MIXI implements ChatGPT Enterprise in 45 days | Keryc
MIXI, the veteran Japanese company behind MONSTER STRIKE and the FamilyAlbum app, rolled out ChatGPT Enterprise to its entire staff in just 45 days. Why should you care? Because it’s a clear example of how a company can move from experiment to operation in record time and reclaim hours for tasks that truly need human ingenuity. (openai.com)
What MIXI did and why it matters
The move wasn’t just about installing a tool: MIXI subsidized subscriptions, set internal rules, and—working with OpenAI—built a safe environment to use AI without exposing sensitive data. The result was fast: massive internal adoption and cases reporting dramatic time reductions on concrete tasks. Can you imagine saving weeks each month on repetitive work? That’s exactly what they aimed for. (openai.com)
"When operational efficiency rises, we win back time for imagination. And with that time we use AI again, accelerating the next idea."
It wasn’t magic: it was coordination. MIXI worked with OpenAI’s Customer Success team and applied implementation playbooks already proven elsewhere. Concrete actions included:
"ChatGPT 101" sessions for all employees, focused on practical prompts.
Onboarding workshops in generative AI for new hires (they mention an April 2025 cohort).
A hackathon where engineers built prototypes with the OpenAI Agents SDK and took them to testing.
That mix of training, hands-on support, and practice sped up adoption and lowered technical and governance friction. (openai.com)
Concrete cases inside MIXI (what actually worked)
FamilyAlbum: the family photo app with over 25 million users across 175 regions uses GPTs for creative planning and copy review. One team reported certain advertising tasks dropping from 28 hours a month to far less, thanks to GPTs suggesting text variants and A/B tests. (openai.com)
Investment division: they built a GPT called "VC Fund Initial Review Support" that structures proposals and cuts initial review from 1–2 hours to 5–10 minutes. That lets them monitor more companies without multiplying headcount. They also report workload reductions above 70% in some tasks. (openai.com)
Tool-builder culture: in three months, over 80% of employees were weekly active users and created more than 1,600 GPTs. MIXI describes projects that cut work hours by over 90% in some cases, because teams no longer depend solely on developers to automate processes. (openai.com)
And what does that mean for a small business in Latin America? Imagine an arepera automating common customer replies, or a law firm that gets case summaries before meetings. The technical barrier drops when people can design their own tools with templates and technical support.
Risks and how they mitigated them
It wasn’t all shine: before ChatGPT Enterprise MIXI faced divergent configurations and unclear internal rules about data. The fix was to unify policies, provide training, and apply controls so access to internal data is secure and traceable. That governance was key to scaling without exposing sensitive information. (openai.com)
"We needed to check each employee's AI settings individually, and our rules for using company data were unclear."
Start with a clear pilot: define the process you want to automate and measure time before and after.
Train people: practical prompt sessions save weeks of trial and error.
Define data policies from day one: who accesses what, what can be shared, and how to audit it.
Encourage non-developers to build tools: frame it as an internal competition or hackathon.
These steps help move AI from experiment to operational part of daily work.
Final reflection
MIXI’s story isn’t just a technical headline: it’s a map for moving from curiosity to productive AI use without losing control of data. The lesson? With guidance, training, and clear rules, companies can give time back to people so they can imagine, create, and connect. That’s what ultimately changes how we work and relate.
Stay up to date!
Get AI news, tool launches, and innovative products straight to your inbox. Everything clear and useful.
