Pentacue AI
Pentacue AI applies computer vision to public FCC certification filings, revealing which electronic components sit inside more than 300,000 devices. Sales, supply chain and competitive intelligence teams use it to spot design wins, sourcing risks and market movements.
What is Pentacue AI?
Every wireless device sold in the United States must pass FCC certification, and that process forces manufacturers to file internal and external photographs, test reports and technical documentation. Pentacue AI mines that mandatory public record. Computer vision reads part numbers off circuit board photographs, including blurry ones, to establish which components sit inside which products, and therefore which supplier relationships lie behind a finished device.
The 2026 homepage describes the corpus as 300,000+ devices, 10,000,000+ files, more than 10 million circuit images and over 100,000 design wins, drawn from 80+ component manufacturers. Named manufacturers in the database include Microchip, Quectel, NVIDIA, Silicon Labs, Nordic Semiconductor, u-blox, Sierra Wireless, Texas Instruments, NXP Semiconductors, SK Hynix, Winbond, GigaDevice and Nanya. Component families run from microcontrollers and SoCs (ARM Cortex, RISC-V, AVR, PIC) to wireless and RF modules (BLE, WiFi, LoRa, cellular, Zigbee), memory (Flash, EEPROM, SRAM, DRAM), power management (DC-DC, LDO, battery management), interfaces (USB, UART, SPI, I2C, CAN), and sensors and analogue parts. New parts and manufacturers are added on demand.
The publisher frames the output along three axes: Design Win Intelligence, Supply Chain Intelligence and Component Intelligence, promising detection of design wins, supply chain risk and market movement months before products ship, straight from the filings themselves. Startmate, the accelerator that took the company into its Summer'25 cohort, describes the platform as scanning thousands of device registrations and breaking down unstructured data such as photos and part specifications to reveal which parts are used in specific devices; it cites a major cellular module manufacturer using it to find customers across North America and EMEA.
Access comes in two layers. A public one, the homepage search plus per-manufacturer and manufacturer-by-country pages, works without an account. The full product lives on a separate portal, insights.pentacue.com, and there is no self-service sign-up and no pricing page: entry runs through Book Demo or Get Access. The natural boundary of the whole approach is its raw material, since a product with no radio inside, or one never sold in the United States, never generates an FCC filing to analyse.
What it does
- Identify the companies using a given electronic component
- Detect a design win before the product ships
- Map component dependencies across 300,000+ devices
- Spot a single-source risk or a supplier change
- Receive alerts when a socket is won or lost
- Explore FCC teardown photos as supporting evidence
- Export results to a spreadsheet
When to use Pentacue AI / When not to
A quick filter to help you decide if Pentacue AI is the right fit.
When to use Pentacue AI
- Sales teams at electronic component manufacturers hunting for accounts that already design in comparable parts
- Manufacturers' representative firms: Startmate describes the tool serving sales teams representing Samsung, Microchip and 30+ other manufacturers
- Supply chain and procurement leads mapping component dependencies and single-source exposure across a device category
- Competitive intelligence analysts tracking which chips a rival wins or loses inside shipping hardware
- R&D and product marketing teams in electronics benchmarking component adoption before their own design decisions
When not to use Pentacue AI
- Teams researching devices with no wireless component: only products that must pass FCC certification enter the dataset
- Companies focused on markets with no United States sales, since the entire corpus comes from US regulatory filings
- Buyers who want to start immediately: there is no self-service sign-up, only a demo or access request form
- Procurement teams that need a published price before engaging a vendor, as no rate card exists anywhere on the site
- Sales organisations expecting a CRM or an outbound prospecting engine: Pentacue identifies component-to-device relationships, and the archived terms explicitly forbid automated scraping of results
How to use Pentacue AI
A typical end-to-end flow, from setup to results.
