MEMBER OF TECHNICAL STAFF @ BAYES CONSULTANT

JAMES KANYIRI

About Me

I am a Member of Technical Staff at Bayes Consulting, where I work on building AI solutions.

I also work on side projects, including PayeLink, a developer framework for agent-to-agent discovery, identity, payment, and evaluation, and Workpods, an AI workspace where teams can collaborate with any AI model, research deeply, and ship polished work.

I did my degree in Computer Science at the Multimedia University of Kenya. Previously I worked as an AI Engineer at Thumbprint Consulting and a Backend Engineer at Payecards. I also run LangChain Community Kenya, a community for AI developers here in Kenya.

Articles

Lessons on filesystems for context engineering and memory
What we learned building a production grant writing agent: context management, workspace-as-memory architecture, and why memory and tooling matter as much as writing quality for knowledge-work agents.
Article · February 15, 2026
LangGraph Assistants: Building Configurable AI Agents
How to use LangGraph Assistants to separate graph logic from agent behavior: configurable prompts, models, and tools with one graph, plus a step-by-step guide with LangGraph Studio.
Article · July 9, 2025
Exposing a LangGraph Agent as an MCP Tool
Learn how to expose a LangGraph agent as an MCP tool so any MCP client can use it—enabling composable, reusable Agent-as-a-Service workflows.
Article · July 7, 2025
Extending Daraja MCP Beyond Payments
A step-by-step tutorial from the Unstructured MCP hackathon. I built DarajaMCP in a week: it extracts invoice data with Unstructured's API, stores it in MongoDB, and automates M-Pesa payments via the Daraja API. This post walks you through the full pipeline so you can build it yourself.
Article · April 1, 2025[GitHub]

JAMES KANYIRI

About Me

Articles

Lessons on filesystems for context engineering and memory

LangGraph Assistants: Building Configurable AI Agents

Exposing a LangGraph Agent as an MCP Tool

Extending Daraja MCP Beyond Payments