Alpha Quantum classifies the web at full scale and turns it into decisions machines can act on: what a page is, which AI tools employees may use and what they do with the data, and which pages an AI agent may open. Our AI block list covers 20,000+ AI-tool domains with risk flags on every row. Our AI agent allow list covers 40M domains with 28 page types each.
Domain intelligence, AI tool control, agent browsing policy, product classification, privacy and talent data. Every number below comes from the platform's own page.
Classify any URL in real time across IAB v2, IAB v3, IPTC and web-filtering taxonomies. 15+ data fields per request and offline databases from 10M to 120M domains.
Explore platform20,183+ AI-tool domains in 18 categories and 165 subcategories, updated daily. Every row carries four risk flags: trains on your data, data sovereignty, abusive purpose, AI-native or AI-enabled.
AI block listFor 40M+ domains, the verified URLs of 28 page types, login, signup, checkout, pricing and 24 more, with 700+ IAB and 59 filtering categories per domain. The policy data layer for agent guardrails.
AI agent allow listAI product classification for Google Shopping, Shopify, Amazon and IAB taxonomies, with gender categorization, buyer persona mapping and custom classifiers. 5,574 Google categories, 200+ languages.
Explore platform120M+ domains classified by IAB and web-filtering taxonomies, enriched with company and firmographic data and web technology stacks. Offline files for firewalls, DNS filters, proxies, SIEMs and data warehouses.
Explore platformDiscover and anonymize PII, PHI and other sensitive data in text, documents, images and video. Multi-technique engine, context-aware detection, real-time streaming for live data, built for GDPR.
Explore platformParse any resume into clean structured JSON: 114+ candidate fields from PDF, DOCX, TXT, RTF, HTML, ODT, Excel and scanned documents via OCR. Normalizes titles, skills and locations.
Explore platformEmployees paste company data into AI tools. AI agents open pages nobody reviewed. Both problems are lists of domains and URLs, with the right flags, kept current. That is what we build.
Blocking AI by name does not work. The row for openai.com says the API does not train on your data. The row for chatgpt.com says the consumer app does. That difference is the product: four risk flags on every one of 20,183+ domains, so policy follows risk, not brand. Real rows from the free sample:
A domain tells an agent nothing about the page. The same site holds documentation it should read and a checkout it must never touch. Every domain in this database carries the verified URLs of 28 page types, so your gateway checks the page type before the request is sent, and unknown sites are blocked by default.
Built at the end of 2025. Replayed request by request against all nine 2026 agent incidents, our database plus egress rules would have blocked almost all of them. Refreshed quarterly so the policy stays true as the web moves.
Our 40M-domain database, our egress rules and our host list, checked by your gateway before any request leaves. This is the mechanism behind every result in the incident replay above.
These are separate systems with separate data. One classifies the web, one classifies products, one finds and removes sensitive data, one reads documents. Each platform runs on the engine built for it.
The domain corpus behind website categorization, the offline database, the AI block list, which is one slice of that 120M-domain filtering database, and the AI agent allow list, which adds verified page-type URLs on 40M of those domains.
Models trained on product titles and descriptions, not web pages. Maps free-text listings to Google Shopping, Shopify, Amazon and IAB taxonomies, assigns buyer personas and gender, and trains custom classifiers on your own taxonomy.
Context-aware detection of PII and PHI in text, documents, images and video, then anonymization, pseudonymization or redaction with a multi-technique engine, including real-time streaming for live data.
Turns resumes in PDF, DOCX, TXT, RTF, HTML, ODT, Excel and scanned images into structured candidate JSON, normalizing titles, skills and locations for recruiting and HR workflows.
Each card is a decision a machine has to make millions of times a day, and the data that makes the decision defensible.
A browsing agent cannot tell a checkout from a product page until it has loaded it. By then the request is out. Our per-URL page types turn that into a rule the gateway checks before the click.
The long tail is where the risk lives: 20,183+ AI-tool domains, most of them tools nobody has heard of, flagged for training on your data, data sovereignty and abusive purpose. Allow by policy, block by risk.
Programmatic moves too fast for manual review. 700+ categories separate market analysis from fraud reports, and real-time classification keeps inclusion and exclusion lists current.
Patient files, KYC documents and contracts carry personal data in every clause. 70+ sensitive data types found and transformed, in 50+ languages.
