Shinhan Squarebridge Daegu × Everyone's Startup, One Day of AX Organization Design
On Tuesday, October 13th, an "AX Organization Design" training was held at Shinhan Squarebridge Daegu. It was an event where ESG Incubation 5th Generation Member Companies nurtured by Shinhan Squarebridge Daegu and applicants from the 2nd generation of "Everyone's Startup" came together.
Representatives from Unique Good Company, CEO In-hyuk Song and Development Team Lead Chan-woo Song, attended.
Morning 10am-12pm | General Education: Unique Good Company's AX Transformation Case, Changes in Work Methods Utilizing AI and Understanding (ESG Incubation Member Companies · Everyone's Startup 2nd Generation Applicants)
Afternoon 1pm-5pm | In-Depth Group Consulting: AX Consulting, AX Index Measurement, SEO/GEO Improvement Practice, Email Automation Connection, Delegation of Repetitive Task Automation (ESG Incubation Member Companies)
Shinhan Squarebridge is a nationwide startup innovation growth platform operated based on the startup nurturing and investment experience accumulated by Shinhan Financial Group over the years. The key achievements introduced that day were 820 member companies nurtured, 932.8 billion won in investment attracted, and 12,995 jobs created and maintained. There is more than one hub location in Daegu. From Seoul (early startup nurturing, open innovation), Incheon (deep tech, biohealth care, ESG, open innovation), Daejeon (deep tech, global expansion), Jeju (ESG, social incubation), to Vietnam (social open innovation), each region has specialized fields.
The Daegu hub is nurturing startups centered on eco-friendly, water industry, and social impact fields. It is a place where companies solving large social and environmental problems with small teams gather.
For ESG startups that need to create large impact with small teams, what does an organization that works with AI agents, an AO (Agentic Organization), look like?
Morning Lecture | Full Speed AX, Taking Flight Again with Nano Venture
The title of the morning general education was "Full Speed AX, Taking Flight Again with Nano Venture," and the subtitle was "A Story of Organizational AI Agent Transformation for 10x Growth."
CEO In-hyuk Song leads the physical AI experience content platform 'Real World' (4.2 million cumulative users) and is also a co-author of 'Nano Venture' written with CEO Eun-young Lee. He also operates the AI agent organization solution 'AXIT'.
There is a reason why 'taking flight again' is attached to the title of the lecture. Unique Good Company once grew to a size of 58 people, but as the number of people increased, the organization became slower and fell into billions of won in deficit. Now a team of 6 people works with AI agents and has created operating profit in just 6 months. The process of that failure and comeback is captured in 'Nano Venture'. So how are our organizations working with AI now?
During the lecture, participants logged in in real-time and solved the 'Nano Index Live' quiz directly. We selected one question and shared it.
If you're not satisfied with the AI's answer, what do you do?
Close the window and do it myself
Repeat "do it again" a few times
Point out what's not good and have it corrected
The corrected standard is saved, so from next time it comes out correctly from the beginning
Most of you are probably familiar with options 1 through 3. Option 4 is a bit different. The corrected standard doesn't disappear and carries over to the next task, so results improve as you work more.
The "AX Organization Design" we shared throughout that day was ultimately about making this option 4 not just an individual trick, but a company's default value.
Working with AXIT, 'Maki' that executes repetitive tasks instead
In the lecture, we also showed how Unique Good Company actually works on screen.
'Maki' in AXIT is a feature where if you set up repetitive tasks or conditional tasks, AXIT executes them instead. For example, every morning at a set time, it checks the blogs and communities of global accelerators, organizes newly released AI news, and posts it to the team channel. The execution process remains as a step-by-step workflow and can be reviewed at any time.
Adding more people doesn't add proportionally more strength
The lecture also addressed the inefficiencies created by organization size. In Ringelman's rope-pulling experiment in 1913, when 8 people pulled together, the strength per person was reduced to 49% of pulling alone. Eight people gathered, but the actual strength was 3.92 people's worth, meaning 51% was lost.
The more people there were, the more unknowingly each person was pulling less.
As the slide title suggests, this phenomenon is hard to avoid once you have more than 10 people. As people increase, meetings, reports, confirmations, and coordination also increase. This is why the formula that you need to hire more people to do bigger work is no longer a solution.
For ESG startups tackling big problems with small teams, this story is both a warning and an opportunity. Because you can design the structure where people and agents work together before the team grows.
From a company utilizing AI to a company moving with AI
"Starting the transition from a company utilizing AI to a company moving with AI."
Just having each member use AI well doesn't make the organization faster. The results and feedback corrected in each person's chat window remain as individual tricks and don't accumulate in the company. In terms of the previous quiz, it's a state of staying at options 1-3.
In a company moving with AI, members' instructions and feedback lead to agent execution, and the corrected standards remain to be reflected in the next execution. This is the structure that 'Nano Venture' calls the 'learning loop'.
Unique Good Company's PR and marketing also work this way. Instead of creating text and images with individual AI chatbots and posting them one by one, simply sharing the work channel results in blog, newsletter, and Instagram content being created and published. Among the thumbnails on the screen, you can also see reviews of this lecture posted on this blog.
From manager to executor, from Task to Goal
When agents take over execution, the role of people also changes. In the lecture, it compared being a manager to becoming an executor (IC, Individual Contributor) like this:
Verified others' work → I directly create results
Managed reporting lines → I see sales and costs
Achievement was the team's → Achievement goes under my name
Organization was evaluated from above → Evaluated by the value created
The capabilities required of new employees also change. Rather than fast document work, 'communication' that actually moves people, 'relationships' extending outside the company rather than tool handling skills, 'need for close' that stays and closes until completion rather than accuracy of assigned work, and 'responsibility' of putting their name on results rather than diligent progress reports.
