Tuple Tech
AI LayerPart of Stratos

AI Agents

Six purpose-built agents for IC design and infrastructure teams — each one understands your environment, not just your prompts.

  • Forge AI — cloud infrastructure from plain language, human-approved
  • Avestra — spec-to-SVA pipeline, <1% hallucination rate, sign-off-ready assertions
  • Analytics AI — deep insights across compute, licenses, and operations
  • Ask AI — conversational access to your full operational knowledge base
  • PV Agent — physical verification triage, traced to the source
  • DV Agent — design verification, test generation, and coverage
Infrastructure Simplified

From a sentence to running, verified infrastructure

Forge AI turns plain-language requests into deployed, verified AWS infrastructure — with a human approval step before anything is provisioned. It generates CloudFormation or Terraform, validates it, presents a plan, and pushes the final IaC back to Git after deployment.

Human-in-the-loop approval

Every plan — resources, warnings, and generated IaC — is presented for review before anything is provisioned. Choose Approve, Discuss, or Reject. Nothing runs without your sign-off.

  • Automated infrastructure provisioning from natural language
  • Drift detection and autonomous remediation
  • IaC generation and review (Terraform, CloudFormation)
  • Multi-cloud resource orchestration (AWS, GCP, Azure)
  • Policy enforcement and compliance guardrails
  • Auditable change history and rollback support
Forge AI · Workflow cd25974b rev 1us-east-2
Using AWS CloudFormation agent, create a VPC with public & private subnets, NAT gateway, and a webserver security group.
Deployment Plan · Workflow cd25974b rev 1
14 to create · 0 to modify · 0 to destroy
createAWS::EC2::VPC
createAWS::EC2::Subnet (×4)
createAWS::EC2::NatGateway
createAWS::EC2::SecurityGroup
Approval required
Review the plan and choose Approve, Discuss, or Reject.
✓ ApproveDiscussReject
Deployment Complete
14 resources deployed and verified. IaC pushed → branch: main
SystemVerilog Assertions

From spec and RTL to sign-off-ready assertions

Avestra transforms specifications and RTL designs into production-grade SystemVerilog assertions using an 8-agent orchestrated pipeline. Built on 35 years of CPU/ASIC/SoC domain expertise, it achieves a <1% hallucination rate — delivering sign-off-quality SVA suites in minutes, not days.

<1% hallucination rate

Proprietary RAG trained on Ashok Mehta's 3 SVA textbooks, 22 US patents, and 35 years of CPU/ASIC/SoC expertise — domain intelligence no generic LLM can match. Deterministic, auditable output every run.

  • 8-agent orchestrated pipeline: spec analysis, RTL analysis, assertion generation, testbench gen, cross-check, and more
  • Spec ↔ RTL inconsistency detection — catches missing states, illegal transitions, and timing violations before first sim
  • Production-grade SVA suite — clocked assert, assume, cover, vacuity-guarded, multi-clocked, every assertion traced to source
  • Coverage-driven SV testbenches with UVM-ready architecture, directed tests, and response checkers
  • LLM-agnostic pipeline — swap the underlying model without affecting output quality or structure
  • Works with Questa (Siemens), VCS (Synopsys), and Xcelium (Cadence) — no simulator vendor lock-in
Avestra · 8-Agent Pipeline · RunningAXI_slave.sv + spec.docx
1/8
Spec Analysis AgentFSM extracted · 14 temporal rules
2/8
RTL Analysis AgentCOI traced · 2 clock domains
3/8
RAG Context AgentDomain knowledge retrieved
4/8
Inconsistency Detection2 undefined transitions flagged
5/8
SVA Generation Agent18 properties synthesized
6/8
Coverage Analysis Agent6 cover points · vacuity verified
7/8
Cross-Check AgentAll assertions validated vs spec + RTL
8/8
Testbench Gen AgentUVM testbench + directed tests ready
Output Ready · Silicon-Grade
18 assertions · 6 cover points · SV testbench · inconsistency report
Questa·VCS·Xcelium

Deep operational insights across your entire environment

Analytics AI surfaces actionable insights from your infrastructure, EDA tools, and operational data. It aggregates usage across cloud, on-prem, and licenses — giving engineering and finance teams a unified view with anomaly detection and natural language querying.

Natural language over your operational data

Ask questions like “which team spent the most on EDA compute last quarter?” or “flag licenses approaching renewal.” Analytics AI queries your data and returns structured insights — no SQL, no dashboards.

