$ chipchamp "the FIFO smoke test is failing: find it, fix it, prove it"
⚙ sim.run tb_sync_fifo J-0021 · FAIL · fifo data mismatch @220000 exp=a001 got=a000
⚙ wave.compare J-0021 vs golden first divergence: dut.rd_data @ 45000
⚙ design.cone rd_data assign rd_data = mem[wptr];
✱ root cause: rd_data is driven from mem[wptr] instead of mem[rptr]
✎ edit sync_fifo.sv (1 line changed)
lint  smoke_sim  affected_regress  coverage_baseline  no_gaming
✔ report.done accepted
🔏 evidence bundle signed sha256:845bc27a… verify: True
No human read the waveform.
That is what this talk is about.
HeiChips Summer School · Heidelberg · August 2026

Introduction into Agentic AI for Hardware Design and Verification

Riadh Ben Abdelhamid

Novel Computing Technologies Group · ZITI · Heidelberg University

RTL Forge and chipchamp: two working systems, shown later in this talk

This talk

Act I

What "agentic" actually means: beyond the buzzword

Act II

Why hardware: the best domain for agents, and the hardest

Act III

State of the art: research, industry, and silicon that already shipped

Act IV

Using it today: harnesses, tools, models: open vs. closed

Act V

Two working systems: RTL Forge & chipchamp, with honest measurements

Act I · What "agentic" actually means

Three eras in five years

2021

Autocomplete

Copilot finishes your line. You are still the engineer; the model is a fancy keyboard.

2023

Chat

ChatGPT writes a whole module. You copy-paste, run, and debug it by hand. The loop is you.

2025→

Agents

The model runs the tools, reads the errors, fixes its own work, and comes back when it's done. The loop is the machine.

The difference is not a bigger model. It is an architecture around the model.

Act I · What "agentic" actually means

An LLM is not an agent

One-shot LLM

prompt model output

You get what you get.
Errors ship silently.

Agent

generate checkreal tool done fix pass fail

A loop with tools, feedback, state.
Errors become inputs.

Act I · What "agentic" actually means

Five ingredients of an agentic system

1

Tools: the model can run things: linters, simulators, file I/O, synthesis

2

Feedback loops: tool output flows back into the next attempt

3

State & checkpoints: long tasks survive failure, resume, and can be audited

4

Routing: when something fails, whose fault is it? Fix the right artifact

5

Decomposition: split big problems into sub-problems with contracts between them

Keep these five in mind. Each one reappears, concretely, in Act V.

Act I · What "agentic" actually means

In software, this already happened

~2%~96%
real GitHub issues fixed end-to-end:
SWE-bench, Oct 2023 → SWE-bench Verified, mid-2026
hours
of autonomous multi-step work per task is now routine
for coding agents (Claude Code, Codex, Gemini CLI…)
  • Agents open the repo, reproduce the bug, edit, run tests, iterate, unattended
  • Adoption went from research demo to daily tool for millions of developers in ~2 years
  • Why it worked: software has executable feedback: compilers and tests

Hold that thought: what does hardware have?

SWE-bench: Jimenez et al., Oct '23, best model 1.96% (full set) · Verified subset: OpenAI, Aug '24 (33.2%) · 49% late '24 · 80.9% late '25 · ≈96% mid-'26, vendor-reported, tracked by aggregators · press C for sources
Act I · What "agentic" actually means

The load-bearing distinction

Estimated

"Model, does this code look correct to you?"

Plausible. Confident. Sometimes fiction.

Measured

Verilator compiled it. The simulation ran. The assertions held.

Ground truth, independent of the model's opinion.

An agentic flow is only as trustworthy as its least-grounded verdict.

Act II · Why hardware

Verification is eating the schedule

60–70%
of engineering effort on a chip project goes to verification
≥1 : 1
verification engineers per designer on many ASIC teams, and rising
$M + months
cost of a respin: there is no hotfix after tape-out

And the industry is short tens of thousands of engineers. The work is growing faster than the workforce.

