In a Plug & Play webinar, Supermetrics showed Claude analyzing Facebook ads, launching campaigns and publishing live dashboards, while Precisely showed Claude finding and fixing SAP material data through an MCP server. Both demos moved AI from reading data to acting on it.
Key takeaways
- Marketing tests read-write AI first. Kaeding argued that marketing's mix of analysis and action makes it the canary in the coal mine for BI.
- Adoption lags pressure. Supermetrics research cited by Kaeding found 90%+ of marketing teams under pressure to adopt AI, but only 6% fully implemented.
- Analysis and action in one chat. Supermetrics' Claude connector pulled Facebook ads data, then created a live campaign in a test account.
- Static AI dashboards need a home. Supermetrics showed a tool, still in internal testing, that turns Claude artifacts into live, shareable dashboards.
- Governed write-back to SAP. Precisely's MCP server let Claude find and fix SAP material weights without opening the SAP interface.
Two vendors showed AI assistants doing more than answering questions in Data Science Connect's Plug & Play: Live AI Workflows in Action webinar, a hands-on series sponsored here by Supermetrics and Precisely. Both demos ran through Claude, and both used it to write changes back into business systems: ad campaigns in one case, SAP master data in the other.
Supermetrics was represented by founder and board chairman Mikael Thuneberg and Evan Kaeding, Global Director of Solutions Engineering. Scott, a senior sales engineer at Precisely, showed the second demo. Here's what each demonstrated.
Why is marketing the "canary in the coal mine" for AI and BI?
Kaeding opened with the argument behind Supermetrics' session title, "AI is disrupting BI." Marketing, he said, needs a high degree of bidirectionality. Teams read data from CRMs, ad networks and analytics tools to judge campaigns and creative, then write back by planning, launching and optimizing campaigns. The most sophisticated marketers, he said, want to do both "in the same workflow."
He cited Supermetrics' own research finding that more than 90% of marketing teams feel pressure from the C-suite and board to adopt AI, but only 6% have fully implemented it in their workflows. "High expectations, low real adoption," he said. He also pointed to Harvard Business Review reporting on "AI brain fry," which he described as a form of burnout, saying more than one in four marketers report it.
Supermetrics: How do you analyze and manage ad campaigns inside Claude?
Thuneberg, who said he has spent recent months building Supermetrics' AI ecosystem products, showed the company's connector for Claude, which he described as a one-click install that needs no existing Supermetrics account. Supermetrics data is also available in ChatGPT, Gemini and Microsoft Copilot, he said, but Claude is where the company sees heavy analytics use.
He asked Claude to list available data sources, then to analyze last week's Facebook ads performance and offer suggestions. Claude asked which account to use, pulled campaign data and compared it with the prior week. "No one wants to go through all of these hundreds of different dashboards," Thuneberg said.
He then had Claude create a summer promotion campaign in a test account, which appeared live in Meta's ad manager, and create ads with images uploaded, picked from Google Drive or AI-generated. Asked about trust, Thuneberg said the best models have become much better with numerical data and that he no longer sees hallucinations with them, while acknowledging "it's not a 100% guarantee."
How can AI-generated dashboards become live and shareable?
Claude can build dashboards on the fly, Thuneberg said, but they are static HTML artifacts that are hard to share. Supermetrics built a tool to upload them into its system as live dashboards that refresh with new data, can be shared and can be edited through a chat panel. He said the feature was still in internal testing at the time of the webinar.
He also rebuilt a Looker Studio dashboard from a screenshot, which he suggested as a way to migrate from other BI tools, and showed a Facebook ads budget manager built in about two minutes that pushes budget changes to the platform when a user clicks save.
Kaeding said the tool grew out of Claude artifacts being passed around in Slack, emails and PDFs where they could no longer be updated. The new version, he said, is "sharable, it's controllable, it's auditable."
Asked whether marketers should automate existing workflows or redesign them, Thuneberg said both, depending on how far teams trust AI. Most are still at the human-in-the-loop stage, he said, with a confirm button before actions go live. Kaeding said Supermetrics keeps a human in the loop for its own live campaigns and recommended manual approvals and an audit trail.
Precisely: How can an MCP server simplify SAP master data management?
Scott, who supports Precisely's Automate portfolio for SAP data processes, said that in SAP even reviewing and fixing a material record means moving between transaction codes. He showed MM03 for read-only viewing and MM02 for changes, re-entering the same material and selecting views each time. Because of network problems before the session, he used recorded backup videos.
He built the workflow end to end. Automate Studio recorded his steps through the SAP screens and turned the fields into a JSON script with little to no coding. He validated it and submitted it to the Automate Evolve platform, which exposed it as an API and assembled it into a tool for the MCP server, linked to an LLM connection and SAP credentials.
In Claude, he confirmed the Evolve connection, listed the available tools, and asked for materials in a particular plant along with their gross and net weights. Claude flagged three records whose gross weights far exceeded their net weights and offered to update them. After one update, it showed the old and new values side by side, all without opening the SAP interface.
The experience feels conversational, he said, but "the execution behind the scenes is still structured and governed," with SAP remaining the system of record.
See the next Plug & Play session
Plug & Play is one of four series in Data Science Connect's 2027 lineup of 24 free live webinars, alongside the AI × Enterprise Panel Series, AI Demo Days and Agentic AI Showcases. See the webinar schedule for upcoming live demos.
Frequently asked questions
Can Claude manage advertising campaigns?
In Supermetrics' demo, Claude used the Supermetrics connector to analyze Facebook ads performance and create a new campaign and ads in a test account.
How does an MCP server connect AI to SAP?
In Precisely's demo, scripts recorded in Automate Studio were exposed as APIs and tools in the Automate Evolve MCP server, which Claude used to query and update SAP material records.
Should marketers keep a human in the loop with AI?
Both Supermetrics speakers said yes for now, recommending confirmation steps, manual approvals and audit trails before campaigns go live.
Why is marketing called the canary in the coal mine for AI?
Kaeding said marketing requires both reading data and writing changes back to many platforms, so it shows early how AI reshapes BI.
Watch: Plug & Play: Live AI Workflows with Supermetrics & Precisely | Webinar Replay on YouTube
Event: Webinars · More coverage: Webinars insights
