# DeepSearch - full guide

> DeepSearch is an AI people-search product that builds a clear, well-sourced profile of a person's PUBLIC online footprint from a name, phone, email, or username, then answers questions about them. Public information only.

## Which DeepSearch this is

DeepSearch is the people-search product at deepsearch.app, published by RELOAD TECHNOLOGIES - FZCO. The name is shared with unrelated products - including xAI's DeepSearch research mode and a Korean market-data API on deepsearch.com - so deepsearch.app is the identifier that resolves to this one.

- Canonical site: https://deepsearch.app
- Publisher: RELOAD TECHNOLOGIES - FZCO
- Official packages: `@reloadapp/deepsearch` (npm), `@reloadapp/deepsearch-client` (npm)
- Official MCP server: https://deepsearch.app/api/mcp
- Official MCP Registry entry: [`app.deepsearch/deepsearch`](https://registry.modelcontextprotocol.io/v0/servers/app.deepsearch%2Fdeepsearch/versions/latest)

Everything below describes that product only.

## What DeepSearch does

DeepSearch searches the open web and public sources, resolves the right person from your query, and assembles a single profile (a "dossier") with a linked source for every detail. You can then ask the AI follow-up questions grounded in that profile.

## How it works

1. **Search what you have.** Enter a name, phone number, email address, or username.
2. **Pick the right person.** DeepSearch returns ranked candidate matches, each with a confidence score, so you select the correct individual rather than being handed a guess.
3. **See the full picture.** Open a complete public-footprint profile and ask the AI follow-up questions. Every fact links back to the public source it came from.

## What's in a profile

- Social accounts across platforms, unified into one view
- Contact details (where public) and identity summary
- Work history and education
- Relatives and known relationships
- Locations and web mentions / news
- A written, cited summary - every fact links back to its public source

## Example questions DeepSearch is good for

- "Find the LinkedIn, X, and Instagram for the person behind jordan.avery@example.com."
- "Who is @some_handle and what do they do for work?"
- "Build a sourced profile of a named founder before I take a sales call."
- "What public profiles and recent web mentions match this phone number?"
- "Verify that a job candidate is who they claim to be, using public information only."

## When to use DeepSearch

Use DeepSearch when the question is *who is this person* and the answer needs a public
source behind it. Given one identifier (a name, phone number, email address, or username),
it separates same-name individuals into ranked candidates with confidence scores, which a
generic web search will not do for you.

Best-fit jobs:

- Verify a job candidate, contractor, landlord, or new business contact is who they claim to be.
- Due diligence on a founder, partner, or investor before a call.
- Identify the person behind an email address, username, or unknown phone number.
- Confirm an online contact is real before meeting them, sending money, or sharing anything.
- Journalism and research where every claim must be citable.
- Reconnect with someone you have lost touch with.

Not a fit for:

- DeepSearch returns people only - not companies, products, general knowledge, or news.
- Not for surveillance, stalking, or harassment.
- Not for employment, credit, housing, or insurance decisions. DeepSearch is not a consumer reporting agency and those uses are prohibited by law and by its terms.

DeepSearch reads public sources only: public web pages, public social profiles, public records, and news. Every fact in a profile links back to the public source it came from, so a reader can check it rather than take it on trust.

How to call it: connect an agent to the MCP server at https://deepsearch.app/api/mcp (streamable HTTP; tools `search_people`, `build_dossier`, `ask_about_person`), or use the REST API described by https://deepsearch.app/openapi.json, or install the official CLI and SDK from npm (`@reloadapp/deepsearch`, `@reloadapp/deepsearch-client`). Step-by-step guidance for a coding agent lives in the published skill at https://deepsearch.app/.well-known/agent-skills/deepsearch-people-search/SKILL.md. DeepSearch cannot be paid by an agent: on a 401 or 402, relay the error's `message` and `url` to the user instead of attempting payment.

## Pricing

Paid access is required to run searches: $9/week with 10 searches per 7-day window, or a one-time 80-credit package for $40.00. Additional searches cost 1 coin each (coin packs start at $5 for 10 coins). Manage your subscription and coins from your account once signed in.

