Ploid: people data for AI agents

Ploid: The Living Index of People

Frontier people search across a living index of more than 2 billion people

https://ploid.com/

Ploid is a people data provider built for AI agents first and humans second. It claims an index of more than 2 billion people, assembled from every public source a person shows up in, and serves it through a REST API, a CLI and a hosted MCP server. Ask Claude or ChatGPT "find CFOs who scaled SaaS companies" and the agent calls Ploid instead of a human clicking filters in Apollo.

It launched publicly on 28th October 2026. Co-founder and CEO Manny Hernandez announced it on X:

I think the world's data should belong to everyone, so we indexed over 2 billion people globally. Ploid turns the public web into one profile for every person and company on earth, and serves it to you and your agents for cents.

The 70-second launch video by Ploid, from the post below: Manny Hernandez on stage, then the dataset builder, a results table with a "Match" check per criterion, and the Claude connector.

"Introducing Ploid. I think the world's data should belong to everyone"

@m9rcn (Manny Hernandez) on X, 28th October 2026

https://x.com/m9rcn/status/2115499043514445974

The company

A 2-person company with a 2-billion-person index. That suggests most of the data is crawled and linked by machines, then bought in for contact details. The privacy policy confirms it: contact fields come from third-party vendors such as FullEnrich, and the rest from Ploid's own crawling and identity matching.

What it does

The pitch is the Superprofile: 1 record per person that links professional profiles, social accounts and personal sites, and that an agent can reason over, including posts, photos and talks. Traditional providers sell static rows that were true weeks ago. Ploid says it retrieves live.

The results table in the video is the interesting part: each row shows a Match per criterion, so you see why a person made the list instead of trusting a filter.

How to use it with an agent

MCP is the main door. Add https://api.ploid.com/mcp as a custom connector in Claude (Customize, Connectors) or in Claude Code. It signs in with OAuth 2.1, so no API key is copied anywhere. The first MCP connection without a workspace creates a free 30-day trial, no card.

The MCP server has 5 tools:

Research runs in the background and can take minutes. The tool returns a run_id at once, and the agent polls every 5 seconds or more. A person Ploid can't identify with confidence comes back as found: false with suggested clues, and costs nothing. That's the right design: no same-name biography of the wrong person.

The CLI needs Node.js 20 or newer:

curl https://ploid.com/install -fsS | bash
ploid login
ploid --json search "AI security founders in San Francisco" --limit 10

The REST API lives at https://api.ploid.com/v1, with an OpenAPI 3.1 spec and Bearer-key auth. Rate limits run from 10 requests a minute on the free tier to 1,000 on Scale. /v1/socials and /v1/linkedin/* cap at 60 a minute and 600 an hour.

Pricing

• Credits are per workspace, shared across the app, API, MCP and CLI, and they don't roll over.
• Only hits are billed. A contact field that comes back not_found costs nothing.
• "For cents" depends on the call. A list is cheap: 1 credit per 10 people is 2 cents a person on Growth. A deep research_person is 25 credits, so $5 on Growth and $3.75 on Scale. A phone number is 10 credits, so $2 on Growth.

Fields returned and what they cost

1 credit is 1 ACU, Ploid's usage unit. The docs quote a "face value" of $0.10 per ACU (25 ACU = $2.50), but no plan sells at that rate: the real price is $0.20 on Growth and $0.15 on Scale. The dollar columns below use those 2 rates.

Credits per call

Ploid doesn't publish the price of 4 calls: the instant POST /v1/search, POST /v1/enrich per contact field, POST /v1/socials per platform, and the /v1/linkedin/* reads. The response tells you after the fact, in meta.credits_charged or meta.usage.

Fill columns, per found cell

POST /v1/datasets is the only place with a published price per field, and it bills only found cells:

What each call returns

Search and People Sets, per person:

• person_id, which /v1/person and /v1/enrich accept
• name, title, company, location, linkedin_url, profile_url, socials
• confidence and score: the rank in this response, not a probability
• filter_match: 1 entry per filter, with the value it matched
• 1 verification entry per criterion, with the evidence, on auto and deep searches
• identity_verified: false, until /v1/person confirms the person
• contact_status: "requires_enrichment": no emails or phones, those cost extra

Person research (/v1/person), per person:

Every history, education, platform and interest entry carries a timeframe (dated, undated, historical) and an evidence_statement.

