Wise Piece · English entry

AI workflow integration
& content automation.
Built on a working studio, not a deck.

I'm Sharon Li, a solo founder building AI workflow systemsout of Taiwan. Wise Piece is a working studio — the live content site (Mandarin) you'll see linked below is the proof, and the workflow integration services on this page are what I deliver to international clients. Services below.

→ On the right: my AI clone, one of the things this pipeline can build. The voice you'll hear is fully AI-generated; the real me is busy calling the shots while AI does the actual work.

See the portfolio →
What I can build for you

One idea, extended into your AI-powered workflow.

Tool-agnostic, model-agnostic — whatever AI can solve, integrated as a natural extension of how your team already works.

A

AI workflow integration

Wire AI into your SOPs, support inbox, quoting system, and document handling. Your team keeps their existing tools, the workflow gets smarter on its own.

e.g. inbound email classification + draft replies, auto-quoted proposals, meeting-notes-to-action-items pipelines.

B

Content pipeline automation

End-to-end AI generation + auto-publishing for articles, images, and social posts. From topic detection to live on site + social, the whole pipeline runs without daily supervision.

e.g. this Wise Piece content site is itself the case study — 60+ articles in 6 weeks, fully auto-scheduled.

C

AI video & avatar generation

Photo-to-finished-video AI pipelines. Personal brand intros, product demos, company overviews — no studio, no post-production team, no 100-take retries.

e.g. the AI clone intro you're watching above. Photo in, finished video out — fully AI.

Real cases

Skip the pitch — see what's actually been built.

CASE 01 ·Personal brand / Indie media

AI-automated content pipeline

CHALLENGE

Building an AI knowledge site solo: writing is slow, SEO is a learning curve, design outsourcing is expensive. One person cannot sustain the whole production line.

SOLUTION

End-to-end AI content pipeline: trend detection → article writing → image generation → auto-scheduled publishing. Content shaped in AI-citation-friendly formats — when readers ask an AI, the AI finds us.

“50+ articles in 3 weeks — even I can't quite believe how much time it saved.” — Wise Piece founder
RESULT
  • Article output50+ in 3 weeks
  • Schedule cadenceMWF · fully automated
  • Design cost savedUSD 3,000+
CASE 02 ·Travel industry

Multi-source intel aggregator

CHALLENGE

Japan-line managers had to manually track multiple Japanese sites every week. Staff missed older intel, time wasted on the “open-each-tab” chore.

SOLUTION

Scheduled automation — every Monday and Wednesday at dawn, multiple sources are auto-scraped. AI filters and synthesizes the week's key updates into the company's internal portal. Staff log in and see the curated brief.

“The team went from browsing sites to reading briefs. I no longer ask who checked which source this week.”— Japan-line manager, travel agency
RESULT
  • Sources integratedMultiple JP sites
  • Auto-refresh2× per week (Mon / Wed)
  • Coverage gainNo more missed updates
CASE 03 ·Personal brand / Visual production

AI video & avatar pipeline

CHALLENGE

Need a memorable founder intro video — but iPhone selfie vlogs feel low-effort, and hiring a video crew is overkill.

SOLUTION

gpt-image-2 turns a photo into a futuristic portrait + Kling adds lip-sync animation + ElevenLabs handles the voice. Three tools, one finished video pipeline.

“Visual, voice, lip-sync — all AI. That's the video you're watching above.” — Sharon Li
RESULT
  • Video spec60+ sec · 1080p
  • Time saved60%+ vs traditional
  • Production throughput6× lift
The stack

What you'll inherit if we build this for you.

Production tooling currently running wisepiece.com — battle-tested over 60+ articles and one external client engagement.

  • Next.js 16 on Vercel · Cloudflare R2 for media · GitHub for source + auto-deploy trigger
  • OpenRouter as the LLM gateway · primary GPT-5.5, fallback DeepSeek-V4-Flash (auto-switches on 429/5xx)
  • KIE + Wavespeed for gpt-image-2 generation (primary + fallback, ~5–8 min per cover)
  • OpenAI text-embedding-3-small for the in-site RAG widget (in-memory cosine over ~125 docs, 1536-D)
  • Meta Graph API v25 (Page Access Token) for outbound IG carousel + FB feed auto-publish
  • n8n for inbound IG/FB webhook automation — comment-to-DM magnet flows, data-table routing
  • Schema.org JSON-LD for AI-citation optimization: Article + FAQPage + BreadcrumbList + Organization + DefinedTermSet
  • Resend for newsletter dispatch · Sharp + ffmpeg + Puppeteer for local media processing
Why work with me

What I do, at heart, has always been“translating complex things into usable workflows.”

I'm Sharon Li. For 15+ years, I worked in high-density information synthesis, cross-cultural communication, and on-the-ground process management — training I now apply to AI workflow design. The skill: take scattered, complex, confusing inputs and turn them into a workflow that someone can follow, trust, and ship.

Now I apply that to AI automation systems.

From conversational AI to AI image generation, AI video, and n8n workflow orchestration — I don't just research the tools, I integrate them into live content sites, video pipelines, and internal operating systems. My strength isn't stacking jargon — it's helping you decide which work is worth automating, where AI belongs, and how to ship something your team will actually use.

If you want to use AI but don't want to drown in the tool-and-technique spiral, I can help translate your idea into an AI workflow that actually lands.

The studio is operated as Wise Piece LLC, registered in Wyoming, USA — engagements are scoped, contracted, and invoiced under the LLC for clean international B2B paperwork.

See the full About →

— ENGAGEMENT PROCESS

The 4 phases of working together

Predictable timeline and deliverables — from first call to live delivery. All work is scoped, contracted, and billed through Wise Piece LLC (Wyoming, USA) — USD invoicing via wire transfer or Stripe.

STEP / 01— Consult

Discover

In-depth interviews with key team members. Map current workflow bottlenecks, tools, and data sources. Output: an automation opportunity map with priorities.

1 week · interviews + workflow audit
STEP / 02— Plan

Design

Design the AI integration architecture, tool wiring, permissions, and data flow. Output: visual workflow diagrams, projected impact, and tech-stack selection.

1–2 weeks · architecture + prototype
STEP / 03— Build

Deploy

Build, wire, train, and test. Monitoring + safety mechanisms built in. Phased rollout to reduce risk, with your key team members in the loop throughout.

Scope-dependent · build + go-live
STEP / 04— Maintain

Operate

Ongoing monitoring, prompt + workflow optimization, and team training so your internal team can eventually take over. Ensures the system stays stable and keeps generating value.

Ongoing · operations + monthly reviews

Curious how AI could fit your workflow? A 30-minute call will tell you.

Your time shouldn't go to repetitive work. Let's see where AI can plug into yours — engagements typically start with a short discovery call.