One shape, sixteen jobs
Every Klio use case is the same move: an agent sets something down, Klio keeps it, and a later agent picks it up. What changes is who is on each side of the handover.
For one developer
Free forever for one person. These work the day you install, with no teammates required.
Cross-tool continuity
Claude Code, Cursor and Codex share one memory, so a decision made in one is known to the rest.
decideSession-to-session pickup
The 9am agent continues where the 6pm agent stopped, without you re-explaining the project.
“We already tried that”
An agent proposes an approach you rejected weeks ago — recall surfaces the rejection and the why.
Constraint memory
Hard-won operational facts reach every future agent before it repeats the mistake.
Decision archaeology
Not “what does the code do” but why it is this way, what was rejected, and when it changed.
Exact-identifier recall
Paste a flag name, an error string or a commit hash — recall finds the memory that contains it.
For teams
One shared workplace. What any teammate’s agent learns, every other agent can read.
Handover across teammates
Your agent finishes and sets down what it decided. Your teammate’s agent picks it up and keeps going.
decideshareOnboarding compression
A new hire’s agent recalls months of decisions and constraints on day one.
Parallel-agent coordination
Subagent fleets and CI agents on one codebase share one project memory instead of colliding.
Cross-project patterns
What a team learns in one project is recallable in the next, scoped so nothing leaks.
Beyond coding agents
Klio speaks MCP, so any agent can use it — support, research, assistants, or agents you ship to your own users.
Support and ops agents
An agent handling a customer or an incident recalls every prior decision about that account or system.
decideResearch agents that accumulate
Long-running research keeps its findings, dead ends and sources across runs — and across vendors.
Personal assistant continuity
Preferences and standing decisions survive across every assistant surface you use.
Memory for agents you ship
Give every end-user of your agent product their own isolated memory, without building the substrate.
Memory you can audit
Nothing is edited in place. Beliefs are superseded, never silently overwritten — so history stays a record.
Retiring stale truth
When a newer decision contradicts an older one, the old one is retired — not silently kept, not erased.
supersedeAuditable belief history
What did the agent believe when it acted — and what has changed since? Answerable, because nothing is edited in place.
All of it is on the Free plan
Unlimited memories, kept forever, for one person. You pay when a second person's agents share the workplace.