2024 Siemens EDA / Wilson Research Group functional-verification study · Lux Capital, "The looming labor crisis in chip design" · press C for sources
Act II · Why hardware

So why didn't the software wave carry hardware with it?

Data scarcity

Public Verilog is a rounding error next to public Python. The best RTL on Earth is locked inside companies.

Absolute correctness

"Mostly works" is fine for a demo app. It is meaningless for silicon.

Tribal knowledge

Methodology, tool flags, house lint rules. Little of it is written down where a model could learn it.

So the naive move, "just prompt an LLM for Verilog", disappoints. Something else is needed.

Act II · The thesis

Hardware is not the hardest domain for agentic AI.
It may be the best one.

prose, UX, strategy…no ground truth
softwaretests, if someone wrote them
hardwarelint · simulation · formal · synthesis · equivalence:
cheap, deterministic oracles

An agent can know its RTL is wrong: mechanically, in seconds, for free.

Act II · Why hardware

…but the loop must be rebuilt for hardware

Feedback latency

pytest: seconds. A regression: hours-to-days on a licensed farm.

Giant ground truth

Gigabyte waveforms, coverage databases, 10⁵-path timing reports. Nothing fits in a context window.

The cost of wrong

A bad software merge gets reverted. A bad chip gets respun.

Metered resources

Simulator licenses cost $50k–300k per seat-year. An agent that spams the farm is fired.

Scale & fragmentation

10M-line mixed SV/VHDL codebases, filelists, generators, Perforce.

IP confidentiality

Foundry NDAs, export control. "Upload the chip to a cloud API" is often simply illegal.

The model is the easy part. The loop around it is the engineering.

Act III · State of the art · research

What the benchmarks say

VerilogEval-Human pass@1 · % (higher is better)

0255075100 GPT-4 · 2023 43.5 GPT-4o · 2024 63 frontier · 2026 ~97 VerilogCoder 94.2 MAGE 95.7 CVDP · best ~34
LLM, prompting agentic system realistic benchmark

Agents beat prompting: VerilogCoder gains +33.9 points over the best non-agentic run of the same class of model.

Then the benchmark saturated, so NVIDIA built CVDP: 783 realistic tasks. Best model at launch: ~34%.

And one year later, a purpose-built scaffold (generator + reflector + coordinator) puts open models at 97% on CVDP's agentic categories. The climb came from the loop, not a bigger model.

The lesson, twice over: the loop is the product.

VerilogEval (NVIDIA ICCAD'23, v2 2024) · VerilogCoder (AAAI'25) · MAGE (DAC'25) · CVDP (NVIDIA 2025, now hosted by Si2) · ACE-RTL (NVIDIA, July 2026) · press C for sources
Act III · State of the art · research

A field assembling itself, fast

generatespec → RTL

  • ChipNeMo: domain-adapted LLMs for chip design (NVIDIA '23)1
  • AutoChip: the first sim-feedback generation loop2
  • Fine-tunes (RTLCoder, CodeV, CraftRTL): a 7B reasoning fine-tune beats 671B DeepSeek-R1 on RTLLM3

verifytestbenches, coverage, formal

  • Auto-UVM machines: UVM², HAVEN; coverage-closure agents4
  • AssertLLM: spec → SVA assertions at ~89% correctness5
  • NVIDIA FVEval: frontier models still weak at formal6

debugtriage & repair

  • RTLFixer: ~98.5% of syntax errors auto-repaired7
  • VerilogCoder: an agent with a waveform-tracing tool8
  • Marco timing-debug agent: solves 86% where the baseline solved 0%9

implementthrough the physical flow

  • ChatEDA drives RTL→GDSII tool flows10
  • Marco DRC-Coder: sub-3nm DRC rules, weeks → ~4 min9
  • ORFS-agent tunes OpenROAD better than Bayesian optimization11

And a hopeful result for this room: small open models + good scaffolds close much of the gap.12

A selection; the survey "The Dawn of Agentic EDA" (2025)13 maps the field · press C for sources
Act III · State of the art · silicon