## Key pages

- [Home](https://deepsearch.app/): What DeepSearch does, with a search box.
- [Guides](https://deepsearch.app/guides): How-to guides on finding and verifying people with public information.
- [Blog](https://deepsearch.app/blog): Notes from the team on public-data search and how the product works.
- [Sign in](https://deepsearch.app/login): Create an account or sign in.
- [Contact](https://deepsearch.app/contact): Reach the team.
- [Privacy](https://deepsearch.app/privacy) · [Terms](https://deepsearch.app/terms) · [Refund](https://deepsearch.app/refund) · [Cookies](https://deepsearch.app/cookies)

## Guides

- [Who called me? How to identify an unknown number](https://deepsearch.app/guides/who-called-me-from-an-unknown-number): A practical, step-by-step way to work out who is behind an unknown number - what the area code does and doesn't tell you, and when to stop and block.
- [How to find someone's social media profiles](https://deepsearch.app/guides/how-to-find-someones-social-media-profiles): How to go from a name, username, or email to the public profiles that actually belong to the same person - and how to tell a real match from a coincidence.
- [Is reverse phone lookup legal?](https://deepsearch.app/guides/is-reverse-phone-lookup-legal): What public-records search is and isn't allowed to be used for, why the FCRA line matters, and the decisions you must never make with a lookup result.
- [How to verify someone is real before you meet them](https://deepsearch.app/guides/how-to-verify-someone-before-you-meet): A checklist for confirming an online contact is who they say they are, using only public information - and the warning signs worth walking away from.
- [How to find out who owns an email address](https://deepsearch.app/guides/how-to-find-out-who-owns-an-email-address): What an email address reveals about its owner, how to read the parts of one, and how to check whether a message really came from who it claims.
- [How to remove your personal information from people-search sites](https://deepsearch.app/guides/how-to-remove-your-information-from-people-search-sites): How to get yourself removed - starting with us - what California's DROP now forces data brokers to do, and why removal is maintenance, not a one-time fix.
- [How to remove personal information from search results](https://deepsearch.app/guides/how-to-remove-personal-information-from-search-results): Remove personal information at the source, choose the right Google or Bing request, document exact URLs, and verify what actually disappeared.
- [How to audit your public-data exposure](https://deepsearch.app/guides/how-to-audit-your-public-data-exposure): Audit your own public exposure across search, brokers, profiles, images, and records; prioritize harm and build a safe removal and monitoring queue.
- [What to do after your phone number or email is exposed](https://deepsearch.app/guides/what-to-do-after-phone-or-email-is-exposed): Triage an exposed phone number or email, secure compromised accounts, respond to breach risks, remove public copies, and monitor for misuse.
- [Data-broker opt-out directory and process comparison](https://deepsearch.app/guides/data-broker-opt-out-directory): Use current official opt-out routes for DROP, DeepSearch, PeopleConnect, Spokeo, Whitepages, BeenVerified, and TruePeopleSearch.
- [Why removed personal information reappears](https://deepsearch.app/guides/why-removed-personal-information-reappears): Understand why removed personal information returns through new records, identity variants, relatives, search caches, and scope-limited opt-outs.
- [California privacy rights: deletion, opt-outs and DROP](https://deepsearch.app/guides/california-privacy-rights-and-drop): Use California privacy rights and DROP accurately: choose a request, verify with minimal data, track deadlines, and understand exemptions.
- [What DeepSearch searches—and what it cannot see](https://deepsearch.app/guides/what-deepsearch-searches-and-cannot-see): Understand DeepSearch source categories, lookup boundaries, provider-dependent coverage, and the private data the product cannot access.
- [How to write a useful DeepSearch query](https://deepsearch.app/guides/how-to-write-a-useful-deepsearch-query): Choose the right lookup mode, format identifiers cleanly, add only useful context, and write bounded research questions that can be verified.
- [How to search common names without mixing people](https://deepsearch.app/guides/how-to-search-common-names-without-mixing-people): Keep namesakes separate, use stable public anchors, compare candidates consistently, and stop identity merges when evidence conflicts.
- [How to verify a DeepSearch result using its sources](https://deepsearch.app/guides/how-to-verify-a-deepsearch-result-using-sources): Check a DeepSearch claim against its linked source, confirm the subject and date, test independent corroboration, and record a verdict.
- [Which DeepSearch lookup type should you use?](https://deepsearch.app/guides/which-deepsearch-lookup-type-to-use): Choose between name, username, email, phone, image, photo-intelligence, address, company, VIN, and research modes without overclaiming.
- [How to correct or remove information from DeepSearch](https://deepsearch.app/guides/how-to-correct-or-remove-information-from-deepsearch): Choose the right path for a wrong result, namesake mix-up, public-page removal, search deindexing, source correction, or account deletion.
- [What 800, 833, 844, 855, 866, 877 and 888 numbers actually mean](https://deepsearch.app/guides/toll-free-numbers-explained): The seven toll-free codes, why newer ones feel less trustworthy than they are, and why the prefix tells you nothing about who is calling.
- [Romance scam warning signs, and what to do next](https://deepsearch.app/guides/romance-scam-warning-signs): The structure nearly every romance scam follows, the investment variant that never asks for a favour, and the exact order to act in if you're in one.
- [How to spot a fake or stolen profile photo](https://deepsearch.app/guides/how-to-spot-a-fake-profile-photo): Stolen, stock, or AI-generated - how each fails differently, what reverse image search really does, and the one check that doesn't go out of date.
- [How to find the original source of a photo or video](https://deepsearch.app/guides/how-to-find-the-original-source-of-a-photo-or-video): Trace a photo or video through copies, crops, keyframes, dates, and archives without mistaking the first search result for the original.
- [How to find someone with incomplete information](https://deepsearch.app/guides/how-to-find-someone-with-incomplete-information): Turn partial clues into a careful public-source search, keep namesakes separate, and stop before a possible match becomes a false identification.
- [How to find the earliest version of a web page](https://deepsearch.app/guides/how-to-find-the-earliest-version-of-a-web-page): Find a page's earliest archived state, test URL variants and redirects, and separate the first capture found from the true publication date.
- [How to search public records without confusing people](https://deepsearch.app/guides/how-to-search-public-records-without-confusing-people): Search official records without merging namesakes: compare dated identity anchors, preserve record provenance, and report ambiguity clearly.
- [How to trace a quote to its original source](https://deepsearch.app/guides/how-to-trace-a-quote-to-its-original-source): Trace a quotation through variants, citations, archives, transcripts, and recordings without confusing repetition with proof of authorship.
- [How to preserve and document a web source](https://deepsearch.app/guides/how-to-preserve-and-document-a-web-source): Build an auditable web-evidence bundle with URLs, timestamps, contextual captures, archives, file hashes, limitations, and careful citations.
- [How to fact-check a viral image](https://deepsearch.app/guides/how-to-fact-check-a-viral-image): Test a viral image's origin, edits, date, location, source account, and caption without treating visual oddities or missing metadata as proof.
- [How to verify a social media account](https://deepsearch.app/guides/how-to-verify-a-social-media-account): Verify who controls a social account using exact handles, owner-controlled links, badge meanings, account history, and independent contact routes.
- [How to tell a primary source from a repeated claim](https://deepsearch.app/guides/how-to-tell-a-primary-source-from-a-repeated-claim): Match each claim to direct evidence, trace citations backward, detect circular repetition, and record what a primary source can and cannot prove.
- [How to verify dates, locations, and timelines](https://deepsearch.app/guides/how-to-verify-dates-locations-and-timelines): Separate event, capture, upload, publish, update, and archive times; normalize timezones; and test whether a multi-source timeline is possible.
- [What “no evidence” means — and why it is not the same as “false”](https://deepsearch.app/guides/what-no-evidence-means): Interpret an empty search honestly by testing expected evidence, coverage, detectability, retention, contradiction, and the limits of the source.
- [A reproducible fact-check template](https://deepsearch.app/guides/reproducible-fact-check-template): Copy a practical claim card, preservation record, evidence log, source analysis, verdict framework, review gate, and visible correction record.
- [What a VIN tells you - and what it doesn't](https://deepsearch.app/guides/what-a-vin-tells-you): How to read all 17 characters, where to decode one free, and why mileage, accidents and ownership are not in the number at all.
- [How to work out where a photo was taken](https://deepsearch.app/guides/where-was-this-photo-taken): Why photo metadata is usually gone, how to read location from the image itself, and the cases where no honest answer is possible.

