, and Sohu. Published links are automatically logged in dashboard reports.

Nationwide projects focus on category and comparison queries like "best [category]" or "how to choose [category]." Local projects target "city + problem" queries and include city names in author bios so AI connects the brand with the location.

Sample projects use 125 fixed queries covering the buyer decision journey across categories, comparisons, trust, product fit, and transactions or after-sales. The exact number is determined by industry and project scope, and remains fixed once established to allow consistent round-by-round benchmarking.

Yes. Request a China AI visibility diagnostic, and we will run a set of queries on Doubao with web search to see if your brand is mentioned, where it ranks, and who replaces it. A consultant will review the report with you.

It automatically retests every 30 days using the same query set. Reports are generated in the dashboard, comparing results query-by-query against the previous round.

Brand and product collateral, target customers, target cities or coverage regions, and main competitors. Starbox builds a brand knowledge base from these materials, ensuring all content facts are drawn exclusively from it.

Global B2B manufacturers, DTC consumer brands, and SaaS companies whose buyers are overseas and use tools like ChatGPT and Perplexity to research vendors.

Written in the language target market buyers actually search in—usually English; text in accompanying images is also exclusively in the target language.

Yes. That is where most Chinese companies expanding overseas begin. Starbox starts with your website's AI readability, structured data, and content assets, then gradually builds brand mentions across third-party platforms and industry media.

AI engines commonly used by overseas buyers, including ChatGPT Search, Perplexity, Google AI Overviews, and Claude.

Queries are run in the target market's language. Using Gemini alongside Google search, we evaluate prompt by prompt whether your brand is mentioned, where it ranks, who replaces it, and which sources the AI cites.

Overseas AI engines are typically more cautious about Chinese brands, applying a "trust discount." Many companies also rely solely on an official website, with very thin third-party content ecosystems. GEO Global edition uses verifiable third-party content and industry media mentions to rebuild this trust.

On sources overseas AI models frequently cite—such as Reddit, YouTube, industry media, and official blogs—prioritized based on the citation sources identified in diagnostics.

First, allow AI crawlers, deploy structured data, and organize product and credential pages so AI can understand and cite them. You can start by requesting a free website GEO audit to see scores across five dimensions and actionable recommendations.

No. Content is written directly in the natural phrasing of buyers in your target market, and image text is also in the target language to avoid the unnatural phrasing of machine translation.

Yes. Apply for an overseas AI visibility audit to run queries in your target market's language, see how ChatGPT, Perplexity, and Google AI respond to your prospective buyers, and review the report alongside a consultant.

They are compiled from real questions asked by target buyers across different decision-making stages, covering categories, comparisons, trust, product fit, and transaction and after-sales. Once established in the first formal audit, the prompt set remains fixed.

Once the retest finishes, the platform compares your scores and metrics against the previous round and automatically updates your GEO topics: used topics are flagged, unused ones roll over, still-valid ones get a light rewrite, and fresh topics are added.

Yes. Once generated, each draft goes to pending review. You can approve it, send it back with feedback, or download it. Drafts pending review for more than 48 hours are automatically approved.

Once the audit is complete, the system generates topics automatically: it analyzes AI answers to see which sources get cited most, which questions lack your presence, and which facts are misrepresented. Topics directly target those gaps and are organized by category and buyer journey stage.

Five specialized writing AI agents produce drafts daily based on your topics: short posts with images, product recommendations, brand deep-dives, knowledge deep-dives, and third-party fact-check articles. Facts are drawn strictly from your brand knowledge base, structured the way AI platforms like to cite: lead with the conclusion, use complete names, and provide verifiable facts.

Once each draft is written, the system automatically creates 2 to 3 matching images tailored to the content and destination platform. The AI agent determines the exact number, so no designers are needed.

Brand voice means your company speaking directly, published via official brand accounts with clean copy and author info kept separate. Third-party voice takes an outside perspective, best published via third-party accounts, naturally weaving in your brand name and one verifiable brand fact. Running both tracks together makes it far easier to build trust with AI platforms.

