Choose Your Track
Most teams do not need the same entry point. Start where your implementation risk is lowest:
If you want deterministic integrations and internal system sync, start with the REST API. If you want Claude, Cursor, or another client to work directly with Lev data and tools, start with MCP. If you want terminal-first workflows, scripting, or to give an LLM coding agent shell access to Lev, start with the CLI. If you want to ship a specific workflow quickly, start with a recipe and then branch into the reference docs as needed.
Use structured endpoints, predictable envelopes, and machine-readable docs when you need sync jobs, back office automation, or product integrations.
Use Lev from Claude, Cursor, or another MCP-compatible client when you want tool use, retrieval, and human-in-the-loop workflows.
Use the CLI for ad-hoc queries, shell scripts, CI/CD pipelines, CSV exports, and LLM coding agents that need structured JSON output from Lev.
Use implementation recipes when you already know the workflow you want to build, from sync pipelines to a broker-facing copilot.
Start by Job to Be Done
Use these starting points if you already know the workflow you want to ship:
Start with Quickstart, then move to Pagination, Filtering & Sorting, and Data Sync Patterns.
Start with Deals, Contacts, Companies, and Account & Team.
Start with MCP Setup, Agent Workflows, Lender Search, and Term Sheets.
Start with CLI Setup, then CLI Commands and CLI Workflows for CI/CD, scripting, and agent integration patterns.
Start with Build a Broker Copilot for architecture, tool selection, retrieval patterns, and implementation order.
Persona starting points
- Platform engineers: Focus on API overview, auth, pagination, and sync patterns first. Use the CLI for quick terminal-based validation and CI/CD integration.
- Product teams: Focus on the core resource pages and start with the objects you need to expose in your UI.
- AI teams: Focus on MCP setup, agent workflows, lender search, and CLI workflows for LLM agent integration.
- Broker operations: Focus on broker copilot, placements, term sheets, and account/team workflows.
What Lev Provides
The hard part is not building a UI anymore. The hard part is trustworthy CRE data, workflows, and intelligence. Lev gives you a shared foundation for all three tracks.
First Builds
These are the fastest useful things to ship if you are starting fresh:
- A first successful API read: Validate auth and fetch a small page of deals in under five minutes.
- A durable sync job: Use cursor pagination, checkpoint storage, and idempotent writes to your own systems.
- An AI-native lender search flow: Let an agent gather deal context, trigger lender search, and summarize the best next actions. (Lender matching via MCP is coming soon; the REST Lender Search endpoint works today.)
- A broker copilot prototype: Combine deal retrieval, lender search, placements, and term sheets in one guided workflow.
The best first step is the one that validates your riskiest assumption fastest, whether that is auth, MCP connectivity, or a production workflow recipe.