It was a busy May here at Bobsled. Spring has sprung for some of our team. Winter is on its way for others. But for everyone, the sled is moving faster than ever.

The big focus in May was all about control in production. We gave every data agent in Bobsled a way to talk to other agents (Analytics MCP), playbooks for common questions (Data Skills), and saved execution graphs (Workflows) so they don't reinvent repeated analyses every time they're asked.

Plus, we shipped a ton of the usual quality of life improvements — including expanded support for the enduring Microsoft Office suite.

Here is a rundown of what went live and what each one does.

Analytics MCP

What it is: A single governed endpoint that gives external agents access to the Bobsled data agent wherever they work.

Why you should care: At Bobsled, we know that people asking questions of the data agent directly is just the start for agentic analytics. More and more, data agents are answering questions from other agents — whether it's a business user working on a presentation in Claude or an autonomous agent working on a background task. Every data agent in Bobsled now comes with its own MCP, so both internal and external stakeholders can let their agents ask questions — wherever they work.

Key features

• Let users chat, anywhere. Your stakeholders can put their agents to work on your data from the tools they already use, like Claude and ChatGPT, instead of switching into another app to ask.
• Integrate with other agents. Beyond people in a chat window, autonomous and background agents can ask, decide, and act on your data programmatically through the same endpoint.
• Unified governance and control. Every agent connects through one governed endpoint that already understands your semantic model and business context, so you control how your data is accessed and interpreted no matter who, or what, is asking.

Read more: Introducing the Bobsled Analytics MCP.

Data Skills

What it is: Repeatable prompts that let data teams teach agents how to handle complex analyses on demand, codified once and reusable forever.

Why you should care: Every data team has a handful of questions it answers over and over, and a right way to answer each one. Until now, that know-how lived in an analyst's head or a stored procedure they ran by hand. Data Skills turn it into a playbook the agent runs itself. A sales leader asks for a quick run-down on March, the agent recognizes it's QBR prep, pulls revenue, CAC, and churn, and drops the numbers into three slides in the approved format, without anyone touching SQL. The analysis your team has already worked out gets applied the same way every time, on demand.

Key features

• Create skills as you work. Notice yourself running the same prompt every week? Tell the agent to save it as a skill, and it invokes that prompt on its own next time.
• Share across the organization. Keep skills private or publish them to your team. Skills built by one analyst become reusable workflows for everyone, and data teams govern which ones get promoted.
• Take action, not just answers. Pair skills with tools to build decks, generate PDFs, or update downstream systems in the same run.

Read more: Introducing Data Skills.

Workflows

What it is: A saved execution graph that gives you the flexibility of an agent with the predictability of a pipeline.

Why you should care: Skills give the data agent a way to reuse trusted prompts and business rules, but until now, the agent still recreated the step-by-step plan — choosing which queries to run, in what order, and how to format the result — each time it was invoked. That's the right behavior when a question needs judgment, and the wrong behavior when the answer needs to look the same every Monday morning. Workflows make the execution graph a primitive, so the steps stay fixed, the inputs are validated, and the outputs are reproducible.

Key features

• Build with the workflow agent. Describe what you want to automate and the agent drafts the execution graph, picking from built-in steps (Run SQL, Execute Python, Fetch URL, Web Search, LLM Call, and Write File) plus any deployed Library Tool your Skills already use, then wiring step dependencies and templating inputs for you.
• Tune in the editor. A graph view to reorder dependencies, edit each node's arguments, and wire data between steps. Independent steps run in parallel, and conditions skip steps that should not run.
• Invoke anywhere. A slash command in chat, on a schedule, or from a Data App tile, a Skill, or an external event.

Read more: Introducing Workflows.

Other quality of life improvements

Alongside the launches, we shipped a handful of smaller upgrades in May that make day-to-day work smoother.

• Bigger CSV exports: CSV export now supports up to 500,000 rows, up from 100,000, so you can pull and share much larger datasets.
• Office formats, including Excel: You can now attach and export .xlsx, .pptx, and .docx files with the analysis and context agents.
• Data App upgrades: Data Apps now support richer, more polished dashboards, including agent-replay tiles that show LLM chat results inside a Data App, plus a set of UI and UX improvements to readability, interactions, attachments, and rendering.

That's the view from May. There's more on the way no matter what the season looks like where you are. If you're working on making your data agent-ready, book a demo or reply and tell us what you're building.