AGNTCon + MCPCon North America 2027
Table of Contents
Basics
| Field | Value |
|---|---|
| Dates | |
| Location | San Francisco, California |
| Host | The Linux Foundation |
| URL | events.linuxfoundation.org |
| Edition | Second; the first was San Jose, October 2026 |
| CFP | Sessionize; cfp@linuxfoundation.org |
Dates
| Date | What |
|---|---|
| CFP closes, 23:59 PST | |
| CFP notifications | |
| Schedule announced | |
| Early-bird registration ends | |
| Event opens |
Submitting
Three submissions per person, maximum. Accepted speakers get one complimentary speaker pass and two attendee passes per submission. Panels must name every participant at submission, and the Linux Foundation does not accept all-male panels. Slides are required before the event.
| Type | Length | Speakers |
|---|---|---|
| Session | 25 min | 2 max |
| Panel | 25 min | 4 max incl. moderator |
| Lightning | 5 min | 2 max |
| Workshop | 90 min | --- |
| Live demo | 5 min | 2 max |
| Poster | 25 min | 2 presenters max |
Two notes from the CFP worth keeping in view when drafting. Sales and marketing pitches "are almost always rejected". And reviewers must be able to judge subject-matter expertise from the proposal itself: submissions that "rely heavily on AI generated or templated content often lack the specificity needed to evaluate technical depth".
Tracks
Six thematic:
- Agentic Engineering — harnesses, multi-agent coordination, interoperability, deployment at scale, local and self-hosted inference, cost, post-training, self-improving agents; context engineering, persistent memory, retrieval, knowledge graphs, tool discovery, agentic search
- Enterprise Adoption in Practice — pilots to production, ownership and governance, measuring outcomes, lessons from regulated environments
- Evals & Reliability — evaluation pipelines, regression testing, software verification, observability, debugging, incidents, failure recovery
- Security, Identity & Control — authn, authz, delegation, sandboxing, guardrails, human approval, threat models
- Agent Interfaces & Interaction — agent-ready applications, browser and voice agents, generative interfaces, computer use, human-agent collaboration
- Agentic AI 101 — foundations for newcomers
Ten project tracks: MCP, Skills, AGENTS.md, goose, agentgateway, Agent Router (the Envoy-based project formerly Envoy AI Gateway), A2A, Open Responses, OpenEnv, OpenSandbox.
Of interest here
The tracks that meet work on this site: Evals & Reliability names software
verification and regression testing outright; Project: AGENTS.md is the
convention this repo follows, including keeping instructions useful as a
project evolves; Project: Skills covers SKILL.md authoring, discovery and
loading; and Security, Identity & Control meets the bot-identity and
agent-trust notes.
Sponsors
Diamond: Google Cloud, Obot.ai, Solo.io, WorkOS. Platinum: Vultr. Silver: Descope, MCPJam. Bronze: AgentCat.
Notes
Notes pending.