Week 1
💬
Simple Chain
An NGO grant strategist assistant with a defined persona and conversation memory. Three nodes, instant value.
From a simple chat chain to a self-correcting multi-agent pipeline — build real AI workflows with Flowise, Groq, Gemini, and SerpAPI. All free tools, all hands-on.
Each module builds on the last. Start with a simple chatbot and end with a multi-agent software house that reviews and fixes its own code.
An NGO grant strategist assistant with a defined persona and conversation memory. Three nodes, instant value.
Turn a PDF bank statement into clean, machine-readable JSON using structured output parsers.
A tool agent that searches the web, does math, knows the date, and posts to Slack — choosing its own tools.
RAG-powered policy expert that answers questions grounded only in your documents — no hallucination.
Four agents in an assembly line — PM, Developer, Tester, Writer — with a self-correcting loop. The capstone.
Each module is a 1.5-hour hands-on lab. Expand any week for full details.
Create four free accounts and copy down your API keys. No payment details required.
Build a live NGO grant strategist assistant. Add a chat model, conversation memory, and a system prompt persona. Test it with starter prompts and file upload.
Key concepts: Chat models, buffer window memory, system prompts, chat app configuration.
Turn a PDF bank statement into clean JSON. Wire a prompt template, LLM chain, and structured output parser. See the exact moment text becomes structured data.
Key concepts: LLM chains, prompt templates, structured output parsers, file upload, autofix.
Build an agent that decides which tool to use — web search, calculator, datetime, and Slack. Write custom JavaScript tools from scratch.
Key concepts: Tool agents, custom tools, SerpAPI, Slack webhooks, tool-calling models.
Build a policy expert that answers questions grounded only in your uploaded documents. Uses Gemini embeddings and a vector store for retrieval-augmented generation.
Key concepts: RAG, document loaders, embeddings, vector stores, grounding prompts, conversational retrieval QA chain.
Four specialist agents — Project Manager, Developer, Tester, Document Writer — collaborate in an assembly line. A condition node reads the Tester's verdict, and a loop sends failed code back to the Developer for up to 3 retries.
Key concepts: AgentFlow V2, multi-agent orchestration, condition nodes, loops, flow state, self-correction.
No payment details needed. Groq, Gemini, and SerpAPI all offer free tiers that cover the entire course.
NGO grant proposals, bank statement extraction, research agents, policy Q&A — not toy demos.
Start with a 3-node chain. End with a self-correcting 4-agent pipeline. Progressive by design.
Built by SOZO LAB — a social enterprise that ships AI solutions. Every lab solves a problem we've actually faced.
20 minutes of setup. Four free accounts. That's it.
Your no-code canvas
Fast, free AI inference
Embeddings for RAG
Web search (250 free/mo)
Program officers, project managers, researchers — no coding background needed. If you can drag and drop, you can build AI workflows.
Developers and students who want to understand RAG, tool agents, and multi-agent systems without wrestling with frameworks.
Organisations looking to automate grant writing, document processing, research, and policy Q&A with limited technical resources.
Corporate innovation labs and social enterprises prototyping AI solutions fast — before committing to full-stack development.
Each module is a 1.5-hour hands-on lab. The full course is about 7.5 hours of lab time, plus 20 minutes for prerequisites. Go at your own pace.
No. Flowise is a no-code visual builder. Week 3 and the Bonus module involve small JavaScript snippets for custom tools, but they're provided — you just copy and paste.
Yes — all the tools used in the course have free tiers. Flowise Cloud has a free plan, Groq offers free inference credits, Gemini has a free tier for embeddings, and SerpAPI gives 250 free searches per month. No payment details required for any of them. The course itself is a paid training — you're investing in the curriculum, guides, and practitioner expertise.
Each module is self-contained with its own guide. You can jump in at any week, though we recommend going through them in order — each builds on the previous.
Absolutely. The materials are designed for both self-study and facilitated workshops. Contact SOZO LAB for team pricing and custom facilitation.
Download the full syllabus and prerequisites guide. No email required.
Download Syllabus (PDF)Or visit SOZO LAB to learn more