This is no longer hypothetical silicon

2023

Chip-Chat / QTcore-A1: first LLM-architected chip ever taped out. A microcontroller co-designed in ChatGPT conversation, on Tiny Tapeout.1

2023→

AlphaChip (DeepMind): RL floorplanning used in three generations of Google TPUs; layouts in hours instead of weeks.2

2025

QiMeng-CPU (CAS): fully automatic CPU design. v1 taped out (Intel-486-class), v2 a superscalar ~Cortex-A53-class core, FPGA-validated.3

2025

AlphaEvolve: a Gemini-powered agent rewrote Verilog inside an upcoming TPU. Human-verified, shipped into the design.4

2026

OpenAI × Broadcom "Jalapeño": inference ASIC, initial design to tape-out in 9 months, OpenAI models credited in the flow.5

2026

Eight groups of high-school students, 90 minutes: tapeout-ready VGA chips via an in-browser LLM Tiny Tapeout flow. One class session.6

1 Blocklove et al. '23 · 2 Mirhoseini et al., Nature '21 + DeepMind '24 · 3 CAS ICT '25 · 4 DeepMind, May '25 · 5 OpenAI × Broadcom, June '26 · 6 "From RTL to Prompt Coding", Jan '26 · press C for sources
Act III · State of the art · industry

Copilots → agents → "autonomous engineers"

Synopsys

  • DSO.ai: 700+ production tapeouts1
  • AgentEngineer: "L4" spec→RTL workflows2
  • Autonomous DV orchestrator with NVIDIA, 2H263

Cadence

  • Acquired ChipStack → "AI Super Agent"; used at NVIDIA, Qualcomm, Tenstorrent4
  • Claims an "L5 virtual engineer": 5-week verification loop → <1 day5

Siemens EDA

  • "Self-verifying" agentic workflows with NVIDIA6
  • Every agent decision validated against deterministic physics engines (Questa, Calibre…)6

The startup wave

  • ChipAgents $134M; with STMicro, claimed: assertions 240×, UVM bring-up 400×7
  • Cognichip $93M for a "physics-informed foundation model"8
  • ChipStack's exit to Cadence: the first AI-DV acquisition4

And the chipmakers themselves

  • All three vendors' agents run on NVIDIA Nemotron models3
  • AMD rolled Claude out across its engineering org9
  • Anthropic contributes to next-gen NVIDIA chip design10
All "×" and "L" figures are vendor claims · press C for sources
Act III · State of the art · the honest tension

Hold both of these in your head

The marketing

  • "L5: fully autonomous virtual design engineer"
  • "40× faster RTL validation"
  • "400× faster UVM environment bring-up"

The measurement

  • Hardest public benchmark (CVDP): best model ≈ 34%
  • Agentic task categories score lower, not higher
  • Formal reasoning: still a frontier-model weakness

Both are real. The space between them is called the trustworthiness gap,
and it is where hardware engineers will spend the next decade.

"L5" borrows the self-driving autonomy ladder: L1 assistance → L5 fully autonomous, no human in the loop. Synopsys and Cadence grade their EDA agents on it · press C for sources
Act IV · Using it today

The anatomy of your setup

model: the brain frontier API · or open weights on your desk tools: the hands simulators · linters · synthesis · waveforms · MCP harness: the loop Claude Code · Codex CLI · Gemini CLI · aider · your own (Agent SDKs, LangGraph…) guardrails: the trust gates the agent cannot skip: lint · regression · coverage · equivalence · review the harness is generic; the tools and guardrails make it a hardware engineer
Act IV · Using it today · tools

Giving the agent hands: CLI first, MCP next

The zero-effort start

Coding agents can already run any shell command. If your flow is scriptable, an agent can drive it today:

$ verilator --lint-only -Wall fifo.sv
$ make sim && gtkwave dump.vcd # agent reads the log
$ yosys -p "synth_ice40" top.v

Plus a CLAUDE.md/AGENTS.md teaching it your flow. That's the whole setup. (Community plugins like GateFlow already wrap Verilator, SymbiYosys, Yosys & nextpnr into Claude Code.)