  • Infrastructure usage attribution across teams, projects, and cost centers
  • Anomaly detection across compute, license, and operational metrics
  • Trend analysis and utilization insights across your full environment
  • Custom dashboards and reports for engineering and finance teams
  • Cross-platform data aggregation — cloud, EDA, and on-prem
  • Natural language queries over your operational data
Analytics AI · Operational InsightsLive · All sources
Analytics AI · Usage Insights
May11/14 seats
Jun13/14 seats
Jul15/14 seats
analytics ai — Projected seat shortage in July (+1 seat). Peak demand driven by tapeout sprint for Block A. Recommend purchasing 2 additional VCS seats before June 15 to avoid queue blocking. Renewal date: Aug 3.
Ask AI · Activeconnected to Stratos
you

Why did our GPU cost spike on March 14th?

···

Querying Stratos cost data and deploy history…

ask ai

Spike correlates with the model-training-v3 batch job that ran 14:00–22:00 UTC. 3× normal GPU utilization. Triggered by the ml-pipeline team. Estimated excess cost: $1,240. Recommend adding a budget alert for this job class.

you

Has this happened in the last 7 days?

ask ai

Yes — 2 incidents in 7 days. Both resolved within SLA. Root cause: Redis connection exhaustion during peak traffic (8–9pm UTC). Alert threshold already adjusted after incident #1.

Ask anything about your infrastructure…

A chatbot that actually knows your environment

Ask AI is the conversational interface to your entire operational knowledge base. It retrieves answers from Stratos runbooks, incident history, monitoring data, and documentation — giving your team instant context without digging through dashboards or tickets.

MCP connector support

Ask AI connects to any data source via MCP — your ticketing system, monitoring tools, databases, and internal wikis. One interface for your entire operational knowledge base, with self-hosted or cloud LLM inference.

  • RAG across Stratos docs, runbooks, and tickets
  • Database and analytics query access
  • MCP connector support for any data source
  • Multi-turn conversation with full context retention
  • Team-scoped knowledge with access controls
  • Works with self-hosted and cloud-hosted LLMs

Physical verification triage — traced to the source

PV Agent triages DRC, LVS, and ERC failures — identifies root causes, traces violations back to the responsible commit, and surfaces fix suggestions with full runbook context. Sign-off moves faster when every error already has an answer.

IP stays on-prem

PV Agent supports self-hosted LLM inference — your design data never leaves your environment. Required for tape-out sensitive and export-controlled designs.

  • DRC, LVS, and ERC triage with root-cause identification
  • Fix suggestions with cross-project runbook context
  • Commit-aware: traces violations back to the responsible change
  • Batch violation analysis across full chip or block level
  • Self-hosted or cloud LLM inference — IP stays on-prem
  • VS Code extension for inline access during sign-off
pv_agent.log
drc_summary.rpt
VS Code Extension
PV Agent · Physical Verification
youPV run failed on VCO_LDO — what triggered it?
pvParsing DRC log and cross-referencing recent commits…
pv agent — Metal-3 spacing violation near bias resistor array (4 instances, rule M3_SP_MIN). Introduced in last commit by @mpark. Fix: increase spacing by 0.04µm on nets VN_BIAS[2:0]. Same pattern resolved in BGR_CORE — runbook #441 linked.
youAre there similar violations elsewhere in the block?
pv agent — Scanning full block for M3_SP_MIN pattern… 2 additional instances found in VCO_CORE near net VBG_REF. Same fix applies. All 6 violations can be resolved in one edit pass — diff ready.
dv_agent.log
coverage.rpt
VS Code Extension
DV Agent · Design Verification
youCoverage is stuck at 78% on block MEM_CTRL. What's missing?
dvScanning coverage report and testbench constraints…
dv agent — 3 uncovered FSM states in mem_ctrl_arb: IDLE→STALL→RETRY path not exercised. Suggested: add directed test for back-pressure + retry sequence. Estimated coverage lift: +9%. Testbench snippet ready to insert.
youGenerate the testbench snippet.
dv agent — Snippet generated for IDLE→STALL→RETRY. Inserted at line 247 of mem_ctrl_tb.sv. Re-run sim to confirm coverage closure.

Test generation and simulation — without the manual effort

DV Agent understands your testbench, coverage reports, and design specs. It generates tests for uncovered paths, runs simulations, and automates regression management — so your team closes coverage faster without manual effort.

VS Code Extension

DV Agent is available directly inside VS Code. Inline suggestions, diff-aware context, and chat — without leaving your editor. IP stays on-prem with self-hosted inference.

  • Test generation for uncovered paths, FSM states, and corner cases
  • Simulation run management and regression automation
  • Coverage analysis and gap identification across functional and code coverage
  • Directed test authoring from coverage holes — no manual testbench edits
  • Spec-aware context from your runbooks and design docs
  • Formal verification setup and result interpretation

Put AI to work in your environment

See how all six agents work together inside your existing infrastructure and EDA workflows.