On September 28, 2026, Synopsys announced the Autopilot Platform and its AgentEngineer portfolio for engineering workflows across six domains. Synopsys plans availability for the end of 2026 and said more than 50 customer engagements were underway at the time of the announcement.

Six engineering domains

Synopsys says the platform is designed to coordinate AI agents across:

  • Verification
  • System validation
  • Implementation
  • Analog
  • Manufacturing
  • Simulation and analysis

Examples of targeted work include coverage closure in verification, software bring-up and validation, multi-die 3DIC assembly, and physical design tasks such as PPA closure. PPA stands for power, performance and area. Other examples include analog layout synthesis, design migration, mask synthesis, and analysis of combustion, resonance and signal integrity.

How the agents divide the work

Autopilot is the orchestration and governance platform; AgentEngineer is the portfolio of engineering agents it coordinates. The workflow is described in layers: long-horizon agents organize work across a broader engineering goal, task agents handle bounded jobs, and engineering tools execute those tasks.

Synopsys describes platform components that include orchestration, skills, memory, telemetry and governance. It also describes a cognitive model that provides context intelligence. In verification, for example, a task agent can work on coverage closure. The broader portfolio is intended to apply the same agent-based approach to several distinct engineering domains.

Human approval checkpoints remain

Synopsys characterizes its agents as Level 5 on its own autonomy scale. People define the guardrails and retain approval checkpoints, and teams can set points to review or redirect the workflow. The Level 5 label is Synopsys’s own framework designation.

What the reported results measure

Synopsys reported up to 50× faster verification closure and 20% higher coverage from July 2026 work with NVIDIA, compared with Synopsys workflows without AgentEngineer. Both figures refer to that verification work.

Fujitsu reported a 10–30% productivity boost in RTL code generation using Verification AgentEngineer capabilities. RTL means register-transfer level, a way of describing a digital circuit’s data flow and operations. The work Fujitsu described included generating SystemVerilog assertions, wrapper and parameterized modules, and refactoring code.

AheadComputing reported reduced manual engineering effort with Implementation AgentEngineer across the workflow from RTL handoff through implementation, engineering change orders (ECOs) and signoff. Its account was qualitative.

Model and infrastructure choices

Synopsys says customers can choose models, infrastructure, data, agents and workflows. The described options include commercial, open-source or fine-tuned models, with deployment on Synopsys Cloud, a customer’s cloud or on premises.

Synopsys lists access controls, encryption and runtime guardrails among the protections for customer, partner and Synopsys intellectual property.