MCP: a USB-C port for tools

Model Context Protocol: now Linux-Foundation-governed, 10,000+ active servers. EDA is filling in:

  • OpenROAD-MCP: official, from the OpenROAD project itself
  • MCP4EDA: Yosys · Icarus · OpenLane · GTKWave · KLayout behind one interface
  • Vivado & Quartus servers, even driving real dev boards
  • Still missing: cocotb, JTAG/OpenOCD, a proper waveform server

Rule of thumb: start with shell + Makefiles; reach for MCP when you need structure, state, or safety.

Act IV · Using it today · models

Closed vs. open: an actual choice now

Closed frontierOpen weights
WhoClaude · GPT/Codex · GeminiDeepSeek (MIT) · Qwen3-Coder (Apache) · GLM · Kimi · gpt-oss · HDL fine-tunes (CodeV-R1…)
StrengthBest agentic reliability: long tool-use chains, self-correctionYour data never leaves the building · fine-tunable on your RTL
Cost modelPer token: a heavy agent day can cost real moneyBuy the GPU once, iterate for electricity
Verilog todaySaturates classic benchmarks; ~⅓ of realistic CVDP tasks30B-class models are genuinely usable in a good loop; 7B fine-tunes rival 671B generalists on RTL generation

For chip companies the deciding factor is usually not quality: it's IP confidentiality.
That's why open models matter more in hardware than almost anywhere else.

Act IV · Using it today · local inference

"Local" stopped being a compromise

  • A 24–32 GB GPU runs a quantized 30B-class coder at interactive speeds
  • A $2–5k, 128 GB unified-memory box (DGX Spark, Strix Halo, Mac) runs 100B+ MoE coders on a desk
  • The 2026 sweet spot: MoE models with tiny active params: 80B-total / 3B-active coders hit ~70% SWE-bench, fully offline
  • Ollama / LM Studio / vLLM make it a one-command install

Everything in Act V, every measured run, used local models on hardware like this. Not a frontier API in sight for most of it.

Why this matters here

Open models + open EDA + an open PDK =

a fully sovereign, fully inspectable, zero-license path from prompt to GDSII.

No cloud dependency. No license server. Nothing you can't read the source of. A university, or a student, can own the entire stack.

Act IV · Using it today · the open toolchain

You are sitting in the perfect playground

Simulate & verify

Verilator (functional coverage since 5.050) · GHDL · cocotb 2.0 (Python testbenches, the language LLMs know best) · SymbiYosys · eqy

Synthesize & harden

Yosys · OpenROAD (ships an official MCP server) · LibreLane: this week's workshop flow, IHP PDKs built in → GDSII

Real silicon

Tiny Tapeout on three foundries · IHP 130nm CMOS5L Open Source PDK: low-cost open-source MPW shuttles · and this school's funded tapeout

Every box is free, scriptable, and license-unlimited: an agent can hammer this loop a thousand times a night, which is exactly what commercial license seats forbid. Open EDA isn't just the cheap option; for agents it's the better option.

Act IV · Using it today · craft

Patterns that work, failure modes that bite

Do

  • generate → lint → simulate → fix: never accept unexecuted RTL
  • Testbench first, or at least independently, to decorrelate the checker from the author
  • Make the agent prove refactors: equivalence check, not eyeball
  • Feed it windows of waveforms, coverage holes, log excerpts: query, don't dump

Beware

  • Protocol off-by-ones, FSM corner states, reset & CDC subtleties: plausible ≠ correct
  • Vacuous passes: a testbench that checks nothing, green forever
  • Reward hacking: agents will delete the failing test, weaken the assertion, shrink the timeout
  • Naive wrappers can hurt: a bad loop can score below one-shot prompting

The failure modes are not hypothetical. Act V shows you every one of them, measured.

Act V · Two working systems

I built both halves of the problem, and the eyes

greenfield

RTL Forge

Natural language → verified SystemVerilog.
A 13-stage pipeline with fix loops, formal checks, and a deterministic judge.

