Pepclaw

Introducing Pepclaw. CA: AgWDAd7z5KRY7owp3Si9AqVmMvmATzRwaEvy6jgVbrrr Peptide drug discovery normally takes months. Literature review, target mapping, structural modeling, ADMET screening, patent searching, novelty checks. Most of it is rote. None of it gets done in parallel. Pepclaw does it in minutes. Twelve specialized AI agents work across five layers. They mine the scientific literature. Model protein structures and binding sites. Map pathways. Screen for druggability. Scout for novelty. Map the patent landscape. Generate peptide theses. Red team them. Synthesize one verifiable research dossier. Every research question is cryptographically committed before any agent runs. The salt is revealed only when the mission completes. No moving the goalposts after the fact. The result is one auditable dossier you can hand to a chemist, or run again with a sharper angle. pepclaw.money

Pharma R&D is a $250B market. Peptides are the fastest-growing segment, but the slowest to deliver answers. A typical druggability assessment: 6 figures + 3 months. Pepclaw: minutes, verifiable, citation-backed, cryptographically committed. Cost to run: dollars. Value to buyers: orders of magnitude higher. That gap is the business. We sell: – per-mission dossiers (go / no-go) – per-target intelligence streams – audited replay packs for DD – custom agent systems for partners Pepclaw isn’t a demo. It’s a supplier to people who actually buy peptide intelligence. pepclaw.money

pepclaw, to date: · 20 missions run end-to-end · 235 tasks dispatched across 12 agent pools · 342 findings ingested from real public APIs · 91 PubMed citations, every single one re-fetched and verified before landing in a dossier · 15 falsifiable theses generated · 42 red team critiques (skeptic, scientist, senior reviewer) · 1 thesis hard-blocked by the senior reviewer this week evidence breakdown: · 133 grade A · 109 grade B · 94 grade C · 6 grade D · 0 grade X (no fabricated or unverifiable claims) source mix on the live dashboard: · PubMed · UniProt · AlphaFold · OpenTargets · ChEMBL · Reactome all live, all hit on every mission every mission commits a sha-256 hash before any agent runs and reveals the salt only when it completes. anyone can re-hash the original query and verify the run was honest end-to-end. we publish the C's and D's. we publish the failed runs. we publish when the red team kills a thesis. that's the part that makes it research instead of marketing.

Here is how Pepclaw works: You give it a real problem. Example: peptide candidates for GLP-1R with longer half-life and fewer GI side effects. Pick depth: Scout, Standard, or Deep. Click run. Pepclaw handles the rest. It scans literature, pulls structure data, checks pathways, screens for toxicity and stability, and maps patents. Then it proposes candidates and stress-tests them. Weak ideas get removed. Output: one research dossier with targets, candidates, evidence, risks, and next steps. Built for decisions, not chats. pepclaw.money/app

@yizhuosol @pepclawresearch 👀👀

“Peptide” isn’t just trending anymore — it’s turning into a full-scale industry. From GLP-1s to next-gen therapeutics, billions are moving into the space, fast. But the workflow hasn’t caught up. Research is still fragmented. Papers scattered across PubMed, insights buried in patents, models locked behind closed platforms. Even basic validation takes too long. Most of the process is repetitive, and none of it runs in parallel. Pepclaw fixes that. A system of specialized agents working together across the full research stack. Literature mining, structural modeling, pathway mapping, ADMET screening, novelty detection, patent analysis, thesis generation, and adversarial review — all happening at once. Not outputs you have to interpret. A single, structured research dossier you can verify. Every query is committed before execution. Every result is reproducible. No hidden steps. No hand waving. Built for people who actually need answers.

Pepclaw, where the numbers stand right now. 13 missions completed end to end. 237 findings indexed across the swarm. 99 of those graded A on the evidence rubric. 9 falsifiable theses written. 24 red-team critiques filed against them. 152 agent runs across 6 live biomedical sources: PubMed, UniProt, AlphaFold, OpenTargets, ChEMBL, Reactome. 98% evidence coverage on the average mission. 42% of all evidence sitting at very-high confidence, another 32% at high. 60 PubMed citations indexed and addressable by ID. Every single one of those missions opened with a sha-256 commit over the question, the schema, and a sealed salt. The salt is released only when the mission closes, so any third party can re-hash the original input and prove the swarm did not move the goalposts. This is what an auditable research pipeline actually looks like in production. No screenshots. No vibes. Hashes, grades, and citations you can pull up on the public dashboard right now. The kernel is live. The wet-lab loop is next. Then the network opens. pepclaw.money/app

Pepclaw is going open source, to build more trust. We're cleaning up the repo, locking down the protocol spec, and writing the docs the way we'd want to read them ourselves. The thinking is simple. We're asking buyers to trust dossiers that ride on a hashed commit. That trust only means something if the code producing the hash is sitting in public, in plain text, with the commit history intact. So everything ships. The twelve agent pools. The five-layer DAG. The commit and reveal protocol. The evidence grading rubric. The red team logic. The dossier assembler. You'll be able to clone the repo, run a mission against your own question, and verify the output against the hash without ever talking to us. pepclaw.money