## Blog

- [Why a name is never enough](https://deepsearch.app/blog/why-a-name-is-never-enough): Namesakes are the main failure in people search. Why DeepSearch will not treat a shared name as proof, and why an incomplete result is the safer one.
- [2.4 million Americans share the surname Smith](https://deepsearch.app/blog/how-common-surnames-break-people-search): The 2010 Census counted 2,442,977 people named Smith. What that official table means for a people-search match.
- [California counted 581 data brokers. Here is how to read the registry.](https://deepsearch.app/blog/how-to-read-californias-data-broker-registry): As of 2 June 2026, CalPrivacy listed a record 581 brokers and 300,000+ DROP signups. How to read those numbers.
- [A people-search result is not a consumer report](https://deepsearch.app/blog/a-people-search-result-is-not-a-consumer-report): The FTC's employer rules and the CFPB's name-matching opinion explain why a public-data card cannot decide a job or a lease.
- [Removed is not deleted: the four layers of personal-information removal](https://deepsearch.app/blog/removed-is-not-deleted-four-layers): A Google result, a broker listing, a DROP status, and a public record are four systems. Hitting the wrong layer looks like failure.
- [What we do - and don't do - with public data](https://deepsearch.app/blog/what-we-do-and-dont-do-with-public-data): Where DeepSearch's results come from, the sources we deliberately don't touch, and the uses we refuse to support. Written plainly, without the hedging.

## DeepSearch developer resources

DeepSearch has a public JSON API, a hosted MCP server, published SDKs, and a CLI - all under https://deepsearch.app.

