Build with the wizard
Describe your topic, add expert-approved text or .pdf / .docx sources,
point at URLs, or enable AI web research. SMEme extracts the decision-critical factors and drafts a branching decision tree.
neuro-symbolic AI at scale
From your expertise to a tool your agent runs
SMEme is the reasoning layer your agent calls when judgment has to be repeatable — not a saved prompt or runbook that re-burns tokens re-explaining your methodology every run. Build and edit your decision tree however you work; your agent calls it on real evidence and gets back a deterministic, attributable answer.
01
You build the reasoning
Two ways in — build and edit through the guided webapp, or directly through your agent over MCP. Both compile to the same deployed decision tree.
Webapp
Guided & visual
Refine in the editor
Tune wording, branches, or conclusions in the graphical/text editor. A live validation panel runs continuously with a “How to fix” hint under every issue.
Connect your agent
Add the SMEme MCP server from your dashboard and sign in. Your agent will be ready to call the decision tree as soon as it’s deployed.
Agent · MCP
In chat or agentic runs
Connect your agent
Connect to SMEme through MCP and authenticate — the authoring tools your agent uses to generate and revise decision trees are MCP tools too.
Generate in chat
Ask your AI chat to help you build a decision tree with SMEme. Add operating procedures, prompts, runbooks, templates, files, or websites to the working context. Your agent harness can also use other MCP connectors (email, cloud drives, CRM) to pull source material. It drafts a decision tree you can refine in chat; once approved, it formats and uploads the draft to SMEme for validation. Drafts are not live for agents until you Deploy from the dashboard.
Revise via MCP
Keep the conversation going to refine it — your agent can revise the draft directly, or you can open it in the visual editor any time to fine-tune by hand.
Self-hosted deployments: ask your admin to turn this feature on.
Manage and Run your decision trees ↓
Manage your decision trees in the SMEme platform
Manage your decision trees and access from your SMEme dashboard. Deploy runs readiness checks — structure, reachability, coverage, logic consistency, node completeness — and compiles the live version your agent calls.
Listed controls whether your agent can discover it. List or hide, download, or delete any decision tree, any time — nothing is locked in.
02
Your agent runs it
Two ways to run decision trees — directly in AI chat harness or agentically through runbooks or workflows like LangGraph or n8n. Either way, SMEme evaluates the same deployed decision tree.
In chat
Leverage SMEme tools directly in chat. You can run decision trees by name. Your AI gathers facts from approved sources, formats them, and sends them to SMEme for deterministic reasoning against your authored decision trees. Structured responses return directly to the chat context for further processing.
5
Find & structure
“Run our enterprise deal-qualification decision tree on this opportunity.”
Task matches a deployed tree. Gathering evidence from your connected sources…
Docs
CRM
arr_band$380,000
strategic_accountYes
discount_request_pct22%
6
Validate & reason
Required fields ✓ · Valid options ✓ · Sources attached ✓
Validated — answers complete
Strategic-account test
Multi-year commit gate
Discount-authority routing
7
Report & continue
VP approval required — 22% discount exceeds AE authority.
“What if they commit to three years at list price?”
Within standard authority — can advance without escalation.
Human-in-the-loop, start to finish.
Agentic trigger
e.g. LangGraph, n8n
SMEme calls are single, deterministic evaluates — not hedges. This run calls two different decision trees: deal qualification up front, and counter-offer authority only if escalation is triggered, several ordinary steps later.
Two decision trees, one agentic run.
Your subscription meters MCP tool calls and AI-assisted builds — deploy once; every run uses your monthly allowance. See pricing for current limits.
On SMEme-hosted, you control whether to edit, download, or delete your decision trees. Deterministic evaluation runs on SMEme infrastructure, not inside the LLM provider. Optional AI-assisted generation sends disclosed content to configured providers; see our Privacy Policy and Terms of Use.
Turn your expertise into a tool agents can run
Build your first decision tree free — in the wizard or straight from your agent. Connect once you're ready to deploy.