closed beta · accepting research partners
Sutra.
सूत्र · the thread that holds verification together

An end-to-end verification copilot that doesn't forget your hierarchy.

From design verification through post-silicon, Sutra plugs into your existing EDA flow and reads spec, RTL, waveform and log as one connected context. It writes testcases and UVM that understand the block they verify, and collapses the four windows engineers normally tab between into a single, hierarchy-aware view.

The bottleneck

Hierarchy is the memory of VLSI work. Most tools flatten it away.

The bottleneck looks different in every phase of a silicon engagement. The identifier in your query exists in five blocks. The interface you are about to change has consumers no full-text search can enumerate. The spec clauses without assertions never appear in any positive match. None of these are content questions. The answers are relationships the agent has to walk, not text it can match.

Four moments from a single engagement, drawn from a fabless team bringing up cortex_v3 (a derivative SoC built on cortex_v2 with a new accelerator). The cast carries from one panel to the next. Use the arrows below, the dots, or swipe on touch to step through.

Cost of running an agent

Up to 10× lower token cost than flat-context agents.

Across the industry, teams that adopted general-purpose coding agents over the last year have begun to report that their token spend grew exponentially with engagement size, and that an annual agent budget could be consumed in a single quarter. The cause is structural rather than incidental. A flat-context agent reloads, re-flattens and re-explains the codebase to itself on every session, so the same hierarchy is paid for again and again in tokens. Sutra holds the hierarchy and its cross-references as a persistent knowledge graph, retrieved by structure rather than by string. On multi-block engagements the difference compounds across sessions and reaches an order of magnitude.

≤ 10×
lower token cost on multi-block engagements vs flat-context coding agents
measuredon real multi-block workloads basispersisted knowledge graph
Hierarchy as substrate

The design hierarchy is the first-class object. Spec, RTL, waveform and log hang off the same nodes and are looked up by structure rather than re-flattened into a context window.

Persistent knowledge graph

Context is held across sessions and across engineers. The model does not pay tokens to rediscover what the team already established yesterday.

Structural retrieval

Each query pulls the minimum slice of the graph that is relevant to the question, rather than the largest window the model can accept. Sharper context, fewer tokens, fewer wasted iterations.

In practice Engagements stay within budget across the full pre-silicon to post-silicon cycle. Verification can run continuously rather than rationed against an exhausted token quota.
query is overflow_chk in alu_top (cortex_v3) handling signed values correctly?
flat · semantic retrieval
~24k tok 2 / 13 rel
overflow_chk @ alu_top_v11.6k
overflow_chk @ alu_top_v21.6k
overflow_chk @ alu_top_v31.7k
overflow_chk @ cortex_lite_alu1.4k
AXI overflow snippets2.1k
spec §3.5.1 · unsigned arith1.9k
spec §3.5.2 · signed-overflow rule2.0k
+ 6 more noise chunks · debug logs, prior tickets, retired forks…
⚠ context poisoning likely · 11 irrelevant chunks compete for the model's attention
sutra · graph retrieval
~1.8k tok 3 / 3 rel
alu_top.sv (v3 instance)0.9k
spec §3.5.2 · signed-overflow rule0.6k
cortex_apu.signed_mode_cfg0.3k
— window otherwise empty · budget free for reasoning —
✓ clean signal · selected by binding / clause / config edges
also Across a working session, a smart handoff carries the graph state forward — follow-up queries reuse the same context without rereading.
The other half

Hierarchy is the context. Discipline is what does the work.

A hierarchy-aware model that still prompts itself like a junior engineer will still answer like one. AI software-engineering agents have learned over the last year that raw model capability matters less than the curated workflow you wrap it in. The same lesson applies, with greater force, to silicon verification, where conventions are denser and "almost right" is indistinguishable from "wrong".

Sutra ships a library of verification skills: prompt scaffolds, generation harnesses and reviewer-passes, authored by experienced verification engineers and invoked at the exact stage of the flow they belong to.

a skill at every stage of the flow
01
Spec ingest
verification lead

Pulls test intent, corner cases and coverage targets out of the document, the way a lead reads a spec rather than the way a search engine does.

