From consumer apps and AI tools to taking over half-finished systems — this is what BotsUP genuinely owns and operates. Client cases are anonymized for NDA reasons, but the technical details and outcomes are real.
AI-Native Enterprise Admin Platform for a Tech Company
Challenge
This client refused to repeat the usual enterprise pattern — strategy in Notion, issues in Linear, accounting in a third tool, contracts in Google Docs, invoicing on yet another platform, with people shuttling data between fifteen tools every day. They did not want a chat bubble bolted onto one of those tools. They wanted an operating system designed from day one for AI agents: agents that could read strategy, create projects, issue invoices, run payroll, parse git commits, and generate daily briefings. Nothing off the shelf did this.
Solution
BotsUP delivered a full AI-native enterprise admin platform on Next.js + Firestore + GCP + MCP — strategy, execution, commercial, and HR — with AI agents as first-class citizens throughout:
Strategy layer: OKRs / objectives / key results + milestones wired to every project, quarterly through annual
Execution layer: projects / epics / backlog / sprints / kanban, from goal down to every git commit
HR layer: employees / AI agents / payroll / expense claims share one permission model — humans and agents unified
Four MCP servers (PM / Dev / CMS / Biz) let Claude.ai and Cursor operate company data directly
AI-generated morning briefings and evening retros, written back into the system automatically
Deep GitHub webhook integration: commits auto-link to backlog items; a scope guard tracks goal alignment
ISO 27001-aligned, multi-tenant permission isolation, complete audit log
Result
AI agents are first-class citizens from day one — not a chat bubble bolted on afterwards. The client team runs at multiples of comparable headcount: morning opens with an auto-generated briefing, afternoons hand invoicing to Claude, evenings let Cursor run review-to-ship — all on the same agent + MCP framework across the whole company workflow. In production for over six months and still iterating. Not a pitch deck — a system in daily use.
Our Own ProductApp Store Best App, 2021
Credidi (水滴發票) — 1.96M-User Consumer Finance App
Our own product, built and operated by the team for three years. AI image recognition on paper and electronic invoices, Taiwan e-invoice carrier integration, and named App Store Best App of 2021. Wound down in 2023 due to macro conditions (rising rates and slower-than-expected open-banking progress) — a business-environment call, not a technical or product failure.
AI image recognition: one-tap scanning of both paper and e-invoices
Taiwan e-invoice carrier integration and webhook reliability with back-fill
Re-architected Firestore / Cloud Run to hold cloud costs at million-user scale
A/B tested an AdMob + subscription dual business model
FlutterFirebaseAI OCRCloud RunAdMob
Our Own ProductIn Long-Term Operation
Taiwan Original Short-Drama App
Our own product, built from zero to one. Flutter dual-platform, BytePlus VOD streaming, AdMob monetization — we walked the full path from product definition and content ops through to monetization ourselves.
Flutter iOS / Android development and long-term ops
BytePlus VOD streaming and licensing architecture
AdMob integration and monetization flow design
Short-drama content ops and retention iteration
FlutterFirebaseBytePlus VODAdMob
Our Own Product5h MVP Record
Intermittent Walking — Idea to Store in 5 Hours
Our own product. Zero to simultaneous App Store / Google Play launch in five hours end to end. Proof that when your stack and product playbook are already wired, speed itself becomes a differentiator.
Strategy, UI, build, packaging, and submission, end to end
Existing Flutter + Firebase stack pushed to the limit
Post-launch retention tracking and monetization iteration
Our own product. Helps site owners measure search visibility, technical SEO, and AI-search (AEO) friendliness, with concrete, actionable recommendations.
The AI task / project / daily briefing system BotsUP uses internally. We plan to spin it out as a standalone SaaS in 2026. AI agents have been part of this system from day one.
Morning briefing and evening retro generated automatically
AI agents read and write tasks, projects, and decisions directly
Running internally in production for 6+ months
Next.jsMCPClaude APIFirestore
Sounds Like Your Situation?
Shipping a new product, taking over a legacy system, or wiring AI into an existing flow — spend 30 minutes with us first. We come back with a concrete answer. No fee. No pressure.