./work
The systems we are building.
MershLab is a research lab with working software behind the thesis. These projects are connected parts of the same attempt: make AI agents more capable, more contextual, and easier to trust.
public / building
Saturday
A zero-dependency agent harness for any model and every surface, with approvals, checkpoints, tool use, and provable trails.
./verification layeropen source / building
MershTrust
A verifiable execution and audit layer for AI systems, from model inference to multi-step agent runs.
./context layeropen source / building
MershBrain
A company knowledge system that turns scattered context into searchable, executable skills for agents.
./applied systembuilding
AutoMarket
An AI lead-acquisition system for Indian SMBs that carries a business conversation from inbound interest to an owner-ready brief.
./applied systemprototype
Dispatch
An after-hours voice intake system for service businesses where every call becomes an auditable run.