Anonymization APICIPA, acceptable use and network security all rest on knowing what every domain is. 120M+ classified domains in the formats firewalls, proxies and DNS filters already read.
URL Categorization DatabaseListings land in the wrong place and search returns noise. Every product mapped to Google Shopping, Shopify, Amazon or your own taxonomy in one call.
Product Categorization APIYour application sends a URL. The response carries the domain's classification across every taxonomy in your subscription, the page type of that exact URL when you license the AI agent allow list, and audience and quality data, from the real-time API or from a local copy you keep synced.
One site can be several things at once, so the engine returns multiple labels rather than a single answer. A streaming platform can be Video, User-Generated and Entertainment at the same time, and its login page is still a login page.
Different teams, different platforms. Every role gets the one built for the problem it owns.
Enterprise IT teams, school districts and managed security providers deploy our databases directly into firewalls, DNS resolvers and proxies. Daily-refreshed data in RPZ, EDL, PAC and hosts formats keeps policies current.
The same enforcement points now govern AI tools: the AI block list adds 20,183+ AI-tool domains with risk flags to the filters you already run, so shadow AI is a policy, not a hope.
Research agents, sales agents, shopping agents and browser automation all land on pages nobody reviewed. The AI agent allow list gives your gateway a per-URL page type for 40M+ domains, so login, checkout, upload and post pages are refused before the request leaves.
Licensed as a database for embedding in your product, fully on-prem, or as a lookup API.
SSPs, DSPs and publishers use real-time website categorization across 700+ categories to keep brand-safety inclusion and exclusion lists current against a web that changes every hour, protecting advertisers without killing reach.
IAB v2 and v3 taxonomies are supported natively, and sub-100ms responses fit programmatic timing.
Healthcare systems, financial institutions and legal teams integrate the Anonymization API into document pipelines. Medical records are de-identified for research, financial narratives redacted for filings, contract reviews stripped of counterparty PII before storage.
Volumes that would need a 50-person team are handled by one endpoint, consistently and auditably.
Marketplaces and e-commerce platforms use the Product Categorization API to normalize millions of seller-submitted listings into clean category trees, with buyer personas and gender labels alongside the category.
Recruiting and HR teams use the Resume Reader API to turn any resume file into 114+ structured fields, from $0.0198 per resume.
Most providers cover the top 1 to 10 million domains. Our corpus covers 120 million, including the long tail of new sites, niche verticals and regional domains. If your use case depends on classifying the unusual, we cover it.
Classification, AI tool control, agent browsing policy, product data, privacy and talent intelligence, each on the engine built for it. One vendor relationship, consistent APIs, one support channel.
Every API offers a 14-day free trial with no credit card. Database products include free sample downloads. Live demos run against production models with your own data, no registration walls.
Before AI data platforms, Alpha Quantum built quantitative software for the financial industry: portfolio optimisation, risk management, security analysis and news analytics used by asset managers, hedge funds and wealth managers. The same engineering discipline now runs our data platforms.
Mean-variance, CVaR and multiperiod optimisation with efficient frontiers, correlation cleaning, backtesting over parameter grids and automated factsheet generation.
View product presentation (PDF)VaR and CVaR across historical, parametric and Monte Carlo methods, stress testing with historical and custom scenarios, risk attribution and pre-trade compliance.
View product presentation (PDF)Test the platforms with your own data. Demos run against production APIs. No sign-up, no credit card, no time limit.
Enter any URL and see IAB categories, technology stack, audience personas and quality scores in real time.
Launch DemoThe 2026 agent incidents replayed request by request, with the page type or rule that refuses each one.
See the replayDownload the free 50-row sample with categories and all four risk flags, and browse the 18 categories.
Get the samplePaste a product title and get instant classification across Google Shopping, Shopify and Amazon taxonomies.
Launch DemoPaste text with personal data and watch the API detect and redact names, emails, IDs and more.
Launch DemoAnonymize and pseudonymize sensitive data in text and documents while keeping the data usable.
Launch DemoSee a resume parsed into structured JSON with 114+ fields, from any common file type.
Launch DemoSix deployments, each drawn as it runs: where the data comes from, which platform decides, and where the decision is enforced.
Start with a demo, download a sample database, or talk to our team about your use case. No commitment, no sales pressure, just data you can test.
