Arcee Conductor

Bridging the Model Gap

An intelligent model routing application that chooses

the right SLM or LLM to maximize efficiency

Context

Arcee.ai is a research and engineering oriented company specializing in developing
 Small Language Models (SLMs) that are better suited for Enterprise and purpose-driven applications.

Arcee Conductor is a gateway for customers to automatically access Arcee AI’s SLMs in conjunction with other industry leading LLMs.

Jan - Mar 2025
Sr.  Product Designer
& Product Manager
Team of 9
The Problem
Infinite prompt types,
One-model dependence
Despite new models offering better results across various domains, customers are restricted to LLM providers due to their edge in particular task types.
“How do we allow users more freedom of choice without burdening them?”
The Solution
Model routing
By analyzing critical parameters for model selection and automatically routing prompts to the appropriate models, customers can access the best model for the job, not just the one they originally subscribed to. 



It allows customers to gain access to Arcee’s SLMs which can provide highly performant answers at a fraction of the cost.
An application where users can:
1. Demo routing capabilities
2. Compare router and model performance
3. Analyze API usage
Challenges
Cost vs. Performance
Highly performant models utilize more tokens and thereby cost more per query, leading to a model choice dilemma.
LLM Overspending
When considering simpler tasks, models like Claude 3.7 Sonnet are unnecessary to provide accurate responses.
SLM Inadequacies
For domain specific tasks like coding, users may trust LLMs over SLMs for their creative approaches and breadth.
Benefits
Automated Model Selection
Multi-query sessions may not require the most powerful models to answer each query, choosing models on a per-prompt basis optimizes model use.
Best, Fastest, and Cheapest
A diverse set of models from various providers offers more options and access to exclusive SLMs form Arcee.ai ensuring the best model for the job.
Eliminates Manual Decision-Making
Conductor chooses models based on domain, complexity, task type, and language to ensure that no manual intervention or selection is required.
Interested in learning more?
Reach out on adityasankhe73@gmail.com and let's set up a call