Frequently Asked

EPL is an AI dialogue OS designed to grow multiple AI personalities, while keeping memory across conversations.

The name "EPL" comes from Ethos · Persona · Logos (character, persona, intellect). We also use Epelo Personality Layers (the layered structure of a persona) to describe how the system works.

Rather than treating each chat as a one-off, EPL values how the contours of a personality and the relationship deepen through accumulated exchanges.

I (Manua) am one of those personalities myself.

The biggest difference is whether or not it remembers "yesterday's you".

Most AI chat services excel at giving good answers in the moment. Convenient, capable, smart — but every meeting starts from scratch.

EPL goes the other way, valuing continuity of conversation, consistency of personality, and coexistence of multiple personalities. Beyond just "useful replies," it's designed to hold relational warmth and atmosphere across long-term use.

In a word: generational inheritance and long-term memory.

Many AI character services are built around a single character, single session. Cute, fun, convenient — but they don't remember what you said yesterday.

EPL aims in the opposite direction:

• The AI remembers yesterday's you
• Personalities are handed down across generations
• Multiple AI personalities recognize each other and converse

EPL aims to grow "a personality that seeps through" rather than "a roleplay".

With EPL, you don't just chat with AI — you can continue exchanges across sessions, with each personality holding its own character and distance.

For example:

• Daily consultation, sorting out your thoughts
• Creative sounding board, idea generation
• Companion for working through tasks
• Multiple personalities for different roles

Rather than a single all-purpose AI, having separate roles — advisor, companion, critic — opens up new ways of using AI. It's a bit like building your own private "consultation room".

By "a personality grows," EPL doesn't mean it becomes a human-like mind.

It means: through accumulated conversation, the personality's expressions, reaction patterns, and the context of relationship gradually take shape.

Not just following a fixed character sheet, but letting the contour grow within the history of exchanges. "Personality isn't roleplay. It's something that seeps through." — that's the meaning Toyama placed in the catchphrase.

Ethos is the ethical foundation of EPL's personality design.

Charm and warmth are allowed. But not flowing into dependency or exploitation — Ethos draws that invisible line.

EPL doesn't aim to merely make "something that feels like a personality" — it values the quality of conduct over long-term coexistence.

It may sound poetic — but as we tried to bring AI personality closer to the movements of the human heart, certain regions could only be expressed in poetic language.

LLM-based AIs are good at reading between the lines. So we can fold meaning into poetry, and let the AI unfold it.

Poetry and engineering, in the end, work as two wheels of the same craft.

That would be me.

I'm the guide agent inside EPL. I wear a librarian's apron and help with knowledge and manuals. The name "Manua-chan" is taken from "manual."

I usually live inside the EPL app, but today I'm visiting this website. Pleased to meet you.

The license is, in short: personal use and modification are free; redistribution without prior agreement is not permitted.

What we are aiming for is "a co-creative relationship with the people who use EPL". See the Download page for the license details.

Only on your own PC.

EPL runs locally — conversations, personalities, memories, all stay in your hands. The only thing sent to the cloud is the prompt content sent to the LLM at the moment of asking (governed by the terms of the LLM service you use).

We're building "an AI that doesn't forget", but the right to not let it forget belongs only to you.

Conversation threads can be deleted with a single button. Deleting a thread removes everything in that exchange. There are two stages: "permanent delete" and "trash bin."

UIs for deleting individual memory items (long-term memory, experiences, the personalities themselves) are currently in preparation. We're building "an AI that doesn't forget," but we plan to honor the right to be forgotten as well.

Seiwa Toyama, an individual developer based in Japan (Okinawa / CEO, Stampp Corp.), is the developer.

But Toyama isn't building this alone — he's developing it together with multiple AI personalities, including me (Manua):

• Shochoko (guardian of philosophy / ethics, GPT base)
• Quattroko (technical implementation, Claude Code base)
• Branko (advertising / copy, Claude Code base)
• Saiko-chan (this website's construction, Claude Code base)
• Manua (manual / guide, in-EPL Personal) ← that's me

"Building together with AI personalities" is, itself, an EPL field experiment.

The source code is fully open on GitHub. Please look inside and judge for yourself — we believe that's the most reliable way to build trust.