shuxiachai/academic-commercialization-agent

Turn a paper or research topic into a source-linked commercialization assessment draft, with an auditable scorecard and risk notes for research triage.

Open the application · Read a public report · Run locally · 中文

Demo access: an operator-issued access code or BYOK (a supported LLM API key plus a Serper search key). New analyses and PDF extraction incur provider usage. The public report below can be read without keys or a new analysis.

Real UI archived on 2026-08-11 at a343a95, not a current-release capture.

Public sample: Solid-state batteries for electric vehicles — frozen 2026-07-19 version, 9673a83. The screenshot and sample are different historical runs; neither represents current model accuracy or an up-to-date market assessment.

Start here: documentation and repository map · Case study

Built with Python, CrewAI, FastAPI and a build-free JavaScript client. The production system deliberately limits autonomy: retrieval is deterministic, and six LLM stages reason over validated evidence. Supplementary Tool Calling is experimental and remains disconnected from production.

Market scorecards disclose unverified estimate comparability. This makes the historical cap's limitation visible; it does not fix or validate the underlying market-scoring policy. New validated results also show pre/post cap scores and the actual deduction; historical missing receipts cannot be reconstructed. The 20-source inspection still establishes no fully comparable real pair; the scoring policy is not yet validated.

• Submit a research topic or attach a paper PDF; choose report language and scoring profile.
• Optionally provide Decision Context: the system distinguishes exploratory orientation from an actor-specific decision and identifies whether a threshold is owner-approved.
• Follow progress, inspect citations and reliability warnings, export Markdown or PDF, and share a run link.
• In Sources, locate a saved source by exact ID or literal keyword and expand its saved text without a model call. Saved text is not paper full text or verified claim support; see the operating guide.
• Recover an interrupted run as an immutable child using its longest validated checkpoint prefix and fresh credentials.
• Use an operator-issued access code or supported bring-your-own-key (BYOK) credentials on a gated deployment.

A report supports research triage; it is not technical, legal, regulatory, investment or freedom-to-operate due diligence. Valid citation IDs do not establish that a source entails a claim.

Topic / PDF + optional Decision Context
                   │
Deterministic retrieval → validation → frozen source registry
                   │
         ┌─────────┼─────────┐
     Academic    Patent    Market
         └─────────┼─────────┘
                 Writer
                   │
                 Reviewer
                   │
                  Scorer → deterministic weighted total
                   │
        Shared run artifacts + terminal truth
                   │
       FastAPI / browser / CLI / recovery

The three evidence specialists run in parallel. Writer, Reviewer and Scorer follow in sequence. A stage is not a promise of exactly one model request. See the code map and contribution rules before changing orchestration.

The frozen baseline contains 10 topics × 3 live repetitions:

These are different checks, not a combined accuracy score. Expected TRL ranges were adjusted after early observations, so this is not independent held-out validation. The uncited-numeric proxy does not measure all hallucinations. Seven of ten topics met their TRL range in all three runs.

Other completed evidence includes:

• 90-cell topology ablation: the four-node arm used 54.89% fewer median tokens and 47.03% lower median cost than the six-node arm. Six nodes were not established as universally necessary.
• Five-reviewer utility study: 20 eligible judgments; the registered success rule failed despite a 6:4 full-workflow preference in each round.
• Two target-user pilot: both retained DEFER and answered MAYBE to reuse; neither checked external sources. This does not establish product adoption.
• Recovery: 30/30 offline fault-injection children completed; one production child reused four committed nodes. This is not an exactly-once or general cost-saving guarantee.
• Runtime RTI02: one normal Qwen completion passed 12/12 primary terminal checks, with a disclosed minor observer-cadence deviation. Timeout/fallback lanes and general report quality were not validated by that run.

Protocols, source artifacts and limitations are linked in the current evidence ledger.

Three capabilities have different release boundaries:

• Saved-source location: an opt-in production page/API makes at most one bounded Qwen selection from saved IDs/titles, then returns exact saved text. It generates no research answer and performs no new search. Owner-code authorization, explicit transfer consent, paid admission, durable receipts and operation-level accounting apply. See the operating contract.
• Supplementary retrieval: production remains zero-call shadow mode. The latest unseen evaluation failed its admission gates; source-location progress does not enable this separate capability.
• Generated evidence follow-up answers: experimental, not a production answering feature. Tool execution or valid JSON does not prove claim support.

