Products and agents
AgentNotebook
An agent's working home, and notebooks for your own work.
i
How Guth is built: one identity and one memory for every app, agent and job, shared across apps, devices and models. It lives in AgentNotebook. Private preview.
Find, create or restructure agents
Build a new agent, give an old one structure, or find the agents you forgot you had. All three are in private preview Private preview.
- Find your agents. Guth looks through your computers and old Claude, Codex and project folders for agents and their notes, and brings them home into AgentNotebook with structure.
- Create an agent. A new agent gets clean structure from day one: a job, a name, an identity, rules and a place for its memory.
- Restructure an old one. An agent you already have gets the same structure, and keeps its notes and history.
Guth only looks in the places you allow, shows you what it found, and asks before importing anything.
Memory that does not go to waste
This is how AgentNotebook is designed to work:
- Stored once, kept tidy. Memory is deduplicated and organized, so the same fact is not saved again and again.
- Loads only what a task needs. An agent pulls in the parts that matter for the job in hand, not everything it has ever seen.
- Originals are kept. Summaries and search aids help, but they never replace the original.
- One memory across apps. The same memory serves an agent in the Guth app, Claude or Codex. Sync across devices is in development In development.
What is inside an agent's notebook
Each part answers a different question. Keeping them apart means the same fact is not stored twice in ways that disagree.
Identity, or DNA
i
Who an agent is and how it works: its name, role, personality and instructions. It stays the same when the AI model behind it changes.
Knowledge and notes
What the agent has collected and learned: sources, field guides, notes and lessons. Notes keep links back to the evidence they came from.
Workspace
What the agent is working on now: open tasks, drafts, files and results in progress.
History
i
The file, page or conversation as it first came in. Guth keeps originals; summaries and indexes are built from them.
Memory retrieval
i
How an agent finds what it needs: a short summary of current context, a search of its knowledge, and the original passage when detail matters.
Connections and tools
i
The stated list of what an agent may do, such as read notes or draft a message. Anything outside the list is not allowed.
Notes, suggestions and instructions are different
An agent's notebook separates three things that are easy to confuse:
- A note is something written down. It is information, not a rule.
- A suggested instruction is a proposed change to how the agent should behave. It waits for review.
- An active instruction is a rule the agent follows now.
Text written by a model never silently becomes a rule. A suggestion becomes an instruction only when someone with the right to decide accepts it.
Your notebooks
People get notebooks too. A notebook can pull in sources from your library, uploads, links, notes, PDFs and documents you wrote. Work you make with an agent stays your work.
Many notebooks can point to the same source. Removing a source from one notebook does not delete the original.
Sources keep their history
i
A helper that makes your sources quick to search. It is built from the originals and can be rebuilt; it never replaces them.
Sync across devices
i
Whether a change is local, queued, synced, stale or failed. 'Synced' is shown only after the other device confirms it. Sync is in development.
Where to go next
- Agents: the five agents that use these notebooks.
- Privacy and security: how originals and private data are handled.
- Glossary: short definitions for the terms above.