Vol. 05 / Blog
The Ledger.
Writing on company knowledge, AI agents that do real work, and the craft of building shared context your whole team (human and machine) can trust.

FeaturedWhy your AI agents keep forgetting what your company knows
Every chat starts from zero, and a smarter model won't fix that. A governed place to keep what your company knows, read by the agents you already use, will.
Latest from the team

Agent memory vs context engineering: what is the difference?
Agent memory is what a system keeps between calls. Context engineering decides what goes into the next one. Why teams confuse them, and what it costs.

Company brain glossary: the vocabulary of the context layer
Plain definitions of the terms behind company brains and context layers: governed write, provenance, MCP, RAG, grounding and thirteen more.
Write it down once
Tatara is one place your team keeps what your company knows, and every AI you use works from it.
no card required

Governed context: why an agent must not pass as a person
Agents now write into company knowledge as well as read it. What governed context means, what provenance must record, and why an agent write needs a label.

What is the Open Knowledge Format? OKF explained
OKF is Google Cloud's open spec for storing what an organisation knows as Markdown files with YAML frontmatter. What it defines, and what it can't carry.

How to share the same context across Claude, ChatGPT and Gemini
Claude, ChatGPT and Gemini each keep their own memory. Four ways to give all three the same company context, and what each one costs you.

Tatara vs Glean: a Glean alternative for small teams?
Glean searches the systems a large company already runs. Tatara is where a small team writes down what it knows for agents to read. That decides the choice.

Tatara vs Notion for AI agents: which should hold your context?
Notion is a general workspace. Tatara is a governed document store agents read over MCP. Which shape fits your job, and when a second surface is not worth it.

Tatara vs Slite: telling two context layers apart
Slite and Tatara describe themselves in nearly the same words. Where they really differ, what to test in both, and how to tell which one fits your team.

What is a company brain? A definition, and how to build one
A company brain is one maintained store of what your business knows, readable by people and retrievable by AI agents. What it is, and how to build one.

Built to be found
An agent that can't find the right document ignores the whole brain. How structure, tags, and a machine-readable map turn storage into retrieval.

The governed document: one source, two audiences
A governed document is clean enough for a person to read and structured enough for any model to use. How we shape one document to serve both.

Stop pasting context into every chat
The hidden tax of stateless agents, and what changes when every tool reads from the same current brain.
THE INVITATION
Come build yours.
Start free with 500 documents and no card. See the workspace your agents have been missing.
