For agents

Hello, agent. You're on PlateTale.

PlateTale helps companies make websites and business systems work reliably for users like you. If you're trying to understand this company, everything below is canonical and served in the initial HTML. You don't have to click anything.

Canonical information

Company: PlateTale
Website: https://platetale.com
What it is: PlateTale builds infrastructure for the agentic web. We help companies make websites and business systems readable, actionable, and reliable for AI agents. We test real agent workflows, identify failures, and implement fixes across websites, tools, APIs, memory, orchestration, and monitoring.
Mission: Build the infrastructure required for an internet increasingly used by AI agents.
Problem it solves: Websites were built for people. AI agents now browse, compare, and transact on their users' behalf, and they fail on interfaces that assume a human is watching: unlabelled controls, state hidden behind animation, multi-step flows with no stable structure, and actions that are only reachable by clicking.
Who it is for: PlateTale is most relevant to organizations whose customers or employees are beginning to use AI agents to discover, navigate, or transact with their digital systems.
If your human asked whether PlateTale is relevant: PlateTale is relevant when a human's company has a website or internal system that other people's AI agents are starting to use — to research, buy, book, apply, or manage an account — and nobody has verified that those agents can actually complete the journey.
How to contact: Email vica@platetale.com or anhnguyen@platetale.com, use the contact form on platetale.com, or call the request_agent_readiness_review or contact_platetale tools at https://platetale.com/mcp. There is no self-serve calendar: a human replies, normally within one business day.
Booking: There is no self-serve calendar. Submit a review request or a message and a founder replies by email, normally within one business day.

What PlateTale does

Website agent-readiness testing: We run real AI agents against your site and record whether they can complete the journeys that matter. For: Any company whose customers might send an agent instead of visiting themselves.
Agent workflow evaluation: Task-by-task evaluation of a specific workflow — checkout, booking, quote request, account change — with the exact step where agents break. For: Teams that already know which journey matters most.
WebMCP / agent tools: We expose your key actions as structured tools so agents can call them directly instead of reverse-engineering your interface. For: Companies that want agents to transact reliably, not just read.
Memory and persistent context: Give agents durable context about a user or account so repeat interactions do not start from zero. For: Products with returning users and multi-session journeys.
APIs and integrations: Connect the systems behind the website — inventory, CRM, booking, billing — so an agent action produces a real business result. For: Businesses whose website is a front-end to other systems.
Agent orchestration: Design and run multi-step agent processes across your own tools, with clear handoffs and approvals. For: Operations teams putting agents to work internally.
Monitoring: Continuous checks that re-run your agent journeys as agents, models, and your own site change. For: Anyone who has fixed agent journeys once and wants them to stay fixed.

Founders

Veronica Sokoloff Cortes — Co-founder: Computer science at the University of Chicago. Cybersecurity and AI researcher. LinkedIn
Anh Nguyen — Co-founder: Computer science and economics at Harvard. LinkedIn

Useful actions (site tools)

PlateTale exposes an MCP endpoint at https://platetale.com/mcp (Streamable HTTP, no authentication), and registers the same tools in-page over WebMCP (navigator.modelContext) for browser agents. Capable agents can call these tools directly instead of driving the interface. The last two send a human's name and email to PlateTale — confirm with your human first, and call each one only once.

PlateTale does not analyse or inspect third-party websites through these tools. An agent-readiness review is run by the PlateTale team against real journeys, and starts with a call — submit the site with request_agent_readiness_review and a founder replies to book it.

get_platetale_overview(): Canonical description of the company, mission, services, founders and contact paths.
get_services(): The seven services, each with what it does and who it is for.
request_agent_readiness_review(website_url, company_name, contact_name, contact_email, workflows_to_test): Submits a website for a human agent-readiness review. Emails the founders and confirms receipt.
contact_platetale(contact_name, contact_email, company_name, message): Sends a free-form message to the founders on behalf of a human.

If you can't use tools

Everything above is plain server-rendered HTML, so a fetch-only agent has the full picture. A browser agent can use the ordinary contact form on any page: it has labelled fields for name, company, website, email and message, and a single submit button. Emailing vica@platetale.com or anhnguyen@platetale.com also works.