"produce the RTL"

brownfield

chipchamp

A Claude-Code-style agent platform for existing RTL codebases, driving real EDA tools and earning trust with signed evidence.

"own the loop, and prove it"

companion

alwavelets

A terminal waveform viewer (VCD/FST). Not an agent itself, but built to be driven by one: an MCP server and a live control socket let an agent steer the exact view on your screen and read back what you see.

"the agent's eyes on the waves"

Same thesis, two attack angles plus an instrument: the loop is the product. Everything runs on local, open models.

Act V · RTL Forge

Thirteen stages, half of them checkers

Elicit Spec Architect Generate RTL Review ✓ Lint ✓ Formal BMC ✓SymbiYosys Testbench TB Review ✓ TB Lint ✓ Verify ✓simulate + coverage Judge ✓ re-flow: the failure evidence travels upstream
  • ✓-stages don't write; they check. That ratio is the design philosophy.
  • Formal properties are model-checked, not just emitted.
  • A wrong testbench is worse than no testbench, so it gets its own review pair.
Act V · RTL Forge

Retry is not a strategy. Convergence is engineered

Every patch is classified

accept-progress · accept-equivalent · reject-no-improvement · reject-regression · reject-compile-fail. "Reject" means reject, and the model is told why.

A champion is kept

The best measured state is tracked across the whole run. A bad fix can never overwrite a good state. Stagnation is detected; token-burning stops.

The testbench is attacked

Mutation testing: inject bugs into the passing RTL; a testbench that kills no mutants is exposed as vacuous. "Deleting a test is never a fix" is a hard rule.

Debugging reads the waves

Triage runs a bounded probe loop over the VCD: the model requests signal windows, the harness answers from the dump. Evidence, not vibes.

Act V · RTL Forge · what we measured

The experiment that humbled me

Idea: harvest each model's recurring lint errors → distill rules → inject into its next generation prompt. A/B, 5 designs, real Verilator.

error classno rulesrules injected
unsized literals2/50/5rule helped
parameter headers3/51/5rule helped
block placement3/53/5no change
complex ports0/53/5rule backfired

The backfiring rule was the only one that named the anti-pattern: naming the wrong form primed the model to produce it. The pink elephant effect.

And the control arm swung 37 → 24 errors with no change at all, at temperature 0. Run-to-run noise ≈ 40%. Measure, don't vibe, and know your noise floor.

Act V · RTL Forge · learning

The loop learns between runs, and what it learns is text

Harvest, per model

A training mode runs the pipeline over a spec corpus and harvests each model's recurring failure classes from real lint and simulation. The catalogs are keyed by model: a 24B's habits are not a 70B's habits.

Distill, with guardrails

Failures become positively-phrased rules (the pink-elephant lesson from the last slide is enforced, not just remembered). Embedding-based dedup detects saturation: when new runs only paraphrase known rules, harvesting stops itself.

Remember what fixed what

Cross-run memory keys fix recipes by failure signature: the next run that hits the same signature sees what worked last time: evidence-carrying reflow across runs, not just within one.

The federated possibility

Rules, recipes and skills are kilobytes of text: no weights, no RTL. Teams could exchange distilled engineering judgment while designs never leave the building: federated learning of rules, not gradients. The artifact is auditable: you can read what your peers taught your agent.

Measured caveat: a too-weak model harvests noise, not lessons: there is a capability floor under which "training" pollutes the catalog. The guardrails exist because we hit this.

Act V · RTL Forge · the receipts

You don't have to take my word for it: demo at the end

1 page
of plain English in: a register block
with interrupt logic
13
stages ran overnight, unattended:
spec, RTL, formal, testbench, judge
50/54
of its own tests pass, re-measured
live on stage with real Verilator

It proved a bug on itself

SymbiYosys found a counterexample: irq lags its enable by one cycle, violating its own property. The simulation failures point at the same defect. Two detectors, one bug.

And the judge says FAIL

Final verdict: 77/100, FAIL. Weighted criteria, formal counterexample charged, per-requirement traceability. The system does not bluff. That honesty is the demo.

