Executive Overview
Alan is an AI-powered hardware design automation agent that accelerates RTL development and verification through intelligent task automation and deep EDA tool integration. It augments engineering productivity while fitting into existing toolchains and enterprise workflows.
- Executable Spec as Source of Truth: Machine-executable specs driving RTL and verification with traceability.
- 10x Faster Verification Setup: Automated testbench generation and configuration.
- Intelligent Design Assistance: AI-powered RTL generation and optimization.
- Seamless Tool Integration: Works with Cadence, Synopsys, and Siemens tools.
- Enterprise-Ready: Native JAMA/JIRA/CI-CD integration.
- Coverage Optimization: ML-guided coverage closure with targeted test selection.
- Requirements Intelligence: PRQ → HWRQ → HWDD/HWA with end-to-end verification linkage.
Full specification: see documentation/ALAN_PRODUCT_SPECIFICATION.md.
Key Capabilities
Executable Specification
- Spec generation and validation (SysML v2, YAML/JSON DSL).
- RTL derivation and verification artifact generation (UVM, SVA).
- Docs and traceability with change impact analysis.
Design Implementation
- RTL generation, FSM synthesis, and interface templates (AXI/AHB/APB/etc.).
- Refactoring and documentation extraction.
Verification
- Testbench and assertion generation; coverage planning and regression management.
- Coverage optimization and requirements-linked verification (PRQ → HWRQ → tests/covergroups).
- AMS co-simulation support (Spectre, CustomSim/PrimeSim XA, AFS).
Architecture
Modular architecture with core agent, knowledge layer (RAG), and MCP-based tool integration. Enterprise connectors include JAMA, JIRA, Git/Perforce, and CI/CD. Simulation tools MCP covers digital and mixed-signal simulators.
- Core: Orchestrator, LLM engine, sequential reasoning, state, execution, spec generator.
- Knowledge: Vector/graph search, continuous learning, executable spec storage.
- EDA: Cadence Xcelium/Spectre, Synopsys VCS/CustomSim, Siemens Questa/AFS.
Requirements & Coverage Flow
Alan derives HWRQs from PRQs, maps them to HWDD/HWA in the executable spec, and generates verification artifacts linked by requirement IDs. Coverage rolls up by HWRQ and PRQ with dashboards and sign-off checks.
- PRQ → HWRQ derivation with acceptance criteria.
- Consistency enforcement: implementation and verification links per HWRQ.
- Coverage optimization: gap analysis, targeted stimulus, regression pruning.
Technical Requirements
- Minimum: 8 CPU cores, 32GB RAM, 500GB SSD, Ubuntu 20.04+/RHEL8+.
- Recommended: 16 CPU cores, 64GB RAM, 1TB NVMe, 1Gbps network.
- LLM Compute: 8× NVIDIA H200 for on-prem inference, or access to state-of-the-art LLM APIs.
- EDA: Licenses for target simulators and analysis tools.
Licensing & Pricing
- Licensing Models: Perpetual, Subscription, Usage-Based, Enterprise.
- Pricing Tiers: Starter, Professional, Enterprise, Academic.
See section 15 in the full specification for details.