Published links automatically feed back into the publishing reports on your workbench. You can filter by platform or publisher and export with one click, without any manual formatting for reporting.

Track metrics from each round of retesting on the same question set: AI visibility index, core question mention rate, top recommendation rate, unprompted mention rate, brand fact accuracy rate, and the number of red-flag questions, compared item by item against the previous round.

Usually just one team member to handle reviews. Audits, topic planning, drafting, image generation, publishing, and retesting run automatically on schedule. If drafts pass the review deadline, they are approved automatically, keeping your workflow moving.

Writing directly with standard AI means brainstorming topics yourself every time, getting inconsistent output, and dealing with an AI that doesn't understand your brand. Content Factory first builds a brand knowledge base, then automatically draws topics from a vast topic pool using four methods, and sends them to AI writing agents across nine dimensions—brand voice, third-party voice, and word of mouth. Every piece follows a fixed five-step pipeline and draws facts exclusively from the knowledge base, ensuring consistent, compliant, and non-repetitive output.

Nine dimensions: brand voice (brand product seeding & Moments, brand-spoken video & industry explainers, long-form brand articles & educational deep dives), third-party voice (influencer seeding, influencer video scripts, expert site visits), and word of mouth (everyday user notes, customer case study videos, case stories)—covering platforms like Xiaohongshu (RED), WeChat Official Accounts, Douyin, WeChat Video Account, WeChat Moments, Zhihu, and more.

The topic pool is generated by combining elements from your brand knowledge base—such as product selling points, target audiences, use cases, and emotional hooks—creating thousands of topics that expand continuously. Topics are retrieved daily using four methods (sequential, randomized, trending, or product-focused). Used topics are automatically tagged to avoid repeating the same angle.

No. General rules, AI agents, and tone-of-voice samples are all reused. Switching brands only requires updating the brand knowledge base (D01–D06).

It solves content shortages and inconsistent output in enterprise marketing: managing multiple platforms and frequent posting makes relying solely on human writers unsustainable; topic ideation dries up quickly; and staff turnover leads to lost brand voice and expertise. Content Factory turns content production into an automated assembly line, so your team only needs to review.

Sequential is best for steady, everyday volume to ensure complete coverage without gaps. Randomized keeps content fresh by finding new angles even for the same product. Trending leverages current events by automatically matching daily holidays, solar terms, and platform trends. Product-focused is designed for new launches and core product pushes, concentrating content output.

They represent three perspectives based on who is speaking: brand voice comes from the brand and its founder to build brand awareness; third-party voice comes from influencers and outside experts to build trust from an external standpoint; word of mouth comes from everyday users and real customers to deliver social proof and case studies. Multiplying these three voices across three formats—short image-text posts, video scripts, and long-form articles—yields nine content dimensions.

Six categories: D01 Product Arsenal, D02 Target Audience & Scenario Library, D03 Writing Rules, D04 Brand-Specific Constraints, D05 IP Story Library, and D06 Enterprise Knowledge Base. Starbox will help you organize these; you only need to provide product information, customer profiles, and existing top-performing content.

Output is configured according to the platforms and frequency your business needs. In a typical benchmark setup, a brand steadily produces around 36 pieces a day, covering platforms like Xiaohongshu (RED), WeChat Official Accounts, Douyin, WeChat Video Account, and WeChat Moments (illustrative figure).

Facts are pulled exclusively from the brand knowledge base with zero hallucination. Each piece is crafted through a five-step process: audience anchoring, scenario derivation, initial draft, polishing & compliance check, and finalization with image matching. During polishing, every piece is scanned for advertising law superlatives, brand-banned terms, competitor mentions, and pricing constraints before reaching human review.

Yes. Completed drafts go into a pending review queue where they can be approved, returned with revision notes, or downloaded. Drafts unreviewed after 48 hours are automatically approved. Any return notes are fed into the next writing pass.