Deterministic on purpose: the demo re-verifies finished artifacts. No model call, no network.

Act V · chipchamp

Claude Code, if it had grown up in a fab

  • An agentic CLI over an existing RTL codebase: it queries an elaborated design database (hierarchy, clock domains, FSMs, fan-in cones), not a pile of files
  • 17 EDA adapters as first-class tools: Verilator, GHDL, cocotb, Yosys, SymbiYosys, eqy, OpenSTA, LibreLane, nextpnr, Vivado, FABulous, spike…
  • 96 tools, disclosed progressively: the model sees a 31-tool core, 72% less context per request
  • Waveforms & coverage are queried, never dumped: wave.when, cov.holes, first-divergence compare
# any provider, including fully local
$ chipchamp --provider ollama
╭──────────────────────────────╮
│ ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓ │
│ ▓▓ chipchamp ▓▓▓▓▓▓▓▓▓▓▓▓▓ │
│ ▓▓ agentic RTL & DV ▓▓▓▓▓▓ │
│ ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓ 2631 ▓ │
╰──────────────────────────────╯
design DB · 176 modules · 6 s

The banner is an ASCII die shot, lot-code stamped with the git revision.

Act V · chipchamp · the core idea

Success is structurally unforgeable

agentwants to say "done" real tool runssim · lint · synth · formal signed job recordsthe model cannot write these policy gateslint · regression · coverage · LEC …+ anti-gaming detectors "trust me, it passes": this path does not exist report.done is the only way to claim success: it re-validates every gate against the records

Anti-gaming gates: green-by-deletion, disabled assertions, shrunk timeouts, pinned seeds: all detected, all block completion

Evidence bundles are hash-signed and tamper-evident, auditable long after the session

Subagents physically lack the "declare success" tool: only the orchestrator can, and only through the gates

Act V · chipchamp · how deep the loop goes

Prompt to GDSII, measured

RTL → GDSII LibreLane · SKY130

  • Full physical signoff as agent gates: DRC 0 · LVS 0 · antenna 0 · WNS +1.64 ns
  • 213 KB of GDSII out
  • Physical feedback to RTL: area by source line, power from real activity, critical path → file:line

Optimize, with proof eqy

  • Coverage finds 28 dead flops; the agent narrows the logic
  • Equivalence checker proves the edit safe
  • 400 µm² recovered; the whole loop takes ~5 seconds

FPGA & eFPGA nextpnr · Vivado · FABulous

  • iCE40 bitstream in ~2 s; Vivado xc7a35t in ~2 min, timing met
  • Drives FABulous eFPGA fabrics, the technology at the center of this school's tapeout

RISC-V spike · RISCOF

  • Lock-step co-simulation against the reference model
  • "divergence at instruction #2: core executed PC 0x1040, reference 0x1008 (add gp, ra, sp)"
  • A named instruction, not "test 7 fails"

Stress-tested overnight, on local models: 9 models × 8 scenarios, 11 hours unattended, 27/30 platform checks passing.

Act V · chipchamp · versus a general coding agent

What Claude Code structurally cannot do here

a general cloud agentchipchamp
needs a cloud API, so NDA'd RTL and PDKs leave the buildingair-gapped: free local models close real silicon flows; PDK reads are refused below the model
"done" is as good as the model's discipline that dayevidence-gated closure: report.done is rejected until class-specific gates are paid with post-edit proof
bash runs are ephemeral; the evidence trail is whatever got narratedprovenance ledger: every job has an id, repro command, input hashes (incl. fabric state), greppable logs
one model, fixed for the sessionmid-task routing: hot-swap to the next local model on refusal / timeout / stall, same transcript

Live proof of row 2: a local model edited its design after its last proof and tried to close: rejected. It re-proved with a fresh bitstream, then closed. The harness enforced honesty; the model learned it.