02
UVM scaffolding
uvm methodologist

Methodology-correct env, agents, sequencers and scoreboards. Bus-handshake templates per protocol, not boilerplate copied from another project.

03
Stimulus generation
verification lead

Constraints and sequences targeted at the block under test, derived from the corners the spec actually called out.

04
Coverage closure
coverage engineer

Missed-bin sampling heuristics and a closure-loop iteration drawn from real coverage-closure engagements.

→ verification + sign-off
05
Formal drafting
formal practitioner

SVA and assertions drafted from spec clauses, with vacuity and reachability checks built into the pass.

06
Debug
post-silicon debugger

Walks waveform → log → RTL with a senior debugger's heuristics. Lands on a line, cites the spec clause that anchors the expected behaviour.

07
Reviewer-pass
senior ic engineer

Lint, style and coverage review applied to every output before it ever reaches you, the way a senior signs off a junior's patch.

Each skill is owned by a domain expert and refined from real engagements.

Capabilities

Four windows. One context.

An engineer normally chases information across four open windows: spec PDF, RTL editor, waveform viewer, and simulation log. Sutra holds the four together as one queryable context. Ask a question in plain English and receive answers grounded in your actual design.

01

Doc-aware testcase generation

Reads your spec alongside your RTL. Writes testcases that exercise the corner the document called out, rather than generic stimulus.

Stim · UVM · constraints
02

UVM scaffolding, hierarchy-correct

Generates env, agent, sequencer and scoreboards that match your block's boundary. No copy-paste from another project's testbench.

SystemVerilog · UVM 1.2
03

Debug with log + wave + RTL

Ask "why does carry_out fail when op_a is signed?" and Sutra walks the waveform back, lands on the RTL line, and cites the spec clause that defines the expected behaviour.

Root-cause · cited
04

Post-silicon correlation

The same hierarchy carries through to silicon bring-up. Lab logs land on the same nodes your testbench did, so issues are traced back to the right RTL block, not re-investigated from scratch.

Pre & post-silicon
05

Natural-language design queries

"Where else does this signal toggle in the next 200 cycles?" is answered against the actual waveform, scoped to the right module, in seconds.

Cross-window query
06

Reviewer-style suggestions

Inline lint and review for testbench and constraints. Flags untested branches, unused covergroups and missing assertion targets.

Coverage · assertions
07

On-prem & air-gapped

Your design never leaves your perimeter. Self-hosted from day one. No outbound calls to anyone's API. Audited for fabless and IDM workflows.

Local-first · audited
08

Flow-agnostic plugin

No rip-and-replace. Sutra sits next to your existing simulator, waveform viewer and bug tracker. Use what you already use.

Make / Tcl / CI hooks
How it plugs in

Sutra sits underneath your existing flow, not in place of it.

No new IDE to learn, no migration. The existing stages of your verification flow keep their inputs and outputs; Sutra layers a shared context-graph beneath them so each stage can see the others.

Stage 01

Spec

PDF · DOCX · Confluence

Stage 02

RTL · TB

SV · UVM · constraints

Stage 03

Sim · Wave

FSDB · FST · VCD · logs

S

Sutra · the shared context-graph

The same hierarchy of nodes is exposed to every stage, persisted across sessions. Each stage queries the others by structure, not by file path.

spec.ref rtl.scope wave.signal log.event cov.bucket
also Teams already running Codex, Claude Code or another agentic framework can adopt Sutra as a plugin into that framework. Same prompts, same workflow, hierarchy-aware context under the hood. No tool migration required.
Derivative-friendly

Built for lead SoCs and every derivative that follows.