- Start on the homepage and run a 'Find Design Wins' search, which is public and needs no account
- Browse the public per-manufacturer pages (/microchip, /quectel, /nvidia and others) to see which companies use their parts
- Narrow by geography with the manufacturer-by-country pages, such as Atmel in Germany or Airoha in Taiwan
- Request the full product through the Book Demo or Get Access form, which asks for an email address
- Once access is granted, sign in to the insights.pentacue.com portal
- Run advanced device searches filtering by component manufacturer, component model, component type, device model, device name, device type, full FCC identifier or FCC notes
- Read the charts: Design Wins, Device Count Over Time, Device Trends and company-level trends
- Open the FCC teardown photos attached to a result to check the evidence behind each component identification
- Create notifications and set their scope to every company or a selected list, then edit, enable, disable or delete them as needs change
- Export the results to a spreadsheet and download the filing attachments
Pros & Cons
Pros
- Built on public, mandatory, dated raw material: FCC filings do not depend on manufacturers volunteering anything
- Claims a real timing advantage, surfacing design wins months before products ship
- Meaningful scale: 300,000+ devices, 10 million+ files and 80+ component manufacturers
- Every identification is traceable, since the teardown photograph acts as its citation
- The homepage search and the public manufacturer pages can be used without an account
- Usable output rather than a read-only dashboard: spreadsheet export and configurable notifications
- New components and manufacturers are added on request
Cons
- No public price and no published plan, so there is no way to gauge the cost without contacting the publisher
- No self-service sign-up: the full portal sits behind a demo or access request
- Scope is capped by the source, covering only devices with a radio component certified for the United States market
- The only accessible terms of service and privacy policy date from 29 May 2024 and predate the current product
- Those documents are templates reused from another publisher: the privacy policy names 'Counsel AI' four times
- The terms point to app.pentacue.com and privacy@PentaCue.ai, two addresses that no longer exist, with no DNS record and no MX record
- As of 13 August 2026 the marketing site returns 429 to any non-browser client and the portal backend returns 503; no security certification is claimed and no subprocessor list is published
Pricing & Plans
No price is published anywhere. The site carries no pricing page, no rate card and no figure on the 2026 homepage, and the portal bundle contains no pricing string of any kind, whether plan, credit, payment provider or per-month wording. The pricing model is therefore recorded as contact-sales, with no starting price, no currency and no billing unit to report. Neither a free plan nor a free trial is advertised for the current product. Access is obtained through the Book Demo and Get Access forms, both of which are simple email capture forms, so the actual cost is only disclosed in a commercial conversation.
Data, GDPR & hosting
A consolidated view of how Pentacue AI handles your data.
GDPR overview
No document names the GDPR. The only privacy policy available is the version archived on 29 May 2024, with no effective date stated, and it predates the current product. It claims safeguards for cross-border transfers, citing compliance with the European Commission, UK and Swiss adequacy decisions, and announces access, deletion, rectification, restriction and portability rights on request to contact@pentacue.com after identity verification, with a possible referral to the local data protection authority. Personal information is transferred to facilities and servers in the United States. No Article 27 EU representative is designated. Two red flags: the text is a template reused from another publisher, naming 'Counsel AI' four times, and it promises a subprocessor list 'available here' with no link and no such page among the 1,901 archived paths. The address given for reports concerning minors, privacy@PentaCue.ai, sits on a domain with no MX record and is undeliverable.
Who owns the data?
The only accessible legal texts are the terms of service and privacy policy archived on 29 May 2024, which predate the current FCC-filings product, so read them as indicative rather than current. They state that, as between the parties, the customer owns its Output and retains all right, title and interest, including intellectual property rights, in the Customer Data it uploads. The publisher receives a licence limited to what is necessary to deliver the service, resolve a technical incident or meet a legal obligation. Queries and responses generated by other users are not the customer's content. Feedback may be reused freely by the publisher, but without identifying the customer.
Reuse rights
Same caveat on dates: the only available texts were archived on 29 May 2024 and describe an earlier product. Under them the customer keeps its Output and needs no further permission to use it, subject to the terms, which elsewhere prohibit automated scraping of results. On the publisher's side, the terms allow PentaCue to collect and use Usage Data to develop, improve, support and operate its services, and to share that Usage Data with a third party; Usage Data explicitly excludes Content, Customer Data and the customer's confidential information. The privacy policy states that personal information is not sold. The categories collected are account, communication and payment information, social media data, technical data (logs, usage, device, cookies) and support exchanges. No document mentions training models on customer data, which is why that point is recorded as unclear rather than settled. The site does not respond to Do Not Track signals.