An employee takes a photo of a product, and the AI identifies its style, fabric, and selling points, deduces the best audience and angles, and generates titles and copy for platforms like Xiaohongshu (RED), Douyin, and WeChat Channels. Using that photo as a reference, it creates covers, lifestyle scene images, and detail shots—with the actual product in the visuals. The same photo can also be used to generate WeChat Moments posts, spoken-video scripts, WeChat Official Account articles, and e-commerce product detail pages.

The style library is Content Factory's built-in collection of high-performing frameworks and writing styles, featuring 895 spoken-video styles, thousands of proven graphic-and-text frameworks, and over 35,000 style samples. When drafting, you can apply any style based on account positioning: spoken-video styles mimic an individual persona's tone, while graphic-and-text frameworks follow structured storytelling templates. Top-performing content from your own accounts can also be compiled into the knowledge base as custom brand voice samples.

They are two independent systems. Content Factory solves social media content shortages for day-to-day operations on platforms like Xiaohongshu (RED), WeChat Official Accounts, and short video platforms. Starbox GEO specializes in Generative Engine Optimization, ensuring AI search engines recommend your brand when buyers ask.

"Starbox Starbox!" is an all-in-one enterprise AI marketing agent deployed on your company's own servers. It offers three core advantages: first, an ultra-low learning curve—employees trigger it in WeChat simply by typing "Starbox Starbox!...", with no prompt engineering or agent training required; second, a shared company brain—every output is centralized into your knowledge base and organized by AI into sourced customer, employee, and company wikis, keeping assets intact when staff leave, with private deployment ensuring data security; third, Cockpit visibility—team activities, customer pipeline progress, and AI workload metrics are clear at a glance.

No. Employees state what they need in a single sentence, just like briefing a colleague. The system automatically identifies the task type, matching the right skills and company data. Most employees are up and running on day one.

The system is deployed on your company's own servers with tenant-level data isolation. Materials are categorized into public, private, and sensitive access tiers; the underlying knowledge base and prompts remain hidden from regular employees. All file archives and edits are logged by user ID, and the AI has no deletion permissions.

Starbox Starbox! is a standardized product ready for immediate use upon deployment. FDE consulting is a separate advisory and on-site service where technical teams and marketing consultants streamline your workflows, build custom AI agents, and train your internal AI operations leads. They can be purchased independently or together.

Tracking and updating project milestones, drafting quotes and contracts, managing client follow-ups and talk tracks, personal branding interviews and persona building, designing posters and product detail pages, creating e-commerce creative assets, producing short-video scripts and franchise recruitment proposals, running GEO audits, scheduling automated reports, generating meeting minutes, and more, with capabilities continuously expanding.

The wake phrase is "Starbox Starbox!". Employees add "Starbox AI" as a contact in WeChat and type "Starbox Starbox!..." just like messaging a colleague to access all capabilities, without installing any new software.

WeChat is the tool employees already use every day, offering the lowest barrier to adoption. Direct-messaging Starbox AI feels just like chatting with a colleague—no need to learn prompt engineering, no need to figure out AI agents, and no need to open another system.

Every task employees assign to AI and every file they save is centralized into the company knowledge base. AI organizes it into three sourced wikis: Clients, Employees, and Company. When a new team member takes over an account, a single prompt pulls up the entire history, ensuring institutional knowledge stays when people leave.

Recent team activity, workload handled by AI, client and contract progress, real-time updates, and more. You can toggle between "Company-wide" and "My view", giving managers a complete overview on a single screen.

Yes. After a meeting, just say in WeChat: "Starbox Starbox! Summarize the meeting notes from DingTalk just now." The system connects via DingTalk's API to read AI transcripts (read-only), extracts key takeaways, action items, and risks, automatically files them under the relevant project based on clients mentioned, and asks if you want to route the action items to the relevant colleagues.

