Question 1

What is Tamacofflow?

Accepted Answer

Tamacofflow is a personal AI assistant with long-term memory across sessions. Unlike ChatGPT or Claude which reset every session, Tamacofflow remembers your projects, decisions, and preferences. Built with multi-provider routing (Groq Llama 4 Scout, DeepSeek, Gemini, OpenRouter) for cost efficiency and reliability.

Question 2

How does Tamacofflow remember conversations?

Accepted Answer

Tamacofflow uses a Library Architecture: structured retrieval instead of dump-everything-context. Conversations are extracted into atomic memory entries (facts, decisions, projects) stored in a Postgres database. When you chat, only relevant memories are retrieved and injected into the LLM prompt, saving tokens and improving accuracy.

Question 3

Is Tamacofflow free?

Accepted Answer

Yes, Tamacofflow offers a free tier in early access (invite-only, limited to 50 active users). Free tier includes 100K token cap, basic chat, document and image upload, web search, and memory recall. Paid tiers planned post-public-launch with extended capabilities.

Question 4

What AI models power Tamacofflow?

Accepted Answer

Tamacofflow routes requests across multiple LLM providers for cost and reliability: Groq Llama 4 Scout (primary, free tier), DeepSeek Chat, Google Gemini 2.5 Flash, OpenRouter (fallback), with optional Anthropic Claude Sonnet/Opus and OpenAI GPT-4o for paid tiers.

Question 5

How is Tamacofflow different from ChatGPT or Claude?

Accepted Answer

Tamacofflow is an application layer on top of existing AI providers, not a foundation model. Key differences: persistent memory across sessions (ChatGPT/Claude reset every chat), multi-provider transparency (no silent model swap), token usage visible in real-time, audit trail for memories used, and structured retrieval for cost efficiency.

Question 6

Can Tamacofflow read PDF documents and images?

Accepted Answer

Yes. PDF text documents are extracted and analyzed. PDF image-only files (scans, screencaptures) are auto-rendered to images and processed via vision LLM. Common image formats (PNG, JPG, WebP, GIF) supported via Llama 4 Scout multimodal vision.