Documentation
Start with the quick start in the chatbot you already use. You connect SMEme, build a decision tree, run it on a case, and ask more about the result. The guides below cover Deploy, the editor, and connector setup.
You're reading version static of the SMEme documentation. Read the changelog.
Why SMEme
Business rules don’t belong in prompts. A prompt can describe your procedure, but the model still decides each time how to apply it. Ask twice and it may weigh things differently, skip a step, or fill a gap with a plausible guess. That’s fine for a first draft. It isn’t fine for a decision you have to make the same way every time and explain afterward.
SMEme is rational AI. It reasons from your rules to the conclusions they force, using logic instead of predicting what sounds right. Your procedure lives in a decision tree you can read and approve. Your chatbot does what it’s good at: finding the facts in your files and apps. SMEme applies your procedure to those facts and works out what follows.
- Same facts, same outcome. Run a case today or next month and you get the same result.
- Your procedure, not the model’s opinion. Every outcome traces to a rule you approved, and you can open the tree and read it.
- It asks only what still matters. After each answer, SMEme works out which open questions can still change the outcome and asks only for those. Your agent doesn’t spend time on evidence that wouldn’t make a difference.
- It won’t guess. If the answers don’t settle the case, contradict each other, or come from sources that disagree, the report says so instead of picking one.
- It shows its work. The report lists the path, the evidence behind each answer, and which answers decided the outcome.
- You can ask before you act. Find out what would change the outcome, or what it would take to reach a different one.
- Changes are checked. Your past cases run every time you Deploy, so an edit can’t quietly change a decision you’ve already made.
Quick start
Connect SMEme to your chatbot, then choose and build your first decision tree with copyable prompts.
Quick start →Run a tree
Run a deployed tree in a new chat. Your agent answers from your files and apps and cites each source.
Run a tree →Ask more
Ask which answers decided the outcome, what would change it, and what it would take to reach another.
Ask more →Guides
-
Quick start
Connect, choose and build a tree, then Deploy and try it.
-
Run a tree
Run a tree on a case and read the report.
-
Ask more
Which answers decided it, what-if, and what it would take.
-
Building decision trees
Create, edit, save, and version your decision trees.
-
Plans & usage
Free and Pro limits, MCP tool-call metering, and AI-assisted build credits.
-
Deploy, list, & delete
Readiness checks, deploy, Listed/Hidden visibility, and permanent decision tree removal.
-
Download your decision tree
Export the saved graph as a JSON file from the dashboard or editor.
-
Upload decision tree
Upload a .smeme.json file you downloaded, or one someone shared with you.
-
Connect your agent
Build or revise drafts through MCP, add the connector, and run deployed trees against evidence.
About
SMEme is a product of Arista Labs — making expert knowledge machine-readable so high-stakes decisions can scale without losing accountability or human control. Visit aristalabs.ai.