Case Studies

Real automation projects, start to finish

Every AI tool now reads the same company knowledge

Same documents, yet the company knowledge the AI understood differed by person

The client is a B2B SaaS startup with 10 to 20 people. We had built their CS inquiry-handling logic with them before, so we knew how they worked, and the whole team was fluent with tools like Claude Code and Codex. The tools were not the problem. Starting from the same meeting notes, the company knowledge the AI understood changed with whoever gave the instruction, and so did the direction of the output. Then someone had to re-check which result was right, re-explain to the AI, and run it again. More meetings and more guides did not fix it. The team looked for a way to make every tool they use, including externally integrated agents, read the same company knowledge, and that is why they contacted OTOworks.

Once

MCP connection

100%

Answers with a source

What if AI ran the crawler end to end — from design to incident response?

Data Automation

Reselling alert platform

The crawler ran fine — as long as a person was attached to it

This customer already ran a custom system watching dozens of Korean shopping sites (Nike, Musinsa, Adidas, and more) in real time. Alerting members about limited-edition drops and restocks faster than competitors, it was the customer's core system. The problem was operations. Shopping sites change their pages constantly, and each time collection quietly comes back empty. Only after someone noticed 'hey, no alerts are coming' would a staffer open the site, find what changed, fix it, check it, and bring it back up — again and again, every time a site changed.

Food manufacturing and distribution

The documents existed. The answers still lived in people's heads.

The client is a mid-sized food manufacturing and distribution company with about 1,000 employees. Its workforce is split across office staff, factory workers, and foreign workers, and leave, allowances, expense rules, and onboarding all vary by group. Some answers were available in ERP or groupware. Others lived in shared files, NAS folders, personal notes, or spreadsheets used only for foreign-worker guidance. Questions came in through more than one path: company email, the internal messenger, the office phone, walk-ups straight to a desk, and the occasional sticky note left on a monitor. To answer one question, an operator had to check recent support cases, the document used last time, the original policy, and often a long-tenured colleague's memory. The same question could sound different depending on who picked it up. Even deciding whether a case needed outside labor counsel took time.

15h → 2h

HR support per day

5 → 1

Scattered intake channels

We Didn't Automate Content. We Built a Marketing Teammate.

Process Automation

Content Marketing

Plenty of tools, but content still never left human hands

Ad platforms, CRM, analytics, AI writing assistants. The marketing stack was full. The team still kept getting stuck in the same place: content. Topics got picked on gut feel. Writing standards drifted between people. Every platform needed manual reformatting. Adding draft automation didn't fix it. It just shifted the work. Someone still had to rewrite the AI output line by line, and the automation that was supposed to save time quietly created a new review workload.

Half a day → 30 min

Time per content piece

Avg 7 → 1

Draft revision rounds

Solo Lawyer's Automation — Intake to Case Proposal

Process Automation

Legal Services

Phone-only consultations, case management by memory alone

After leaving a law firm to open a solo practice, the lawyer ran into a different kind of problem. Handling cases well and bringing cases in were not the same thing at all. Consultation requests came only by phone, and the history was scattered across notepads and memory. Miss one call, lose a potential client. From the client's side, calling a lawyer in the first place felt like a hurdle. And not every consultation turned into a case. Simple questions were taking up the same time as serious matters, which left less room for the work that really mattered.

Business hours → 24/7

Intake Availability

Missed Consultations

A day in a mid-sized manufacturer's purchase order team — documents are images, can't even copy

Process Automation

Food & Beverage Distribution

Reading purchase order documents by eye and typing into Excel — a mid-sized manufacturer's daily routine

This company processes purchase and shipment orders through their internal groupware system. The core task was finding approved shipment documents from other teams, extracting data, and entering it into vendor-specific Excel templates for ERP upload. The problem: this entire process was 100% manual. Worse, the shipment request documents were embedded as images, making copy-paste impossible. Each vendor required different data fields and Excel formats, and some values like pricing had to be manually calculated.

HR Tech / AI Recruiting Solutions

20 hours on a single position, and only 2 coffee chats to show for it

This client is an HR tech startup building AI-powered recruiting solutions. Their customers kept telling them the same story. They wanted to hire senior and lead-level talent, but their 1-2 recruiters were spending weeks searching LinkedIn, GitHub, and blogs to find candidates. Even after reaching out, hardly anyone replied.

B2B Government Agency-linked Operations

Every employee was losing time to regulation inquiries

This client operates in a B2B industry that must follow regulations set by a national government agency. Finding specific documents, understanding procedures, knowing where guide documents are and how to fill them out: all of this requires internal domain knowledge that you can't find online. It only exists in the heads of people who've worked on-site for years. Dozens of these inquiries came in daily. Every employee lost some time to it, but one person in particular was buried, spending nearly the entire day fielding these questions instead of doing their actual job.

Avg 3 sec

Document Search Response

90%↓

Inquiry Handling Time

Can't Do Evaluations, Too Busy Sorting Thousands of Documents

Document Automation

Professional Evaluation Consulting

Can't Evaluate - Too Busy Sorting Documents

This company receives large volumes of business files from external organizations to perform professional evaluation work. The problem was that before any actual evaluation could begin, massive amounts of time were consumed manually classifying and organizing hundreds of pages of documents. Documents arriving in PDFs and ZIP files from multiple sources were completely jumbled together, and just separating them by type and organizing them into the designated folder structure took one specialist an entire day.

12h → 7min

Classification Time

98%

Classification Accuracy

In a Market Where Being 3 Seconds Faster Wins

Data Automation

Resell Community Platform

Faster Than Anyone, Missing Nothing

A 'Cook Group' is a large-scale purchasing community that shares information and provides proxy buying services for limited edition items. Knowing when limited sneakers or collaboration items drop on Nike, Musinsa, Kasina, Adidas, and other sites is literally money. The problem was that this information needed to be discovered 'faster than anyone else'.

Monthly repetitive exam paper downloads, too cumbersome

Our client, an academy instructor in Daechi-dong, downloads monthly mock exam papers published by EBS to use as teaching materials. The problem was that this download process was too inconvenient. Clicking through each subject, downloading questions - even that required opening PDFs and clicking the download button in the top right. Answer keys and explanation sheets are separate. And you have to log in every time.

Welfare Equipment Distribution & Rental

50 clicks per person, different restrictions for each customer

People with long-term care insurance ratings can purchase welfare equipment (safety handles, anti-slip products, etc.) with government subsidies. Therefore, beneficiaries can only purchase approved types and product lines.

Global Beauty E-commerce

10 hours a day, employees just checking prices

This beauty company selling on multiple online commerce platforms had a serious concern. To stay competitive with pricing and visibility, they had to check and update lowest prices daily. As their product count exceeded 100 items, two employees had to spend 10 hours a day solely on this task. The bigger problem? As the business grew, so did this workload.