The unit of delegation also grows. When AI was an auxiliary tool, work was broken down into small pieces and delegated like "research materials," "write documents," "create screens". As execution capacity grows, goals are delegated like "find ways for us to win in this market," followed by research, hypothesis, alternatives, experiments, comparison, and verification. This is a shift from Task Delegation to Goal Delegation.
Afternoon In-Depth Group Consulting | Seeing 'Our Company' Again with Agents
In the afternoon, we conducted in-depth group consulting with ESG Incubation member companies. It was a time to apply the stories shared in the morning to each company's actual challenges.
Are the strengths mentioned in the introduction also on the homepage?
The first order was the introduction of member companies. The screen showed each company's homepage address and 'known facts' confirmed from public information, and below that was one question.
Are the strengths mentioned in the introduction also on the homepage?
It's a question asking whether the company the CEO directly explains and the company customers encounter through search are telling the same story. In an era when AI summarizes search results to answer, a homepage becomes a company's first impression and the material AI refers to when introducing our company.
Next, we examined the corporate diagnosis results of 7 member companies together. The diagnostic cards were prepared in advance the day before the training. For each company, we included 'what customers can see now', 'what was actually verified', 'three things to fix first', and the materials that formed the basis of the judgment on one page.
In one member company's case, the product description appeared well on the browser screen, but the HTML body received when first loading the page was 0 characters. This means that AI or collection tools that read pages without executing JavaScript cannot read the product description at all. A homepage that looks fine to the human eye can be a blank page to AI.
Practice | Homepage Buildup, Two Requests Delegated to Agents
The diagnosis results immediately led to practice. There were two requests made to the agent.
Request 1. Discovery — How to be discovered better?
"Check if there are any areas that need to be strengthened so our company can be discovered better when AI searches for it. Let me know the priorities based on the shared materials and today's diagnosis results."
Request 2. What kind of homepage is right?
"Propose a homepage design suitable for expressing this content. Show me the page structure, first screen text, and sections that answer customer questions."
The inspection results of Request 1 become the design basis for Request 2. Inspection results like missing introductions, cases, numbers, Korean/English notation, and award/certification dates lead to page structure drafts, first screen text, and FAQ/case/basis sections. It's not about "create a homepage" work but about delegating the goal of "help us be discovered better," exactly the Goal Delegation we talked about in the morning.
Transmission. Don't wait, inform first
There is one more step equally important as preparing to be discovered better. That is 'transmission'. Rather than just polishing the homepage, you must continuously communicate the company's news and achievements through press release distribution channels like Newswire, media, and social media. The more records accumulated outside, the easier it is for both customers and AI to find and introduce our company more accurately.
The deliverables were also confirmed on the spot. For one member company, desktop and mobile mockups were proposed with the first screen text "Familiar wheelchair, new balance." It's a 6-step structure: problem, product/service, usage by customer type, stories with sources, FAQ, and inquiry preparation.
A noteworthy part is the 'To fill / Confirmation needed' section. Information not confirmed by public sources, like pricing, certification, or evidence of effectiveness, was not fictionally created but left as a space for people to fill in. The agent creates drafts quickly, while people take responsibility for facts and judgments. This is a division of labor.
The workflow of that day's practice was also shown at the bottom of the screen.
Discovery → Understanding → Customer Questions → Evidence to Disclose → Record in AXIT → Reflect in Channel → Confirm Results
A workflow designed so that diagnosis and correction don't end in one session but remain as records to serve as standards for the next execution. It's as if the 4th item of the morning quiz, 'the corrected standard is saved, so from next time it comes out correctly from the beginning', has been translated into practice.
AX Checklist You Can Do Right Away
Check if the corrected AI answer standard remains in the next task. If you're still re-explaining from the beginning each time, you're still at options 1-3.
Compare side by side whether the strengths the CEO introduces verbally are also written as-is on the homepage.
Check whether the company introduction visible on the screen is also in the page's original text, and in a form that AI and search engines can read.
Select one repetitive task and set it so that the agent executes it at a set time instead.
Try delegating goals to the agent instead of tasks. Like "make our company be discovered better when AI searches" instead of "create a homepage."
You don't need to change everything at once. The experience of one task's corrected standard leading to the next, that one small loop is the starting point for changing how the organization works.
Small Team, Big Impact
For ESG startups that need to create big impact with small teams, AX is not a distant future challenge. In that you can design the working structure before people increase, this might actually be the best starting point.
It was a day when I felt that 'Nano Venture's' slogan 'From Nano to Giant Impact' was the most fitting words for the member companies solving large social and environmental problems with small teams.
I'm grateful to Shinhan Squarebridge Daegu and the related parties of Everyone's Startup for providing this valuable venue, and to the CEOs and participants of ESG Incubation 5th Generation member companies who thought through this together from morning to afternoon.
Meet Stories of Organizational Innovation in 'Nano Venture'
How an organization that once grew to 58 people collapsed and rose again with AI agents. You can see the process of failure, comeback, and organizational redesign by the two authors In-hyuk Song and Eun-young Lee in the book introduction and table of contents.
'Nano Venture' — Written by In-hyuk Song and Eun-young Lee, Human Cube | View on YES24
Learn more about Nano Venture
AI Lecture and AX Organization Design Education Inquiry
We also conduct training for member companies of corporations and public institutions as well as accelerators and startup support organizations. Configurations that create customized tasks and deliverables for each company and team along with 'Nano Venture' lectures are also possible.
Email: create@realworld.to
Explore Corporate Education and Lecture Programs