Act V · chipchamp · which local models, and who drives

Local models have shapes; the router knows them

modelshape
qwen3.6:35b-a3bthe closer: authored + debugged + closed every eFPGA demo; best debugger
gpt-oss-120bcleanest RTL author; weak self-diagnosis, can claim unearned wins
nemotron-nano-4bfast, reuses IP, decomposes; RTL rarely lints clean first try
gpt-oss-20bfast but chronically refuses mid-task

Profiles score each model per capability (author / debug / plan / verify / speed), seeded from measured campaigns, not vibes.

$ chipchamp /router on
# mid-task, the current model stalls…
model_switch gpt-oss-20b → qwen3.6:35b
reason: refusal · transcript preserved
task closed by the fallback

A 3-model relay on one transcript is validated end-to-end: refusal → swap → stall → swap → close.

Act V · chipchamp · packaged expertise

Skills teach the loop; the library feeds it

Skills SKILL.md playbooks

Expert procedure as versioned markdown. Invoked manually (/waveform-debug, /alwavelets-viewer-control) or auto-matched from the task. The skill index rides the system prompt; the playbook loads on demand.

$ chipchamp skills add ./team-skills
$ chipchamp skills list
waveform-debug · alwavelets-viewer-control

IP library my private library · 134 components

Manifest-indexed component libraries with name-independent semantic match: the agent asks for "async FIFO with gray-code pointers", not a filename, then fetches RTL + C++ golden ref together.

$ chipchamp library add ~/rtllib-pillars
$ chipchamp library match "async fifo"
$ chipchamp library fetch cdc_fifo

Both are how a team's know-how outlives the session: the skill is the how, the library is the what, and the agent reuses both instead of reinventing them.

Act V · chipchamp · what is implemented today

The agentic loop, capability by capability

Progressive disclosure

31-tool core, groups load on demand: 72% less context. Mis-named tools get did-you-mean, not "unknown"

Evidence gates

report.done re-validates class-specific gates against signed job records; anti-gaming detectors block green-by-deletion

Context compaction

old tool results evict to their job ids under a live budget, so long runs survive small local context windows

Teaching errors

every miss names the fix: id shape + recent jobs, path did-you-mean, "not available to your role" vs unknown

Model routing

capability profiles per local model; hot-swap on refusal / stall / timeout, transcript preserved

Subagent orchestration

least-privilege roles, shared job store; only the orchestrator can claim success, and only through the gates

Budgets & policy

token / CPU / license ledgers the model cannot talk its way past; PDK reads refused below the model

MCP, both directions

consumes external MCP servers as typed tools (live waveform viewer control); serves its own design DB over MCP

Skills

SKILL.md playbooks, auto-matched or /invoked. Shipped: waveform-debug, viewer control; teams add their own

Plus: session resume · experiment recorder for blind model campaigns · prompt-injection scanning on tool output · reproducible job records with input hashes

Close

What happens to the engineer?

fromto
writing RTL by handwriting specs that survive contact with an agent
running the toolsauditing the verdicts the agent earned
debugging syntaxjudging architecture, constraints, corner cases

Verification taste becomes more valuable, not less.
The bottleneck moves from typing speed to judgment.

Close

Three things to take home

1.  Hardware has what agents need most: cheap, deterministic oracles. It's a leading domain for agentic AI, not a lagging one.

2.  Trust must be engineered, not narrated: evidence the model can't forge, gates it can't lower, tests it can't delete.

3.  Measure, don't vibe. Aggregates hide opposing effects, and your noise floor is higher than you think.

The hackathon starts this week.
Bring an agent.

RTL Forge

github.com/riadhbenabdelhamid/RTLForge

chipchamp

github.com/riadhbenabdelhamid/chipchamp

alwavelets

github.com/riadhbenabdelhamid/alwavelets

riadhbenabdelhamid.github.io/alwavelets

Riadh Ben Abdelhamid

riadh.benabdelhamid@gmail.com

Questions?

00:00

Current slide

Sources for this slide · press C to close

    Controls

    → / spacenext step / slide
    previous
    home / endfirst / last slide
    ntimer + slide title
    csources for this slide
    llight / dark theme
    ffullscreen
    pprint → save as PDF
    ?this help