Unleash truly agentic AI with executable Python workflows, custom RAG-powered agents, and a supercharged PostgreSQL backend.

From intent to execution

Today we’re launching JamAI Base v2, a major upgrade focused on real-world performance, stronger security, and developer control.
Highlights: Python Column to run custom Python inside workflows, JamAI Chat to compose chat agents, and a re-architected backend that’s faster and more reliable.

Below: what’s new, why it matters, and how to get started.

TL;DR

• Python Column: Execute your own Python in a secure, isolated runtime to process data, manipulate media, or run bespoke logic—no external servers required.
• JamAI Chat — Custom Chat Agents: Build RAG-powered agents with Knowledge Tables, lock them down with project-based RBAC, and ship them to end-users in a clean chat interface.
• Performance & Intelligence: Smarter Action Tables and a unified PostgreSQL backend (migrated from LanceDB + SQLite) deliver faster, more reliable automation.
• Templates: Prebuilt Action, Knowledge, and Chat Table sets to jumpstart common agent patterns.

Meet Python Column: custom code where it counts

Python Column lets you choose per step whether to use an LLM result, your own Python, or both—so you can fine-tune accuracy, cost, and control.

Highlights

• Flexible execution: Swap or combine LLM logic with Python on a column-by-column basis for deterministic transformations, validations, or post-processing.
• Multi-format I/O: Read and write text, images, or audio to power use cases from data cleanup to waveform analysis and image pipelines.
• Pre-installed libraries:
  • pandas, numpy for data manipulation
  • Pillow, opencv-python for image processing & computer vision
  • matplotlib for quick plots & visualizations
  • beautifulsoup4 for HTML parsing & web scraping workflows
• Architecture & Security: Python Column executes code inside isolated V8-powered containers and is further hardened by gVisor. Kubernetes orchestrates workloads for reliable, elastic scaling, while each run is confined to its own namespace with strict CPU and memory limits—preventing cross-tenant interference and ensuring consistent, secure performance.

Why it matters

Use Python where precision is non-negotiable—schema-aware data cleaning, deterministic scoring, media processing, or compliance checks—while still benefiting from LLMs for fuzzy tasks and natural-language steps.

JamAI Chat — Build, govern, and ship custom AI agents

Turn domain expertise into production-ready assistants—fast. JamAI Chat lets you design agents that are grounded in your data, governed by clear roles, and easy for teams to adopt.

• Grounded answers with Knowledge Tables (RAG): Anchor responses in your structured data for precise, context-aware outputs.
• Advanced RBAC: Project-based roles ensure the right people access the right agents—and nothing else.
• Frictionless end-user experience: Colleagues chat with your agents directly in JamAI, no complex setup required.

Great for: customer support copilots, operations assistants, research aides, and internal knowledge helpers.
Plays nicely with: Python Column for running custom Python right inside any step.

Performance & intelligence upgrades

• Smarter Action Tables: Enhanced multimodal prompts add richer step-level context, boosting answer quality and end-to-end reliability.
• Unified PostgreSQL backend: We’ve consolidated from LanceDB and SQLite to PostgreSQL for faster responses, stronger consistency, and better scalability.
• Open-source commitment: We’re continuing to invest in an open foundation to keep innovation moving.

Templates — start in minutes

Open the gallery and pick a ready-to-use template with pre-configured Action, Knowledge, and Chat Tables. Connect your data and ship an agent in minutes.

Perfect for:

• FAQ assistants & help-desk triage
• Report generation & data QA
• Media processing pipelines

Try JamAI Base v2 today

JamAI Base v2 bridges the gap between no-code speed and full-code power. With Python Column, custom AI agents, and a faster, unified backend, you can ship reliable automations and assistants—without leaving your workspace.

👉 Start building with JamAI Base v2 →

Have feedback or want to share what you’re building? We’d love to hear from you.