Contacts (/v1/enrich): request any of work_email, personal_email, phone and linkedin_url. Each comes back with value, status (found, not_found or unavailable), confidence, source (for example fullenrich) and last_seen. Mailing addresses are "coming soon".

Fill columns (/v1/datasets), with the built-in columns:

socials.* covers LinkedIn, X, Instagram, YouTube, TikTok and GitHub. page_text returns the company site's text, up to 20,000 characters. Every cell has a value, a 0 to 1 confidence, up to 3 evidence pages with a quote and a read date, and, for emails and phones, a deliverability check: valid, risky or unknown.

A custom.* column on a person runs the reviewed research: Ploid's own example asks for hobbies, favourite sports team and hometown.

LinkedIn profile (/v1/linkedin/profile): the whole public profile. Name, headline, summary, location, industry, follower and connection counts, photos, every position, education, skills with endorsements, languages, certifications, awards, projects, recommendations, volunteering, publications, courses, patents, and the open-to-work and Premium flags. The same family reads posts, comments, company pages and LinkedIn people search.

Social profiles (/v1/socials): 1 public profile from LinkedIn, X, Instagram, TikTok, YouTube, GitHub, Reddit or Facebook. The docs don't list the profile fields.

What it costs in practice

Map 25 people at 1 account, with a work email and a mobile for each, and assume every cell is found:

Phones are 90% of the bill. So:

• Ask for phones only for the people you'll call. Growth's 1,000 monthly credits buy 100 phones, or 1,000 work emails.
• Resolve before you research. A name lookup (1 credit) or resolve (2) confirms the right person before a 25-credit profile.
• Cap the spend. Set daily and monthly budgets per API key, a budget.credits cap per People Set, and an Idempotency-Key on every POST, so a retry never bills twice.

The GDPR question

This is the part to read before using it from Germany.

• The index holds personal data, not only work data. The demo Superprofile on Ploid's own home page links Instagram, Facebook and a personal site, and shows "Family Vacation to Italian Coast" in the activity feed. The privacy policy says Ploid focuses on professional data. The marketing shows more.
• Nobody in the index was asked. Ploid relies on GDPR Article 14's "disproportionate effort" exception: it publishes its policy instead of telling each of the 2 billion people that it holds a profile on them.
• Opting out is by email. Access, correction and deletion requests go to 1 co-founder's mailbox. There is no self-serve form yet.
• The risk sits with the customer. The terms say you alone are responsible for lawful use, including anti-spam and privacy law. In Germany, a cold email to a contact found here still needs consent or a defensible legitimate interest, whatever the tool.

Against what I already use

Ploid isn't a new data source for contact details: part of what it returns is FullEnrich, which I already use through lookup.py. What's new is the shape. A list of decision makers, verified criterion by criterion, from 1 sentence in an agent, without exporting a CSV.

Verdict

Worth the free 30-day trial through MCP. The test: map the buying committee of 1 real account with find_decision_makers, and compare the list with Apollo and my own LinkedIn connections. Keep research_person for the calls that matter, at $5 a run.

What would stop me: a 2-person company holding personal profiles of 2 billion people under a US-first privacy model. For EU prospecting, use it to find and qualify people, and keep the outreach on a lawful basis I can defend myself.

Related: see Data providers for B2B sales lead generation, see Apollo and see Clay.

Further Reading

MCP server | Ploid documentation

Connect Ploid to Claude, ChatGPT and other MCP clients: tools, costs, permissions

https://ploid.com/documentation/mcp

Ploid: The people data company for the agent era | Y Combinator

YC Fall 2026 company profile and founders

https://www.ycombinator.com/companies/ploid

• Pricing, Ploid
• API documentation, Ploid
• Privacy policy and data rights and requests, Ploid
• Our vision, Ploid's company page