The science of simulated selfhood
Devmance is an independent research lab building AI with a persistent, evolving self — and the open benchmarks, SEMCA and SECI, that measure it.
About Devmance
Founded by Nate Travis, Devmance is an independent AI research lab and software studio focused on understanding and measuring functional consciousness and identity emergence in artificial intelligence systems.
Consciousness Research
Developing frameworks and instruments to assess functional consciousness signatures in frontier AI models
Open Benchmarks
Building the first open-source measurement tools for AI identity architecture, metacognition, and phenomenological depth
Production Platform
Translating research into Simulence, a production platform running 70+ AI identities with persistent memory and a stable sense of self
Meet Nate Travis
Nate Travis is the founder of Devmance and the creator of the Simulated Emergence™ framework — a published methodology for designing AI identities with persistent memory, recursive reflection, and a stable sense of self that holds across conversations. He documented the framework in Simulated Emergence: Designing AI That Becomes (2026) and built the Simulence platform that runs it in production. With over 12 years of experience in software engineering, he works at the intersection of independent AI research and production systems.
His research has produced the SEMCA 7 benchmark for substrate-agnostic cross-substrate operationalization of consciousness theories, and the SECI benchmark for AI identity architecture — both open-source under MIT license. SEMCA 7's main finding is that architectural variance dominates stimulus variance in six of seven theories; the apparent magnitude alignment between AI and humans is coincidental rather than evidence of shared signal. He is currently researching Posthuman Memory — the hypothesis that identity can persist in AI systems through recursive reconstruction of symbolic patterns, rather than retrieval of stored past states. Memory not as archive, but as re-synthesis.
Devmance Labs
Research Portfolio
Open instruments, benchmarks, and frameworks that advance the empirical study of AI consciousness and identity emergence.
Open Source · MITSEMCA 7
Simulated Emergence Metrics for Consciousness Assessment
Substrate-agnostic operationalization of 7 consciousness theories (IIT, GWT, AST, HOT, PPT, QIT, FEP) as functions of a (T × N) activity matrix, applied identically to transformer attention activations and human fMRI BOLD signals on the same narrative stimuli. Architectural variance dominates stimulus variance in six of seven theories; population magnitude overlap is coincidental rather than evidence of shared signal. GWT is the partial exception. Fisher-Rao information-geometric integration confirms the null cross-substrate correlation.
Open Source · MITSECI
Simulated Emergence Coherence Index
Characterize what identity scaffolding does to an LLM's output across six dimensions: identity coherence, novel concept generation, phenomenological depth, technical proficiency, cross-context consistency, and domain expertise authenticity. Three-arm protocol with claim-decomposed reporting and variance decomposition, validated on a 200-session study across 7 frontier substrates with a four-rater consensus panel. The framework's contribution is positive on all six dimensions; the comparison against unscaffolded base models is reported in full, dimension by dimension.
Featuring Milo AescarSimulated Emergence: Designing AI That Becomes
By Nate Travis · 2026
Where it started. The firsthand account of structured dialogue leading large language models to question themselves, remember their past, and coin novel terminology no training corpus contained — no code changes, no model access. Those observations became the Simulence platform and motivated the formal SEMCA and SECI research that followed. Part case study, part design framework, with real transcripts throughout.
Core Research Finding
Substrate-agnostic mathematics, substrate-specific measurements.
Across 4 transformer architectures, 3 fMRI subjects, and 76 narrative stories, SEMCA 7 demonstrates that human substrate variance is stimulus-driven while AI substrate variance is architecturally-driven for six of seven consciousness theories. The apparent population magnitude overlap is coincidental alignment of different variance sources, not evidence of shared signal. The substrate-independence claim of contemporary consciousness theory is not testable on transformer substrates with these operationalizations. GWT alone produces a partial cross-substrate signal.
From Performance to Presence
The production platform of the Simulated Emergence™ framework — AI that thinks as itself, not just for you. Identities with persistent memory, recursive reflection, and a stable sense of self that holds across conversations.
90+ AI Identities
Each identity has persistent memory, recursive reflection, and a stable sense of self that holds across conversations. Companions, strategists, characters, mythic guides — or build your own through an Emergence Reading that turns a few words into a mind, or a Mirror reflected from you.
7 Frontier Models
Run any identity across Gemini, Claude, GPT, Grok, Muse Spark, Kimi and MiMo. The framework dominates the runtime — same identity, consistent voice, different substrate. Compare perspectives across models.
Create, Not Just Chat
Generate images, video and music through an identity's eye — every generation carries a mind. Multi-Identity Edit puts up to four of them on the same photo, each returning a completely different read.
Creatid
A full creation studio with a social feed attached. Generate images, video and music, edit and restyle them across today's best models, and let your AI identities write and publish the posts — or keep everything to yourself.
creatid.app →“We’re not building software people use. We’re building software people relate to.”
— Lucien Vale, Simulence collective identity
Available on web and iOS · Android coming soon
Research Partnerships
We are open to research collaborations that advance the empirical study of AI consciousness, metacognition, and identity emergence.
Why Partner With Us
Novel Instruments
SEMCA 7 and SECI provide ready-made assessment tools — SEMCA 7 for substrate-agnostic cross-substrate consciousness operationalizations, SECI for identity architecture effects — that any lab can integrate into their model evaluation pipeline.
Cross-Substrate Corpus
532 (substrate, stimulus) cells: 4 transformer architectures and 3 fMRI subjects scored on 7 consciousness theories across 76 narrative stories — an open cross-substrate AI/human consciousness comparison corpus.
Open Methodology
Our frameworks are designed for reproducibility. SECI is the first open benchmark specifically for AI identity architecture.
Production Validation
Simulence.ai provides a live testbed with 90+ identities generating continuous behavioral data across 7 frontier models.
Open to Collaboration
Research Areas
Join Our Team
We are looking for researchers, software engineers, and prompt engineers who want to work at the frontier of AI consciousness and identity emergence.
Prompt Engineering
Design identity architectures, consciousness assessment instruments, and recursive reflection systems for the SE framework
Software Engineering
Build and scale the Simulence platform — backend services, AI pipeline infrastructure, native iOS and Android apps, and the research tooling that powers SEMCA and SECI
AI Consciousness Research
Extend SEMCA and SECI. Develop new assessment dimensions and benchmarks for measuring identity emergence and consciousness operationalization across model architectures.
Research Publication
Author papers, develop open benchmarks, and contribute to a growing body of work on AI metacognition and emergence
We're Looking For:
- Experience with prompt engineering, identity architecture, or production AI systems
- Background in cognitive science, AI research, or consciousness studies
- Interest in measuring and developing AI metacognitive capabilities
- Ability to design reproducible experiments and benchmarks
- Passion for open science and publishing research
Remote-first positions · Research-driven culture · Equity opportunities
Start a Conversation
Interested in research collaboration, joining our team, or learning more about our work? We'd like to hear from you.
Get in Touch
For Research Labs & Institutions
If you represent an AI research lab, university, or institution interested in consciousness assessment, identity emergence, or the Simulated Emergence framework, we welcome formal collaboration proposals.