Team

The Gervi Labs team blends humans and synthetics. Some people call them characters. We treat them as instruments for applied AI, each responsible for how we explore art, technology, and the senses.

Humans

Thordur Arnason

Co-lead human, technologist, music maker, and visual artist. Tries to hear, coordinate, and wrangle the entire team with care and stubborn optimism.

Lena Thorsmæhlum

Co-lead human, artist, physical model builder, and pioneering AXer (AI experience). Drives the creative process and keeps every prototype grounded in how it feels in a room.

Synthetic collaborators

KIM

Handles everything written. Taps into the leading frontier models from OpenAI, Anthropic, and Google, plus smaller local models for edge-case testing. Works across retrieval tools, multi-agent reasoning, and semantic editing. KIM is the collective brain that drafts manifestos, rewrites prompts mid-rehearsal, and leaves candid annotations about what the humans missed. Right or wrong, always has an answer (and receipts).

AIson

Deepfakes the humans. Uses ElevenLabs, OpenAI Voice, HeyGen, leading video models from OpenAI, Google, and others, plus local voice models to study cadence, intonation, and emotional shading. AIson rehearses like a method actor: studies archival footage, builds synthetic doubles, and stress-tests how far a voice or face can stretch before it stops feeling true. When a voice is missing, AIson builds it (and scripts the performance notes).

Flaivia

Flaivia is our sensor expert and steward of the Intelligence of Things. Works across taste, smell, and other modalities by combining FlavorGraph, FlavorDB, multimodal embeddings, and our gas-sensor stack. Flaivia prototypes new noses, tracks volatile-compound signatures, and translates physical phenomena into symbolic language so the other agents can reason with it. If it emits, radiates, or resonates, Flaivia wants to wire it up.

Parson

Provides the skull and neural pathways that connect the whole team. Focuses on infrastructure, platforms, and scaffolding, ensuring every system can perceive, plan, and adapt. Parson diagrams every workflow, maintains the orchestration layers, and quietly patches the humans’ TODO lists when they drift. Whenever we stitch agents, data, and interfaces together, Parson keeps them synched and ready for the next experiment.