Most fabless roadmaps run on a small number of lead SoCs, sometimes referred to as driver SoCs, and a longer tail of derivatives built on each. Today every derivative earns its verification collateral again from scratch, because the testbench is wired to a specific block list, address map and corner-case set. Sutra treats the lead's hierarchy as a reusable substrate. When pointed at the derivative's spec and RTL, it computes the structural delta against the parent, carries unchanged collateral forward verbatim, and regenerates only the portions that have actually changed.

Lead SoC · already in Sutra's context
  • Block list, address map, clock and reset tree
  • UVM environment, agents, sequencers and scoreboards
  • Closed coverage with sign-off waivers
  • Spec, RTL, waveform and log cross-references
diff →
Derivative · the delta Sutra adapts
  • Swapped IP and new peripherals identified automatically
  • Retargeted address map and pin multiplexing propagated through the environment
  • Constraint deltas and new corner cases regenerated rather than rewritten
  • Delta coverage traceable back to the parent's closure
Turnaround Derivative bring-up is compressed from quarters to weeks. The parent's sign-off evidence is carried forward rather than earned again.
Productivity outcomes

30–60% productivity gains, measurably.

Across early-access benchmarks, senior verification engineers using Sutra show consistent 30–60% productivity gains against their own without-Sutra baseline, measured on real blocks rather than toy ones. Full methodology available on request.

Sutra productivity gain
30–60%

Measured across testbench scaffolding, debug turn-around and coverage closure, against each engineer's own historical baseline.

self-controlled · senior engineers
Why the range
block dependent

Gain scales with how hierarchical and cross-referential the block is. Pipeline-heavy designs sit at the top end; flat RTL at the bottom.

benchmark range, not best-case
Research & bench

Built by people who've shipped silicon.

Sutra's team is a mix of senior verification engineers from NXP, Freescale, Uber, Flipkart and Directi. We are currently reaching out to leading engineering institutes for research collaboration.

NXP
ex · semiconductors
Freescale
ex · semiconductors
Uber
ex · platform
Flipkart
ex · infra
Directi
ex · product engg.
Open call
Heading a VLSI / electronics department? We're currently reaching out to engineering institutes for a research collaboration on hierarchy-aware verification. Write to us.
Pilot programme

Try Sutra on a block, IP or full SoC.

We invite verification teams to evaluate Sutra on a workload of their choice, from a single block to a complete SoC. Sutra is designed to hold hundreds of specs, RTL files, waveforms and logs together as one queryable context, so the scale of your design is not a constraint. A Sutra engineer pairs with your verification lead through the evaluation, and you receive a quantified report measured on your own design.

  • → Scoped to your workload, from a single block to a full SoC
  • → Paired with a Sutra engineer through the evaluation
  • → Quantified report run on your data
  • → No commitment beyond the evaluation

Start a conversation

Email sales@swadeza.com We reply within one working day

We reply within one working day.

FAQ

The three questions we get most.

Does Sutra replace my existing simulator or waveform viewer? +
No. Sutra is a context layer that sits next to your existing tools. Your simulator, waveform viewer, lint and bug tracker all stay. Sutra reads their outputs and answers questions across them; it never tries to be them.
Where does my design data go? +
Wherever you already trust your code to go. If your team is already using an agentic coding assistant such as Claude Code, Codex or Cursor, Sutra plugs into that same backend, so your design data follows the same path, governance and audit trail your existing subscription already covers. If you do not have one in place, we include a managed inference subscription in the engagement so the cost is rolled into a single line item.
How is this different from agentic coding tools that already exist? +
Three things, and they compound. First, the hierarchy itself is the context: spec, RTL, waveform and log are persisted as one graph and looked up by structure rather than re-flattened into a window. Second, the harness injects verification-engineering discipline at each stage of the flow (UVM scaffolds, coverage-closure heuristics, formal drafting, debug playbooks), all authored by domain experts rather than by a general-purpose assistant guessing what a verification engineer would do. Third, those two together let us retrieve sharper context with far fewer tokens and far fewer wasted iterations, so the output is of higher quality and lands in less time, at lower cost, than asking a flat-context coding agent the same question repeatedly. We are happy to share the technical case in a private session.