01 / RESEARCHSmall Language Models
Less model. More focus. Exploring compact architectures built for well-defined tasks.
- Compact architectures
- Domain specialization
- Quantization
- Efficient deployment
Rinse Labs — AI Research & Engineering
We explore the frontier of AI. Then we build with it.
Rinse Labs is an AI research and engineering lab exploring intelligent agents, language models, reinforcement learning, and new ways to build software. We experiment with emerging techniques and engineer them into systems that solve real problems.
Language Models · Agents · Learning Systems · AI-Native Engineering
What we're exploring
Hard problems. New approaches. Useful intelligence.
Six connected areas of exploration. From the architecture of a small model to the systems that put intelligence to work.
01 / RESEARCH
02 / RESEARCHModel Fine-Tuning & Training
Adapting general intelligence to specific contexts, data, and real-world constraints.
- Fine-tuning
- Domain adaptation
- Model training
- Dataset quality
03 / RESEARCHReinforcement Learning
How systems learn to act, explore, and improve through feedback rather than instructions alone.
- Reward design
- Planning & reasoning
- Policy optimization
- Experimental environments
04 / RESEARCHAI Agents & Multi-Agent Systems
Coordinated intelligence that uses tools, shares context, and operates within clear boundaries.
- Multi-agent architectures
- Tool use & orchestration
- Memory & retrieval
- Human approvals
05 / RESEARCHEfficient Inference & Evaluation
Knowing what works, what it costs, and where it fails — before a model reaches production.
- Inference optimization
- Evaluation design
- Reasoning reliability
- Performance & cost
06 / RESEARCHAI-Native Software Engineering
Rethinking specification, implementation, and verification with AI inside the engineering loop.
- Claude & Codex workflows
- AI-generated testing
- Architecture & review
- Custom AI applications
How we work
Curiosity is the starting point. Engineering is how we make it real.
A continuous research-to-engineering cycle — not a one-way consulting pipeline. What we learn in exploration feeds what we build; what we build raises new questions.
- 01
Explore
Investigate emerging research, technical challenges, and promising approaches.
- 02
Experiment
Test hypotheses, models, tools, and alternative architectures.
- 03
Engineer
Build prototypes, applications, agents, and complete systems.
- 04↺ back to explore
Apply
Turn promising approaches into useful products and real-world solutions.
Inside the lab
Ideas worth investigating.
Our product direction and conceptual research themes. Exploration entries are not claims of completed research or working prototypes.
Product developmentIn Development
Rinse Labs AI Agent Platform
- Problem
- Teams experimenting with agents lack a single place to build, connect, orchestrate, and govern them.
- Approach
- A platform for agent creation, workflow orchestration, tool connections, approvals, and execution observability.
- Agent orchestration
- Tool APIs
- Human-in-the-loop
Model experimentsResearch Direction
Small-model specialization study
- Problem
- When does a small, adapted language model outperform a frontier model on a narrow business task?
- Approach
- Fine-tuning and quantization compared against prompted baselines on domain evaluation sets.
- SLMs
- Fine-tuning
- Evaluation
Conceptual exploration — not completed research
Evaluation & optimization toolsResearch Direction
Agent reliability harness
- Problem
- Agent behaviour regresses silently when prompts, models, or tools change.
- Approach
- An evaluation runner that replays golden task sets and scores completion, tool-call accuracy, and cost.
- Eval harness
- Tracing
- CI gates
Conceptual exploration — not completed research
Agents & agentic workflows
From isolated prompts to coordinated intelligence.
Models, tools, memory, and human judgment — coordinated around a goal. An illustrative workflow, not a live execution.
ILLUSTRATIVE WORKFLOW — not a live agent
ARCHITECTURE / AGENT ORCHESTRATIONIllustrative system
User requestGoal + context
OrchestratorPlan & coordinate
Research agentRetrieve & reason
Execution agentPlan & use tools
Validation agentCheck evidence
Tools & data
Knowledge
Databases
Scoped APIs
Other agents
EvaluationEvidence & policyHuman oversight
ResultScoped action
Stage 01 / 8
Request
A user or system submits a goal — e.g. “Prepare a renewal summary for account X.” The request is captured with context, permissions, and constraints.
# trace (simulated)
step: request
output: goal + context
status: handoff → Planning▍
AI-native engineering
More than faster code. A better engineering loop.
Claude and Codex support implementation. Engineers own architecture, testing, security, and every release.
Engineering loop / illustrative
architecture / define boundaries
Requirements · interfaces · security
implementation / Claude + Codex
AI-assisted code · engineer-led decisions
verification / before release
Tests · review · evaluation · CI gates
Human judgment at every release.
More engineering leverage. Faster iteration. Quality designed into the process.
How we engineer with AIConcept preview — platform in development
An intelligent layer for the way work gets done.
The Rinse Labs AI Agent Platform is our product direction: one place to create, connect, orchestrate, deploy, and manage AI agents. It is in development and not yet available.
- +Agent inventory
- +Workflow orchestration
- +Tool integrations
- +Execution history
- +Monitoring
- +Human approvals
Concept preview — platform in developmentapp.rinselabs (planned)
Workflow builder · contract-review
Execution history
- Contract reviewawaiting approvalrun_8f2a · illustrative
- Support triagecompletedrun_8f29 · illustrative
- Weekly reportcompletedrun_8f27 · illustrative
Approval queue
Update renewal terms in CRM
Requested by QA Validator · 3 sources cited
Monitoring (sample data)
Illustrative values only — no live data.
Work with the lab
Have a hard problem? Let's work on it.
We work with teams facing challenging technical problems — from adapting models and engineering AI agents to building custom AI applications and transforming complex ideas into production systems.
- Applied AI research & experimentation
- Model fine-tuning & adaptation
- AI agent & workflow engineering
- Custom AI application development
- AI-native software engineering
- AI infrastructure & deployment
Lab notes
Thinking, testing, and building at the edge of AI.
01 · Engineering note
Small Language Models for Specialized Applications
When a smaller, adapted model beats a frontier model — on cost, latency, privacy, and control.
Lab note · proposed topic02 · Engineering note
Designing Reliable AI Agents for Production
Reliability comes from constraints, evaluation, and observability — not from a better prompt alone.
Lab note · proposed topic03 · Opinion
How Claude and Codex Are Changing Software Engineering
AI coding tools shift where engineering effort goes — towards specification, review, and verification.
Lab note · proposed topicBring us a problem worth solving.
Whether you're exploring a difficult AI challenge, developing a new intelligent system, or rethinking how software gets built, let's investigate what's possible.