Arsenal
Sharp solutions rarely require invention, they just need revelation. We stay within our zone of expertise and reapply proven ontological patterns; we ship faster, with less risk, and with fewer surprises. Every tool in the Arsenal exists because we've solved the same class of problem more than once, refined the approach, and made it reusable. The result is a system that works on day one, not a prototype that needs six more months.
Proven patterns, composed for speed. Every tool battle-tested in production.
Speed and certainty don't come from cutting corners. They come from not having to reinvent the wheel every time. The AKOS Arsenal is our deployable set of tools, compute modules, and methodologies that we install, compose, and orchestrate across engagements. It's how we ship production-ready systems in weeks, not quarters. Every module listed below has been built close to operators, validated in production, and designed to snap into Palantir Foundry or run independently.
Tool
Type
Input / Output
Where it fits
Industries
AI Chatbot SDK App
External React/Next.js app layer for Palantir Agent Studio agents, built with Ontology SDK and Platform SDK.
User Query + App Context→Grounded Answer + Agent Actions
YOLO Modeling Objective
Complete setup for YOLO object detection model inference directly inside Foundry.
Foundry Media Sets→Bounding boxes + confidence
YOLO Mutation Manager
Manages context-specific model variations without creating entirely new tools.
Base Model→Context-Specific Weights
Paddle Detection + Cropping Pipeline
Paddle-driven detection pipeline to find regions of interest and output clean crops for downstream tools.
Image / PDF→Bounding boxes + cropped regions
Paddle Table Structure Extractor
Extracts rows, columns, and cell structure so messy tables become parseable.
Table Image / PDF→Table Structure + Coordinates
Handwritten Row Extractor
Specialized OCR model for extracting and structuring data from handwritten tables in legacy documents.
Image / PDF→Text + confidence
Signature Verification
Compares two signatures to validate authenticity with a similarity score.
Signature A + B→Similarity score
Human Validation Station (HITL)
Lightweight human-in-the-loop review and correction interface with audit trail.
AI output→Human-verified data
Data Matching Engine
Row/column matching across datasets with exception handling.
Dataset A + B→Matches + confidence
Write-Back Orchestrator
Secure, controlled write-back to systems of record.
Validated data→ERP update + audit
LLM Context Filter
Uses language models to filter low-value retrieval results and improve grounded generation quality.
Vector Results→Relevant Context
Agentic Quality Conductor
Evaluates AI output and routes low-quality results back for regeneration.
AI Draft→Improvement Feedback
Ontology Graph Explorer
Interactive graph view of linked objects, relationships, and operational context.
Ontology Objects + Links→Graph View
AI Chatbot SDK App
External React/Next.js app layer for Palantir Agent Studio agents, built with Ontology SDK and Platform SDK.
Input / Output
User Query + App Context→Grounded Answer + Agent Actions
YOLO Modeling Objective
Complete setup for YOLO object detection model inference directly inside Foundry.
Input / Output
Foundry Media Sets→Bounding boxes + confidence
YOLO Mutation Manager
Manages context-specific model variations without creating entirely new tools.
Input / Output
Base Model→Context-Specific Weights
Paddle Detection + Cropping Pipeline
Paddle-driven detection pipeline to find regions of interest and output clean crops for downstream tools.
Input / Output
Image / PDF→Bounding boxes + cropped regions
Paddle Table Structure Extractor
Extracts rows, columns, and cell structure so messy tables become parseable.
Input / Output
Table Image / PDF→Table Structure + Coordinates
Handwritten Row Extractor
Specialized OCR model for extracting and structuring data from handwritten tables in legacy documents.
Input / Output
Image / PDF→Text + confidence
Signature Verification
Compares two signatures to validate authenticity with a similarity score.
Input / Output
Signature A + B→Similarity score
Human Validation Station (HITL)
Lightweight human-in-the-loop review and correction interface with audit trail.
Input / Output
AI output→Human-verified data
Data Matching Engine
Row/column matching across datasets with exception handling.
Input / Output
Dataset A + B→Matches + confidence
Write-Back Orchestrator
Secure, controlled write-back to systems of record.
Input / Output
Validated data→ERP update + audit
LLM Context Filter
Uses language models to filter low-value retrieval results and improve grounded generation quality.
Input / Output
Vector Results→Relevant Context
Agentic Quality Conductor
Evaluates AI output and routes low-quality results back for regeneration.
Input / Output
AI Draft→Improvement Feedback
Ontology Graph Explorer
Interactive graph view of linked objects, relationships, and operational context.
Input / Output
Ontology Objects + Links→Graph View
13 of 13 tools