Jan Curn
Founder & CEO of Apify
Prague, Czechia
Actions
Jan Curn is the founder and CEO of Apify (https://apify.com), the world's largest marketplace for web data tools that power (not only) AI agents with up-to-date data. He has a lifelong passion for software engineering, which earned him an MSc and PhD in computer science and eventually led him to founding Apify. Jan lives between SF and Prague, is active in the tech community in both cities, organizes meetups, and talks about software, startups, or AI.
Area of Expertise
MCP doesn’t suck — your agent does
Most AI agents misuse MCP and treat tools as prompt-time function calls: tool definitions and results are repeatedly injected into the context, tokens are wasted, and context rots. The result? Slower, less reliable agents, and the misleading conclusion that “MCP sucks, CLIs are better.”
To challenge this narrative and show how agents can get the best of both MCP and CLI, we’ve built `mcpc`, an open-source universal CLI client for MCP. It maps MCP operations to intuitive CLI commands, which agents quickly pick up through `--help` without external skills.
It turns out, CLI is the perfect local interface for agents to interact with MCP, giving them access to full protocol capabilities including modern features like code mode or progressive tool discovery through a single `Bash()` tool call, while leveraging MCP’s standard remote interface for server discovery, auth, payments, and access control.
To once and for all kill the MCP vs. CLI debate and show those two technologies are not exclusive but complementary, we’ll present evals comparing the performance of agents using naive MCP, modern MCP, native CLIs, other MCP CLIs, and `mcpc`, in various real-world scenarios.
Code mode is best served in the shell
MCP is exploding, but most agent integrations still treat tools as a prompt-time API, burning context tokens on tool definitions, re-copying intermediate results, and losing accuracy along the way.
In this talk, we’ll introduce 𝚖𝚌𝚙𝚌, a universal CLI client for MCP that puts “code mode” where it shines: the terminal. With persistent MCP sessions, JSON outputs designed for scripting, and an MCP proxy for sandboxing, mcpc is an invaluable tool for AI engineers working with MCP servers. We’ll live-demo a practical workflow that invokes multiple MCP servers in parallel, filters/transforms results locally, and then turns the interaction into a reusable script.
You’ll leave the session with a new Swiss Army knife for MCP in your toolbox—one that’s not tied to Claude Code, Codex, OpenCode, or any coding agent you might be using this month.
MCP doesn’t suck — your agent does
Most AI agents misuse MCP and treat tools as prompt-time function calls: tool definitions and results are repeatedly injected into the context, tokens are wasted, and context rots. The result? Slower, less reliable agents, and the misleading conclusion that “MCP sucks, CLIs are better.”
To challenge this narrative and show how agents can get the best of both MCP and CLI, we’ve built mcpc, an open-source universal CLI client for MCP. It maps MCP operations to intuitive CLI commands, which agents quickly pick up through --help without external skills.
It turns out, CLI is the perfect local interface for agents to interact with MCP, giving them access to full protocol capabilities, including modern features like code mode or progressive tool discovery through a single Bash() tool call, while leveraging MCP’s standard remote interface for server discovery, authentication, payments, and access control.
To once and for all kill the MCP vs. CLI debate and show those two technologies are not exclusive but complementary, we’ll present evals comparing the performance of agents using naive MCP, modern MCP, native CLIs, other MCP CLIs, and mcpc, in various real-world scenarios.
The rise of the agentic economy on the shoulders of MCP
Thanks to MCP and all the MCP server directories, agents can now autonomously discover new tools and other agents. This lays down the foundation for the future agentic economy, where businesses will sell to autonomous agents (B2A) and eventually agents will sell to other agents (A2A).
But one key part is still missing: agents do not have a standard way to subscribe to external services and pay for them.
In this talk, we’ll show how to give agents full autonomy to discover and pay for new external MCP-enabled services, even if those services don’t support it, using a little-known MCP server nesting capability. We’ll also cover how to monetize AI agents and the B2A/A2A business models.
How to scrape modern websites to feed AI agents
It was web scraping that quietly enabled the AI revolution.
All major generative AI models and large language models (LLMs) have been trained on massive datasets extracted from the web. But the story doesn’t end there. Today’s AI agents still rely on up-to-date information from online sources to stay relevant, reduce hallucinations, and go beyond their training data. Once again, the web becomes the richest source of context.
Yet scraping the modern web is far from trivial. From anti-bot protections to dynamic, client-side rendering and long-term maintenance issues, building scalable data pipelines has become increasingly complex.
In this talk, we’ll demonstrate how to:
- Set up resilient, maintainable data extraction workflows using cloud-based web scraping Actors,
- Clean messy HTML and structure it into meaningful content,
- Feed that data into a vector database to power a Retrieval Augmented Generation (RAG) pipeline,
- And integrate scraping tools directly into AI agents using the Model Context Protocol (MCP).
Whether you're building copilots, smart assistants, or knowledge-augmented LLM apps, this talk will equip you with the techniques and tools to make the web your AI's best friend.
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top