An AI developer, not a casual user
Lisarah teaches AI development to software engineers, freelancing for multiple companies out of Germany. Outside of teaching, she runs her own IT project management and consulting work — split between individual client projects and a company she is building with a co-founder. Claude, in her words, "is like my terminal for everything nowadays."
That background matters for how much weight her opinion carries here. She originally started using xTiles for an ADHD-friendly weekly planner — colors, tiles, a place to brain-dump so nothing got forgotten. Later, building a company with a co-founder, she chose it again, deliberately, as their shared workspace — this time picking it directly over Notion.
Replacing Notion as the shared source of truth
Her own words on the decision: "I also decided, hey, let's use xTiles for the two of us to make sure that we get everything in one place essentially — so we use it as a single source of truth, as an alternative to Notion basically."
Two concrete reasons came up clearly on the call. First, the practical one: a two-person company needs exactly one place both people trust, and xTiles became that place instead of Notion. Second, the reason that mattered more given how she works: xTiles projects can run scheduled automations directly inside them, not bolted on as an external script — which fit a workflow already built around Claude far better than a static database-and-page structure would.
Projects, not pages: how she structures the work
Everything lives in Projects — one per client, one for her own company — typically running six months to two years without ever fully "closing." Rather than fighting that always-open nature, she built directly on top of it. Inside her company's project, a scheduled task pulls in market signals and news and hands her co-founder a running lead-generation list — a real automation living quietly inside the project itself, not a separate script somewhere else.
She goes further than most: writing her own custom Claude skills and instruction files to reuse across sessions, and periodically pulling numbers or checking whether specific milestones already happened — confirming an email marketing send landed, for example — directly from inside the relevant project. It functions less like a task list and more like a lightweight operations layer sitting on top of a workspace that also happens to hold notes and plans.
What the Claude connector adds on top of that
Lisarah was among the first to test the newer xTiles MCP connector: it reads her projects and Planner, creates tasks, sets priorities, and — notably — can assign work to someone other than her. One of the first things she used it for was turning a loose "I want to focus on a few things this week" into an actual prioritized task list in her Planner. "It helped me a lot," she said — even if, as she put it, the main result was realizing "I have much more to do than I thought."
The feature she was most looking forward to next: piping meeting transcripts straight into Claude and having the output land back in xTiles automatically, instead of her current routine of copying transcripts out of Google and pasting them in by hand. "I can finally put my transcripts in Claude and everything will be created next time, so I don't have to copy-paste it all the time from Google."
The habit that actually saved her time: delegating to agents
Recording meetings straight into Claude and having it turn action items into scheduled tasks on her Google Calendar already saves her roughly 10 hours a week on its own. But the bigger shift came from a change in how she prompts: instead of doing tasks herself, she started asking Claude to divide large tasks across specialized agents and plan before executing — a habit that, combined with everything else, adds up to more than 20 hours saved most weeks.
There is a distinction she draws that is worth repeating: her team's founder once pushed back when everyone described AI's impact purely in terms of tasks completed — more emails sent, more items checked off. His question was simple: what results did any of that actually produce? That reframing stuck with her. Tasks finished is not the same thing as a project moved forward, and she now tries to judge her own AI workflow by the second measure, not the first.
Why Claude, specifically, over ChatGPT
Since this is a comparison piece, one more choice is worth including: she picked Claude over ChatGPT for reasons beyond xTiles. On raw output, her comparison was blunt — ChatGPT, next to Claude, "feels like talking to a three-year-old." She also mentioned Anthropic declining a defense-related government contract that OpenAI accepted as a factor in her decision, alongside the practical output difference.