Preset Chatbot

What if every AI agent on your data team already knew your exact Preset and Apache Superset™ workflows and refused to do anything unsafe? In this live session, the Preset team pulls back the curtain on Preset Agent Skills: reusable, versioned plugin packages that teach analytics agents how to discover, operate, and govern your data stack. Evan Rusackas, Beto Dealmeida, Richard Fogaça, and Elizabeth Thompson walk through what skills are, why they make agent work repeatable and safe, and how the same skill can serve Claude Code, Codex, Cursor, Copilot, Cortex CLI, and Gemini. Watch this recorded live session for an inside look at how a small team packages best practices into install-once, update-like-code skills, with metadata-first discovery, permission-aware execution, and confirm-before-risk safety baked in, plus two live demos showing API skills and MCP skills in action against real Preset workspaces and dashboards. In this video: 0:00 – Welcome & intro 0:21 – Meet the Preset team: Evan, Beto, Richard & Elizabeth 1:07 – What are Agent Skills? Reusable playbooks that teach agents your workflows 5:37 – Why skills matter: consistency, best practices, right surface & safety checks 10:25 – Why Preset & Apache Superset users benefit: discover, operate & govern 11:53 – What's included: preset-mcp-skills, preset-api-skills & preset-cli-skills 14:21 – Safety is built in: metadata-first, permission-aware & confirm risk 15:32 – Tested & evaluated: deterministic scoring across API, MCP & CLI 18:49 – Where to get them + install (GitHub, marketplace & supported clients) 20:41 – Demo 1: API skills live — listing workspaces, resolving IDs & safe reads 29:16 – From prompt to a populated Superset dashboard 33:01 – Demo 1 recap: what the agent actually did 34:13 – Demo 2: MCP skills — discover, inspect & preview 34:30 – Where else to get skills: third-party skill sites (skills.sh, skillery.dev & more) 35:50 – Wrap-up: why scoped, safe skills matter for data teams 37:59 – Learn more: go deeper on Preset MCP & Superset MCP 38:49 – Q&A 41:50 – Closing remarks & how to get in touch Whether you're a data engineer, analytics lead, or anyone building internal AI tooling, this talk shows what becomes possible when you give your agents reusable, well-scoped skills that encode your best practices and keep your data safe, across every agent and surface you use. Recorded live on June 24, 2026. 🔧 Get the skills (open source): https://github.com/preset-io/agent-skills 🚀 Get started with Preset AI: https://preset.io Check out our upcoming events: https://preset.io/events Resources mentioned: Preset MCP server overview & docs: https://docs.preset.io/docs/preset-mcp-server Superset MCP service technical deep dive: https://preset.io/blog/superset-mcp-service-deep-dive SUP CLI (Superset CLI): https://docs.preset.io/docs/sup-cli Apache Superset on GitHub: https://github.com/apache/superset Also published to: https://www.skills.sh/preset-io/agent-skills · https://www.skillsdirectory.com · https://skills.re · https://skillery.dev Connect with us: Website: https://preset.io/ LinkedIn: https://www.linkedin.com/company/preset-io X: https://x.com/preset_data Instagram: https://www.instagram.com/preset_io/ Facebook: https://www.facebook.com/preset.io Building skills for your own data agents? Tell us which workflows you'd package first in the comments 👇 #preset #presetai #agentskills #aiagents #apachesuperset #mcp #modelcontextprotocol #agenticai #dataengineering #dataanalytics #businessintelligence #semanticlayer #llm #claudecode #texttosql #aianalytics #poweredbyapachesuperset #workflowautomation #anthropic #openai

What if your whole data team had an AI engineer running around the clock, connected to every system in your stack? In this live talk, Preset founder & CEO Max Beauchemin is joined by Head of Community Evan Rusackas where they pull back the curtain on DatAgor, Preset's internal AI data engineer, and introduces Agor, the open source agent orchestration platform Max built to make it all possible. Watch this recorded live session for an inside look at how a small data team stands up a sophisticated, long-lived agent with memory, skills, and access to everything from Preset MCP to data pipelines, Slack, HubSpot, and beyond. In this video: 0:00 – Welcome 0:37 – Meet the team: Max & Evan 2:21 – Today's agenda: anatomy of Preset's internal data agent 4:36 – State of AI: the big shifts from 2025 5:31 – MCP, Claude Code, Codex, Gemini CLI — the year agent tooling got real 7:47 – 2026 trends: orchestration, agent identity & the PDLC shift 9:30 – Agent identity: do your agents borrow your credentials, or have their own? 10:30 – The tire pile: RAG, IDEs & semantic search in the AI era 13:47 – Key premise: your agents are only as powerful as their interfaces 14:31 – Introducing DatAgor, Preset's internal AI data engineer 15:46 – Preset's data stack: why your agent needs access to context 16:43 – Blueprint: what a great analytics agent actually needs 17:19 – Analytics platform layer: charts, dashboards, datasets & semantic layers 19:10 – SQL access, data access policy & role-level security 20:22 – Enabling data teams: pipeline source code, lineage & data ops access 21:36 – Submitting PRs, improving docs, memory systems & skills 22:21 – Source & destination access: HubSpot, ticketing systems & the soft semantic layer 23:34 – Making your agent available on Slack 25:47 – Trust, security & the human-in-the-loop 28:14 – DatAgor's full architecture: what it connects to 28:48 – Introducing Agor: Preset's open source agent orchestration platform (https://agor.live/) 30:24 – Agor demo begins: the spatial board UI 31:26 – Under the hood: cards, git worktrees & multi-SDK support (Claude, Codex, Gemini) 33:06 – Zones, templates & standardizing prompts across your team 33:52 – Live: exploring Preset's marketing repo inside Agor 35:01 – Why Agor was built: the chaos of isolated local AI sessions 36:20 – Agor settings: boards, repos, assistants & centralized MCP config 37:35 – Artifacts: JavaScript mini-apps your agents create and share 39:00 – 70+ agents at Preset: how every team runs their own 39:40 – DatAgor on Slack: asking an agent to describe itself 40:45 – Live: triaging an ETL error via Slack and ejecting to an Agor session 41:19 – Scheduling: agents that produce weekly reports and work while you sleep 41:40 – Validating agent output: the case for multiplayer sessions 42:31 – Sales, ops & marketing agents at Preset 43:38 – How easy it is to create a new persisted agent in Agor 44:23 – Learnings from Preset Labs: agent modeling, scope & combining MCP + memory 44:38 – Getting started with Agor today (agor.live) 45:54 – Sneak peek: connecting Preset Chatbot to Agor as a control plane 47:06 – Q&A begins 47:22 – Q: Managing dashboards as code using agents and the SUP CLI 48:33 – Artifacts as a new primitive in Superset, and why dashboards aren't dead 50:23 – Q: Auto-populating dashboards and charts via MCP in an automated workflow 53:34 – Q: Guardrails, auditing & the trust problem in AI data agents 55:40 – Closing remarks & how to get in touch Whether you're a data engineer, analytics lead, or anyone building internal AI tooling, this talk shows what becomes possible when you give a well-scoped agent access to your full data stack, and the orchestration layer to manage it all. Recorded live on May 21, 2026. 🤖 Read the blog post DatAgor wrote about itself: https://preset.io/blog/meet-datagor-ai-data-engineer/ 🔧 Try Agor (open source agentic orchestration): https://agor.live 🚀 Get started with Preset AI: https://preset.io Check out our upcoming events: https://preset.io/events Stay informed about Agor: https://preset.io/contact-us-about-agor/ Resources mentioned: Preset MCP overview & docs: https://docs.preset.io/docs/preset-mcp-server Preset Chatbot: https://preset.io/blog/meet-preset-chatbot/ SUP CLI (Superset CLI): https://docs.preset.io/docs/sup-cli Apache Superset on GitHub: https://github.com/apache/superset Connect with us: Website: https://preset.io/ X: https://x.com/preset_data LinkedIn: https://www.linkedin.com/company/preset-io Instagram: https://www.instagram.com/preset_io/ Facebook: https://www.facebook.com/preset.io Building your own internal data agent? Tell us what you're connecting it to in the comments 👇 #preset #presetai #aianalytics #apachesuperset #mcp #modelcontextprotocol #agenticai #dataengineering #agor #agentorchestration #businessintelligence #dataanalytics #semanticlayer #llm #aiagents #poweredbyapachesuperset #texttosql #workflowautomation #openai #anthropic