Blend Bunjaku
Blend Bunjaku is an independent developer and Senior Frontend Engineer who publishes open-source developer tools. His principal current project, optimAIzr, is a local CLI that analyzes Claude Code, Codex, opencode, and API usage to identify wasted LLM spend, apply cost-saving changes, and measure their effects without uploading user data.
At a Glance
- Individual software developers using Claude Code
- Developers using Codex or opencode
- AI application developers using Anthropic or OpenAI APIs
- Privacy-sensitive users who want local usage and cost analysis
- +1 more
AI Tools by Blend Bunjaku
(1)
optimAIzr
AI Coding Agent Token Cost Analyzer
Discussions
No discussions yet
Be the first to start a discussion about Blend Bunjaku
Latest News
optimAIzr 0.12: Guard holds an expensive message to an expired cache once; live adds open-conversation cache countdowns and results reports trade-offs.
optimAIzr 0.11: results measures whether changes worked against prior usage; live reports per-turn cost; version 0.11.1 adds opencode support.
optimAIzr 0.10: cache countdown under Claude Code's prompt and sessions view for locating where re-reading takes over.
optimAIzr 0.9: automatic plan detection, confidence labels for savings, and one-command earlier context compaction.
Products & Services
Free, open-source MIT-licensed Node CLI distributed through npm. It profiles local Claude Code, Codex, and opencode sessions; imports CSV/JSON usage; prices token waste; recommends and applies fixes; and measures post-change results.
Optional Claude Code integration that shows answer cost and five-hour-window usage in-session, supports switching to a cheaper model from live, and provides an in-session handoff command.
A planned hosted product described on the website as separate from the open-source repository; no public pricing or launch date was provided.
Market Position
optimAIzr is positioned as a free, open-source, local-first cost-optimization and usage-analysis tool for coding agents rather than a hosted observability platform or an API gateway. Its differentiators are reading existing local agent histories, pricing waste in dollars, applying reversible fixes, and measuring whether a fix worked. Adjacent alternatives include Langfuse and Helicone for hosted/open-source LLM observability and cost tracking, and LiteLLM or Portkey for gateway-based routing, budgets, and provider management; those tools generally focus on routed application traffic, while optimAIzr focuses on local coding-agent histories and intervention.
Leadership
Founders
Blend Bunjaku
Independent developer and Senior Frontend Engineer. His GitHub profile lists him as a Senior Frontend Engineer; LinkedIn search results identify him as a Frontend Engineer at Twin Harbour Interactive and associate him with Universiteti i Prishtinës "Hasan Prishtina." He is the owner/maintainer of the blendbunjaku/optimaizr repository and creator of optimAIzr.
Executive Team
Blend Bunjaku
Creator, maintainer, and independent developer
Senior Frontend Engineer according to GitHub; LinkedIn search results identify him as a Frontend Engineer at Twin Harbour Interactive. He owns the GitHub repository and publishes the optimAIzr website, npm package, documentation, and releases.
Founding Story
Bunjaku built optimAIzr from his own experience with long Claude Code sessions, where repeated context reads and cache rewrites created substantial API-equivalent cost. The initial vision was a free, open-source, local-first tool that shows where LLM tokens and money go, lets users apply practical fixes, and verifies the results on their own usage data.
Business Model
Revenue Model
The current optimAIzr CLI is free and open source under MIT. The website describes a separate hosted optimAIzr Pro product as coming soon, but does not publish a price or revenue model.
Pricing Tiers
Open-source local CLI; installation is available through npm and the source is released under MIT.
Target Markets
- Individual software developers using Claude Code
- Developers using Codex or opencode
- AI application developers using Anthropic or OpenAI APIs
- Privacy-sensitive users who want local usage and cost analysis
- Open-source developer-tool users and small engineering teams
- Reducing token and API-equivalent spend in long Claude Code coding sessions
- Monitoring and optimizing Codex and opencode agent usage
- Finding expensive cold-cache returns and repeated context loading
- Applying earlier compaction, model selection, retry, input-size, and cache-related fixes
- Instrumenting a developer's own Anthropic or OpenAI application with a wrapper
- Analyzing exported AI-usage CSV or JSON data locally and privately