Warning: Cannot modify header information - headers already sent by (output started at /home/xs301118/sparx.blog/public_html/wp-content/themes/blogus-child/single.php:26) in /home/xs301118/sparx.blog/public_html/wp-content/themes/blogus-child/functions.php on line 66

We will build a DX foundation for small and medium-sized businesses with a VPS-resident AI agent for ¥390,000.

We installed a meeting minutes tool. We also installed a task management tool. We’ve equipped ourselves with chat tools and reminders.
Even so, after a meeting, “summarizing tasks, contacting someone, and tracking progress” remains a human job.
Even with more tools, I’m the one bridging the gaps.
You can have an AI agent take over that bridging work.

“Adding Tools” vs. “Deploying an Agent” are Fundamentally Different
General SaaS tools streamline specific tasks. A meeting minutes tool transcribes faster. A task management tool makes managing tasks easier.
However, the bridging tasks such as “reading transcription and extracting tasks,” “registering tasks in the management tool,” and “notifying the person in charge” are done by humans across multiple tools.
What makes an AI agent different is its ability to automate this bridging.

The moment transcription data enters, it is analyzed -> tasks are extracted -> registered in the management tool -> notified to the person in charge.

The essence of an agent is “end-to-end processing” that operates across multiple tools, not as a standalone tool.

What’s Actually Working
On April 26, 2026, we began test operations of the meeting minutes AI pipeline we built.
We placed a 25-minute meeting transcription into Google Drive and registered it in Notion’s meeting minutes DB. Five minutes later, a VPS-resident AI agent began processing, and a notification arrived on LINE WORKS.

“Meeting minutes analyzed. 8 tasks registered in Notion. Individuals in charge have been notified.”

The AI wrote out the tasks. The AI assigned the people in charge. The AI registered them in Notion and notified them via LINE WORKS.
This system went into full operation in May 2026 for our clients in the real estate industry.

Technical Core: The Entry Point Doesn’t Matter, the Exit is One
The structure is simple.
Transcription (Entry point is flexible: Google Meet / Notta / Manual input, etc.)
↓
Notion Meeting Minutes DB (Status: Unprocessed)
↓ Monitored every 5 minutes
AI Agent (VPS Resident)
↓
1. Analyze meeting minutes
2. Extract tasks (assignee, deadline, priority)
3. Register in Notion Task DB
4. Notify assignee via LINE WORKS

A simple configuration with Claude Code resident on a VPS, receiving and processing LINE WORKS webhooks. Operates with Gmail / Google Calendar / Drive / Docs / Sheets / Notion via MCP (Model Context Protocol).
Processing Performance (Actual Measurement):

Response Time (p95): 12 seconds
Concurrent 10 Parallel Processes: Completed in 31 seconds
Meeting Minutes Analysis -> Task Extraction: 10-15 seconds (8 tasks for a 25-minute meeting)

VPS Costs Start from ¥1,150/month
The term “AI agent foundation” might give an impression of high costs. The actual cost is as follows.

Since you will prepare your own VPS, both initial and monthly costs are minimized.
Even with a full configuration (operating Gmail / Calendar / Drive / Docs / Sheets / Notion entirely), the Xserver VPS 6GB plan (¥2,200/month) can run it comfortably.

Meeting Minutes AI is Just the “Entry Point”
We’ve focused on meeting minutes AI so far, but this is just one example of a “easy-to-understand use case.”
With the same foundation, simply changing the role of the worker allows for the operation of business-specific AI assistants.
Examples of Implementable Business Customizations:

Sales: Lead List Management (Sheets) -> Automated Follow-up Email Generation (Gmail) -> Pre-Visit Briefing Creation (Docs)
Hiring: Applicant Response (Gmail) -> Interview Scheduling (Calendar) -> Job Posting Generation (Docs)
Administrative Tasks: Deadline Management (Calendar) -> Application Draft Creation (Docs) -> Notification to Stakeholders (LINE WORKS)
Real Estate Operations: Generate Introduction Articles from Property Information -> Manage Showing Schedules -> Customer Follow-up

“A foundation that operates across multiple tools, with instructions given to the agent directly from LINE WORKS” — Meeting minutes AI is the most easily understandable entry point for this.

Two Options
We offer two ways to acquire this foundation.

For those who want to build it themselves:
Claude Code × VPS × LINE WORKS Complete Implementation Guide (¥15,000)
A 65,000-character complete implementation guide that reveals all six pitfalls encountered during implementation, with “symptoms -> causes -> solutions,” starting from VPS selection, covering Claude Code MAX OAuth authentication, LINE WORKS Bot construction, SQLite queue design, Google MCP integration, Notion integration, and production operation.
It consolidates information that cannot be found through searches or in official documentation into a single volume.
→ Build Your Own AI Agent Foundation for ¥15,000 — Note Paid Article

For small and medium-sized businesses who want us to handle the construction:
We handle everything from environment setup to business customization.

Construction Agency Lite: Initial ¥80,000 + Monthly ¥10,000 (VPS environment setup + LINE WORKS + Google MCP connection)
Construction Agency Standard: Initial ¥200,000 + Monthly ¥15,000 (Includes 1 type of business customization)
Construction Agency Pro: Initial ¥390,000 + Monthly ¥25,000 (Full 3 types of business customization + Meeting Minutes AI Pipeline)

We will set up an environment where you can give direct instructions to the AI agent from LINE WORKS, which your company uses daily. There’s no need to learn new tools.
→ Entrust Your SME DX Foundation to Professionals — VPS AI Agent Implementation Package Details

In Conclusion
The era has arrived where humans no longer need to handle the bridging work after meetings.
With just the cost of VPS and Claude MAX at $110/month (approx. ¥17,500), you can place a 24-hour AI assistant at the center of your operations.
Your company’s operations must also have tasks that are bridged by humans.
As you read this article, did one come to mind? There might be room for an agent there. Please contact us for a consultation first.
→ SPARX AI LAB Service Details

SPARX CREATIVE STUDIO
SAYONG PARK
X: @sparxcs
Mail: info@sparx.blog