About MyKanik

Systems-first applied AI

Overview

MyKanik is an intelligence platform created by Brenden Portolese.

While the initial application focuses on automotive repair, MyKanik is fundamentally a procedural knowledge platform — designed to retrieve, canonicalize, and present complex, multi-step instructions from multiple sources in a way that is structured, explainable, and trustworthy. (Patent Pending)

This is not a generic AI chatbot. It is an intentionally designed system built to address known failure modes in large language models when applied to step-based, safety-sensitive domains.

About MyKanik

Most AI assistants struggle with procedural tasks:

  • They summarize or paraphrase critical steps
  • They lose context across follow-up questions
  • They merge conflicting sources without attribution
  • They hallucinate missing details
  • They treat safety warnings as optional text

MyKanik was built to solve those problems directly.

The system assembles responses from authoritative manuals, structured data sources, and community knowledge, determines procedural equivalence across sources, and produces a canonical, structured answer that can be safely referenced and refined through follow-up interaction.

What makes MyKanik different

Canonical procedures, not free-form answers

MyKanik constructs a structured Answer Plan for each procedure, rather than generating ad-hoc text. This plan persists across the session and anchors all follow-up questions.

Multi-source by design

OEM documentation, professional guides, and community insights are preserved with explicit provenance. Conflicts are surfaced, not hidden.

Anchored follow-up interaction

When a user asks a follow-up question ("What about step 2?"), the system resolves it against the existing procedure instead of re-querying the entire knowledge base.

Safety-aware architecture

Safety warnings, constraints, and risk-sensitive steps are treated as first-class system components, not optional annotations.

Technical focus

MyKanik is intentionally built as an intelligence platform featuring automotive repair:

  • Hybrid retrieval (vector search, structured sources, web ingestion)
  • Canonicalization and equivalence resolution across procedures
  • Structured data modeling and serialization
  • Provenance tracking and conflict handling
  • Session-level state and anchored follow-up logic
  • Safety enforcement at the procedural level

Technology stack

Backend: Python
AI orchestration: LangChain
Embeddings: OpenAI (Ada)
Vector database: Qdrant
Web layer: Java
Frontend: Flask (for rapid iteration)

About the creator

MyKanik was designed and built by Brenden Portolese, a product leader and systems architect with over 15 years of experience building and modernizing large-scale platforms across fintech, data, and infrastructure-heavy domains.

This Product reflects a personal belief that successful AI products require intentional system design, clear constraints, and respect for the underlying problem domain — especially when real-world outcomes matter.

Why this matters

MyKanik is both a working product and a living portfolio of how to design AI systems that go beyond prompt engineering.

The same architecture applies to other domains such as:

  • Medical procedures
  • Industrial maintenance
  • Laboratory protocols
  • Cooking and manufacturing workflows

The core problem — turning fragmented procedural knowledge into something reliable and interactive — is universal.