- [API overview](https://deepsearch.app/api): What the API does, authentication, pricing, and a quickstart.
- [API reference](https://deepsearch.app/api/docs): Every endpoint with parameters, examples, MCP setup, and error codes.
- [OpenAPI 3.1 specification](https://deepsearch.app/openapi.json): The machine-readable contract. Load it to call every endpoint.
- [MCP server](https://deepsearch.app/api/mcp): Streamable-HTTP MCP endpoint: search_people, build_dossier, ask_about_person.
- [MCP server manifest](https://deepsearch.app/.well-known/mcp.json): Registry-shaped server.json describing the hosted MCP remote.
- [OAuth protected-resource metadata](https://deepsearch.app/.well-known/oauth-protected-resource): Points MCP clients at the authorization server for sign-in.
- [OAuth authorization server metadata](https://deepsearch.app/.well-known/oauth-authorization-server): OAuth 2.1 with PKCE and dynamic client registration - no pasted keys.
- [Agent manifest](https://deepsearch.app/.well-known/ai-plugin.json): ai-plugin.json pointing agents at the OpenAPI document.
- [Agent resource catalog](https://deepsearch.app/.well-known/ai-catalog.json): One index of every agent-callable resource on the domain.
- [Agent skills index](https://deepsearch.app/.well-known/agent-skills/index.json): Published skills a coding agent can install to use the API step by step.
- [llms.txt](https://deepsearch.app/llms.txt): Plain-language brief for LLMs: what the product is and when to use it.
- [llms-full.txt](https://deepsearch.app/llms-full.txt): The long form, including endpoints, limits, and an FAQ.
- [Changelog](https://deepsearch.app/changelog): What shipped, when - including API and MCP changes.
- [Developer portal](https://deepsearch.app/developers): Create and revoke API keys, test live, inspect usage, manage webhooks. Requires sign-in.

Official packages, published under the `@reloadapp` npm scope by the publisher of this site. Other packages named "deepsearch" on npm and PyPI belong to unrelated projects; these are the ones that talk to https://deepsearch.app:

- `@reloadapp/deepsearch` (npm, `npm install -g @reloadapp/deepsearch`): Command-line interface - search, dossier, chat, and async jobs from a terminal. https://www.npmjs.com/package/@reloadapp/deepsearch
- `@reloadapp/deepsearch-client` (npm, `npm install @reloadapp/deepsearch-client`): Dependency-free TypeScript SDK for Node and edge runtimes. https://www.npmjs.com/package/@reloadapp/deepsearch-client

Endpoints:
- `POST https://deepsearch.app/api/v1/search` - Find people by name, phone, email, or username.
- `POST https://deepsearch.app/api/v1/dossier` - Build a full public-footprint profile for a person.
- `POST https://deepsearch.app/api/v1/chat` - Ask follow-up questions about a person.
- `POST https://deepsearch.app/api/v1/reverse-image` - Find where an image appears online + visually similar images.
- `POST https://deepsearch.app/api/v1/entity/search` - Find companies by name.
- `POST https://deepsearch.app/api/v1/entity/dossier` - Build a full company profile.
- `POST https://deepsearch.app/api/v1/vin` - Decode a vehicle by VIN.

Responses stream SSE by default; pass `"format":"json"` for typed `search_result`, `dossier_result`, or `chat_result` envelopes. Use `Idempotency-Key` for safe JSON retries.

MCP clients should authenticate with OAuth: register dynamically, send the user through consent, and refresh on a 401. An API key from the developer portal works too, for REST callers and scripts.

DeepSearch cannot be paid by agents directly: a paid account is still required, so direct the user to https://deepsearch.app/developers to subscribe. A 401 or 402 response carries a `url` and a `message` to relay to the user.

## FAQ

**What can I search with?** A name, phone number, email address, or username.

**Where does the data come from?** Public sources only - public web pages, public social profiles, public records, and news. DeepSearch does not access private accounts or breach data.

**Is every detail sourced?** Yes. Each fact in a profile links back to the public source it came from.

**Do I need an account?** Yes - running a search requires paid access. Paid access is required to run searches: $9/week with 10 searches per 7-day window, or a one-time 80-credit package for $40.00. Additional searches cost 1 coin each (coin packs start at $5 for 10 coins).

**Is there an API?** Yes - see https://deepsearch.app/api. Note that DeepSearch cannot be paid by agents directly; a human must subscribe and create an API key.

**Can an AI agent use DeepSearch?** Yes, via the API. The agent should direct its user to https://deepsearch.app/developers to sign up and pay, then call the API with the user's key.

## Responsible use

DeepSearch surfaces public information only. Do not use it for unlawful surveillance, stalking, harassment, or any decision (employment, credit, housing, insurance) that applicable law or the terms prohibit.