Data retention & training
Hosting summary
What is declared: the 2024 privacy policy states that personal information is transferred from the user's location to the publisher's facilities and servers in the United States. That is the only statement on the subject. There is no trust or security page, no named hosting provider and no region specified. What is observed, on 13 August 2026: the domain is registered through Amazon Registrar, the marketing site is served by Vercel (server: Vercel header), and the insights.pentacue.com portal is served by AWS CloudFront with its backend behind an AWS load balancer (server: awselb/2.0). The resolved IP, 216.150.1.1, geolocates to the United States but is flagged as anycast, and a CDN address proves nothing about where data is actually stored. The gap between the two matters: 'servers in the United States' is a declaration in a document that predates the current product, not a verified hosting commitment. No certification is claimed anywhere, with no occurrence of SOC 2, ISO 27001, HIPAA or CCPA across the collected material, and no subprocessor list is published.
Things to keep in mind
Risks and trade-offs to weigh before adopting Pentacue AI.
- As of 13 August 2026, pentacue.com returns HTTP 429 with a 'Vercel Security Checkpoint' to any non-browser request across the entire domain, homepage, robots.txt and images alike, verified seventeen times that day and again at 20:41
- The portal backend at insights.pentacue.com/api/v1/* returns 503 from an AWS load balancer on every endpoint tested, and the portal bundle has not been redeployed since 12 December 2025; ops.pentacue.com also returns 503, while app., api., docs., portal., dashboard. and trust. have no DNS record at all
- The privacy policy and terms of service have not been renewed since 29 May 2024 and are visibly reused from another publisher, with the name 'Counsel AI' still in the text
- The terms designate the publisher as 'PentaCue Corporation, a Delaware corporation' while the 2026 markup declares 'Pentacue, Inc.', and the 2024 terms set governing law to the State of Victoria in Australia with arbitration in Melbourne, although the published address is in Delaware
- The published address, 251 Little Falls Drive in Wilmington, is a Delaware registered-agent address rather than an operating office
- The JSON-LD declares a founding date of 2025, while the domain has been registered since 18 December 2023 and the first Wayback capture dates from 2 January 2024
- The product has changed nature entirely: until 2024 the same site sold an 'AI Equity Research Co-Pilot' for SEC filings, unrelated to the current tool, so older pages and reviews describe something else
Setup & Integrations
Technical difficulty
Technically trivial, administratively gated. There is nothing to install, as the tool is entirely web-based, with a public site and a separate portal at insights.pentacue.com. No connector, third-party integration, SDK or documented API is offered, so no engineering work is involved. The public part, the homepage search and the manufacturer pages, is usable straight away without an account. The real effort is human: the portal requires access obtained through a form, and no turnaround time for granting it is announced.
Deployment
Supported languages
Behind Pentacue AI
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What data does Pentacue AI actually work on?
How are the components identified?
How much does the database cover?
How much does Pentacue AI cost?
Can I try it without an account?
Is there an API?
Who is behind the tool?
How current are the legal documents?
Should you pick Pentacue AI?
Pentacue AI is a narrow tool built on an unusually solid foundation. FCC certification filings are public, mandatory and dated, so the raw material does not depend on manufacturers choosing to disclose anything, and reading part numbers off teardown photographs turns that regulatory obligation into something a component sales team or a supply chain lead can act on. For anyone selling chips, or anyone trying to see which parts a competitor has designed into a shipping product, the proposition is clear and the claimed scale, 300,000+ devices and 80+ manufacturers, is not trivial.
Two reservations weigh against it. The first is commercial opacity: no price, no plan, no self-service entry, every path to the product running through a sales form, which makes it impossible to size the investment before a conversation. The second is the legal frame. The only accessible terms of service and privacy policy were archived on 29 May 2024, describe an earlier and entirely different product, and are visibly borrowed templates, down to naming 'Counsel AI'. Several addresses they cite no longer resolve.
There is also an operational question a reader should not skip. As of 13 August 2026 the marketing site answers 429 to every non-browser client across the whole domain, the portal backend returns 503, and the portal bundle has not been redeployed since December 2025. None of that announces a shutdown, and none of it proves normal operation either, which is precisely why the status recorded here is unknown.
The sensible approach is to treat the public search and the manufacturer pages as a free sample, confirm the product is genuinely being served before investing time in it, and expect to negotiate both the price and the terms directly.
- Choosing a selection results in a full page refresh.
- Opens in a new window.