Co-founder & COO @ Squadbase. Building AI data workflows. Product design, engineering, and the decisions behind them.
Tokyo
Joined May 2020
- Tweets101
- Following89
- Followers72
- Likes154
Imagine asking an AI assistant to dig into one number in a weekly report.
At Squadbase, I’m exploring how that request could guide what it looks into later. It’s a design goal, not a finished workflow.
An AI can help prepare a weekly report without having permission to change the source data.
That’s a boundary I want to preserve as we design workflows at Squadbase: access to read should not imply access to write. People should remain responsible for the final action.
In a Squadbase beta pilot, the AI suggested looking at related pages alongside a page a website operator had updated. It gave them another place to look when deciding what to change next.
In our Squadbase prototype, receiving an AI-generated dashboard is only part of the workflow.
Seeing it can prompt a more specific request: “Make this a slide” or “Change just this part.”
We’re exploring how notifications and chat can support that back-and-forth.
At Squadbase, currently in beta, we’re connecting data loading, analysis, and delivery with Automation. For a weekly report, those handoffs matter as much as the chart.
squadbase.dev/en/blog/introd…
Keita Mitsuhashi retweeted
[We compared Streamlit and Plotly Dash]
- Streamlit: A framework that lets data scientists build web apps
- Dash: A framework for building scalable BI dashboards
Want to learn more? Check it out 👇
squadbase.dev/en/blog/stream…
RT @squadbaseHQ: Launch Week Day 4 🚀
We’re thrilled to launch our AI App Showcase at Squadbase!
🔹 Sales Coach AI Chat (Streamlit + Attio)
🔹…
Keita Mitsuhashi retweeted
Launch Week, Day 3 🚀
We just rolled out “Deploy @nextjs on Squadbase”
Why not Vercel? — Squadbase offers built-in user authentication, log monitoring, and more for private team deployments—no extra code required!
🔗Read the full post
dub.sh/myH74H2
#LaunchWeek
Keita Mitsuhashi retweeted
Launch Week, Day 2 🚀
We just rolled out “Deploy @streamlit on Squadbase”
Import your GitHub repo, push code, and ship. Built-in auth, role-based access, logs, and analytics come standard—no extra code. Move from prototype to production in minutes!
squadbase.dev/en/blog/deploy…
#Streamlit #LaunchWee
Keita Mitsuhashi retweeted
Launched Squadbase today 🚀—the new cloud platform for AI apps built with Streamlit & Next.js. GitHub integration for auto-deploy, user auth, RBAC, logging & feedback collection included!
🧵 More in our blog about the journey & challenges solved.
#LaunchWeek
Keita Mitsuhashi retweeted
v0.3.0 Release Highlights 🎉
🚀 Deploy to AWS and GCP. Customize CPU, memory, and region settings.
🚀 Supports @streamlit deployment.
🔧 Compatible with @shadcn/ui library / UI components updated.
💪 Up to 5 Viewer users now available on the free plan!
morph-data.io/changelog
Posted a tutorial to transfer data from @NotionHQ to @pinecone using @triggerdotdev to make internal data "AI-ready"
dev.to/morph-data/making-not…
Keita Mitsuhashi retweeted
Morph’s cloud platform now features CI/CD automation integrated with Github! No more setting up Docker on your local environment; just a simpler and more efficient workflow for developing and using internal apps.
morph-data.io/blog/introduci…
Thank you for partnering with us at the @BuildersWeekend hackathon! We're thrilled that so many of you tried out the Morph framework!
Thanks to your questions, we merged 19 PRs into our docs – thanks for your contributions!
docs.morph-data.io/
#BuildersWeekend2025
Sharing my great experience building shadcn CLI compatible components: dev.to/keitam83/building-rea…
