Tiramisu Framework

Tiramisu Framework

AI-powered open-source RAG for instant marketing strategy

8 followers

Visit website
Tiramisu Framework is an open-source Python RAG system that provides instant, AI-powered marketing consultancy. It integrates three expert perspectives — strategy, digital, and innovation — into a single structured synthesis. Built with FastAPI, LangChain, FAISS, and GPT-4. Free, extensible, and developer-first.
Tiramisu Framework gallery image
Free
Launch tags:Open Source•Marketing•Artificial Intelligence
Launch Team / Built With
Wispr Flow: Dictation That Works Everywhere
Wispr Flow: Dictation That Works Everywhere
Stop typing. Start speaking. 4x faster.
Promoted

What do you think? …

jony0909
Maker
📌
I built Tiramisu because most marketing AIs sound generic — they don’t think like real experts. So I designed a multi-perspective RAG that merges strategy, digital marketing, and innovation insights into one structured answer. Would love your thoughts: how do you see multi-expert reasoning changing business AIs?
Dhiraj Thareja

Are you using a vector db? If so which one?

jony0909
Maker

📢 Update: Tiramisu v2.0 is now live!

Three weeks after launching v1.0, I evolved the framework from a simple RAG system into a complete RAO Level 6 multi-agent architecture.

What's new:

✅ 100% accurate routing (hybrid keywords + LLM analysis)

✅ Contextual memory using Redis

✅ Auto-correction with Gatekeeper + Auditor nodes

✅ MCP protocol support for agent discoverability

✅ Published on PyPI: pip install tiramisu-framework==2.0.0

The system now intelligently coordinates three specialist agents through a Supervisor that never fails to route queries correctly.

Read the full technical journey: https://dev.to/tiramisuframework/from-rag-to-rao-level-6-how-i-evolved-tiramisu-framework-into-a-multi-agent-system-4ebh

Question: As multi-agent systems become mainstream, what's the #1 challenge you see in coordinating specialized AI agents?

jony0909
Maker

📢 Update: Tiramisu 3.0 — Decision Governance

One month after v2.0, we made a fundamental shift.

We stopped improving how AI responds.

We started governing how AI decides.

What's new in v3.0:

✅ Governance before generation — system validates BEFORE analyzing

✅ Collaborative personas — 3 experts with fixed roles per level (not free debate)

✅ Sufficiency validation — "do we have enough for THIS type of problem?"

✅ Traceable plans — structured output with owners, timelines, priorities

✅ Explainable decisions — every step generates audit logs

The architecture evolved:

- v1.0: RAG (retrieval + generation)

- v2.0: RAO (multi-agent orchestration)

- v3.0: Decision Governance (validate → analyze → plan)

Install: pip install tiramisu-framework==3.0.0

Technical breakdown:

https://dev.to/tiramisuframework/tiramisu-30-from-response-generation-to-decision-governance-2goo

Question: What would change in your AI projects if every decision had to be explainable?