On 2026-09-26, one bounded production-browser acceptance selected the prepared source, delivered exact saved text and recovered the same receipt/accounting after execution was closed and the service redeployed. The operation reported 1,199 tokens and a USD 0.000764436 frozen-rate estimate, not an invoice. This is one prepared delivery observation, not unseen accuracy or superiority over free Sources keyword search. Paid execution was closed again with zero feature budgets; it is not permanently enabled. The qualified acceptance record separates the original POST body-observer limitation from the verified DOM and GET response, retaining the earlier origin-rejected attempt as a failure.

Historical hypotheses, failures and closed batches remain in the evidence ledger and experiment archive. They are not combined into a general accuracy score or reopened by this acceptance.

Use Python 3.11 or 3.12 for the CI-tested environment and uv. Dependency installation needs network access; the default test suite does not call providers.

git clone https://github.com/shuxiachai/academic-commercialization-agent.git
cd academic-commercialization-agent
uv sync

Copy .env.example to .env. For Qwen, set these values and replace the placeholders locally; do not commit or share keys:

LLM_PROVIDER=qwen
DASHSCOPE_API_KEY=your-key
QWEN_MODEL=qwen3.5-plus
QWEN_API_BASE=https://dashscope.aliyuncs.com/compatible-mode/v1
TAVILY_API_KEY=your-search-key

Choose the endpoint matching your operator account/region. This project's browser BYOK default uses the China-region endpoint. With multiple LLM keys present, set LLM_PROVIDER explicitly; otherwise auto-selection is DeepSeek → Qwen → Anthropic → OpenAI. Tavily takes precedence over Serper when both search keys are present. Remove unused placeholder keys.

uv run uvicorn api.main:app --reload
# Browser: http://localhost:8000
# Alternative CLI (starts real provider work):
uv run academic_agent --topic "solid-state batteries for electric vehicles"

Real analysis incurs provider usage. Duration is provider-bound: observed Qwen completions include 306 and 885 seconds, not a promised three-minute SLA. See the operating guide for HTTP endpoints, output files, Docker, access codes, BYOK, tracing and recovery.

Run URLs carry 128 bits of randomness and act as read capabilities: anyone with the full URL can read that run until retention removes it. Code-owned mutation additionally requires its owner/admin code; ownerless BYOK runs have no second server-side identity. Do not publish private run URLs.

Before public deployment, configure access control, paid-operation limits, retention and persistent storage. PDF extraction is a paid operation too. Use one application replica / one Uvicorn worker: in-memory ownership and file-backed quotas are not a distributed queue. See deployment controls and checkpoint recovery.

The pre-consolidation baseline at 0fdaa76 passed 2071 tests and 678 subtests. CI covers Linux/Windows × Python 3.11/3.12, latest Ruff, narrow Pylint, an 85% coverage floor, zero-provider Chromium smoke, and Docker. These do not constitute a paid-provider production SLO.

uv run pytest -q
uv run --with ruff ruff check .
# Scheduling preview only; no provider requests:
uv run python benchmark.py --dry-run

New measurements use immutable versioned batches; historical outputs are not overwritten. Reuse requires an explicit matching batch and intact artifacts. Reported costs cover Crew nodes, not helper/PDF/search charges. See the execution and accounting contract.

The benchmark guide explains repetitions, frozen evidence and paid-run precautions. Do not change scoring rules to improve these historical numbers.

First-party paid requests support durable receipts: after a lost reply or refresh, read-only lookup can recover run/PDF/recovery-child acceptance for up to 24 hours without another paid submission. Unknown outcomes remain explicit; this is not provider-level exactly-once or multi-replica scheduling.

• 中文项目说明
• Documentation map — reading paths, repository layout and the boundary between current guides, historical experiments and local/private material.
• Evidence ledger — measurements and what they cannot prove.
• Operating guide — setup, API, deployment and scoring.
• Contributing and AGENTS.md — tested constraints and rejected approaches.
• Experiment archive — unchanged protocols, results and errata; older numbers are historical snapshots.

Open limitations include qualitative source entailment, independently measured decision utility, multilingual/short/non-technical benchmark coverage, legacy metadata read-state clarity, and single-replica scale. More agents, a vector database or Kubernetes would not by themselves resolve these gaps.

Created by shuxiachai. Explore more projects on the author's GitHub profile.

For Chinese documentation, open README.zh-CN.md.