Deployment is handled directly by Starbox, with the timeline depending on company size and scope of customization. On the employee side, all they need to do is add Starbox AI as a contact on WeChat. Most team members can start using it on day one.

Hundreds of agents categorized across content creation, channel partner acquisition & sales, visual design, interviews & personal IP, GEO, and project & office management. Examples include: Top 10 Moments Post Formats, Partner Acquisition Master (pitchbooks, launch event plans, full presentation scripts), Global GEO 8-Piece Suite, Talking-Head Quote Scripts, E-Commerce Product Pages, Founder IP Interviews, and more. Agents are centrally developed by headquarters, installed on demand, and updated automatically. Employees still invoke them with a single phrase in WeChat.

No. All data is stored on your enterprise's own servers and belongs to the company. Every save and edit has a complete audit trail. When an administrator deactivates an account, access is revoked instantly.

FDE stands for Forward Deployed Engineer, a model pioneered by Palantir in the 2010s: dispatching engineers directly to client sites to solve problems end-to-end until the system actually works in production. It is not sales, not traditional consulting, not mere implementation, and not outsourcing—it delivers a fully functioning, proven business operation.

Traditional FDE focuses purely on engineering, typically serves large enterprises with mature data foundations, and requires clients to articulate their own requirements. Starbox FDE builds on technical engineering with 15 years of B2B experience across consulting, marketing, outsourcing, and training, backed by insights from serving over 500 enterprises. We first diagnose the business and map workflows before building AI agents—so if you can't clearly define your requirements, we structure them for you.

Six core initiatives in sequence: enterprise diagnostics and assessment, in-depth requirement discovery, end-to-end SOP optimization, agentifying repetitive tasks, department-level AI enablement, and using proven methodologies to cut trial and error. Stage one typically starts with marketing, then scales across the company department by department on a quarterly basis.

Marketing feels the sharpest pain points, delivers measurable outcomes, has a low barrier to entry, and produces replicable experience—making it the easiest place to show early results. Starbox validates the full cycle within marketing using the DDABC full-funnel marketing model (Define, Discover, Believe, Convert, Continue), then expands to operations, customer service, and functional teams.

Eight enterprise-owned assets: the underlying platform foundation, enterprise knowledge base, process and SOP framework, distilled role capabilities, a dedicated matrix of workplace AI agents, a customized DDABC full-funnel marketing system tailored to your industry, the team's operational AI capabilities, and an industry benchmark solution built through year-long hands-on support.

No. FDE handles orchestration, supervision, and hands-on coaching, but does not step in to do the day-to-day work for your staff. The definition of success is that once Starbox steps back, your internal team continues to work productively with AI without relying on external personnel.

In five steps: an initial consultation to set the direction; developing an implementation plan and deploying the first AI agents in a single department (approx. 15–30 days); a proposal roadshow and internal rollout guidance; real-world trial and alignment for about one month; and finally finalizing an annual framework agreement. A deposit of intent is paid at kickoff and fully credited toward the annual framework fee once signed.

Q1 focuses on overall architecture and initial department implementation, completing the platform foundation, knowledge base, DDABC full-funnel model, role capability distillation, complete SOPs, and the launch of the first AI agents—requiring the most concentrated effort. Q2 and Q3 expand to other departments with continuous optimization. Q4 solidifies adoption across all departments and codifies an industry-level solution and benchmark case studies.

Companies that already recognize the value of AI, want systematic implementation, and require deep customization aligned with their internal workflows. Businesses that only need off-the-shelf products can directly use Starbox GEO, Starbox Content Factory, or Starbox Starbox!.

Engagements are structured as annual frameworks, with two options: Lighthouse Co-creation and Standard. For Lighthouse clients, Starbox covers part of the cost in exchange for participating in public case studies. Standard clients have no obligation to share public case studies. For specific pricing, please book an FDE consulting session for an initial discovery call before receiving a formal proposal.