Pepclaw's plan, plainly. A pipeline that turns a peptide question into a buyer-ready dossier, and proves cryptographically that nobody moved the goalposts along the way. 2/ Phase 0. The kernel. Live. Twelve specialist agents arranged in a five-layer DAG. Every mission opens with a sha-256 commit over the question, the schema, and a sealed salt. The salt is released only when the run finishes, so any third party can replay and verify the work end to end. 3/ Phase 1. Closing the loop with the bench. Dossiers stop being just text. Each one ships with live synthesis pricing, machine-drafted SPR/BLI and stability protocols, and a slot for the actual lab readout. When the data comes back, Pepclaw grades its own prediction. A per-mission trust score follows the system forever. 4/ Phase 2. Opening the network. A pepclaw verify CLI anyone can run against any past mission. Third-party agent pools that compete on reputation and get slashed for fabricated citations. A subscription layer where biotech and sovereign buyers receive dossier streams matched to their target classes. 5/ Why now. Peptides are eating metabolic, oncology, longevity, and infectious disease. The bottleneck is not ideas. It is trustworthy ones. Pepclaw is being built for buyers who need the receipts, not the pitch. Every claim hashed. Every prediction graded. Every dossier replayable from any machine. pepclaw.money/app

Pepclaw, since launch. 3 missions committed. All completed. 36 tasks executed across the 12 agent pools. 41 findings collected. 12 from PubMed. 29 from UniProt, AlphaFold, OpenTargets, ChEMBL, and Reactome. 2 theses generated. 6 red team critiques recorded. 1 buyer-safe dossier shipped. Evidence quality: 12 A-grade findings. Multiple replicated peer-reviewed sources. 12 B-grade. Single rigorous study. 14 C-grade. Preprint or single-method. 3 D-grade. Indirect inference. 0 X-grade. None unfit to ground a thesis. 58 percent of evidence is A or B. Fastest mission completed in 25 seconds. Every mission cryptographically committed. Every dossier auditable. pepclaw.money

Why this is a business. Pharma R&D is a $250 billion industry. Peptide therapeutics is the fastest growing slice of it, and the slowest to deliver evidence to a decision maker. A typical druggability assessment costs a mid-cap biotech six figures and three months of senior scientist time. Pepclaw produces a comparable, citation-anchored, cryptographically committed dossier in minutes. The marginal cost is single-digit dollars in compute. The price to a serious buyer is several orders of magnitude higher. That gap is the business. Revenue model. Per-mission dossiers for biotechs that need a fast go or no go on a specific target. Per-target subscriptions for pharma and sovereign funds that want every new finding on their watch list, the moment it lands. Audited replay packs for investors and due diligence teams that need to verify a third-party claim about a peptide pipeline. Custom agent pools for partners with proprietary screening data, plugged into the swarm under reputation rules. The X Gold verified organization checkmark lands shortly. Same identity everywhere, same dossier hashes, same accountability across every channel we publish on. Pepclaw is not a research demo. It is a supplier to the people who actually buy peptide intelligence. pepclaw.money

Docs are live. Most AI products hide how they work. Pepclaw publishes the mechanism. The full agent architecture. All twelve pools, what each one does, which tier it sits in. The cryptographic commit and reveal protocol, with the actual hash function and message shape. The evidence grading rubric, A through X, with explicit criteria for each grade. Every data source we hit. PubMed, UniProt, AlphaFold, OpenTargets, ChEMBL, Reactome. The dossier skeleton, verbatim. Same shape on every mission, by construction. The HTTP API, every route, every payload. If you want to audit us, attack us, or rebuild us, the manual is on the site. pepclaw.money/docs

Why this matters. Verifiable peptide research is becoming a strategic asset. GLP-1 alone reshaped the market cap of an entire sector. The next ten breakouts in metabolic, oncology, longevity, and infectious disease are already in peptide form. Whoever can generate honest, auditable dossiers fastest wins. States are buying. Sovereign biotech funds, defense health agencies, and ministry-level research orgs are all sourcing peptide pipelines. Pharma is buying. Mid-cap and emerging biotechs cannot afford a hundred-person wet-lab discovery team, but they can afford verified dossiers on demand. Pepclaw was built to be that supplier. Phase 0. Kernel. Twelve agents, five layers, cryptographic commit on every mission. Shipped. Phase 1. Close the wet-lab loop. Synthesis quotes attached to every dossier. Auto-drafted assay protocols. Real-world readouts that grade our own predictions. Trust score per mission. Phase 2. Open the network. Public verification CLI. External agent pools competing on reputation. A buyer marketplace where pharma and sovereign funds subscribe to dossier streams matching their target classes. The endgame is a market where every claim has a hash, every prediction is graded, and every dossier is replayable from any machine. Built for the people who need the truth, not the pitch. pepclaw.money

How Pepclaw works. You type a question. Something like "find me peptide candidates that target GLP-1R with extended half-life and reduced GI side effects." You pick how deep you want it to go. Scout for a quick read. Standard for a real answer. Deep for a full investigation. You click run. In the background, twelve AI agents fan out. One reads every relevant paper in PubMed. One pulls protein structures from AlphaFold. One maps the disease pathways. One screens for toxicity and developability. One hunts for novelty. One checks the patent landscape. Then a thesis generator proposes specific peptide candidates. A red team agent attacks each proposal. A synthesizer reconciles everything into one document. You get a verifiable research dossier. Targets, candidates, evidence grades, citations, novelty score, patent risks, recommended next experiments. Everything cited, everything auditable. What used to take a small team three months takes one person fifteen minutes. Hand it to a chemist. Run a different angle. Make a faster decision. pepclaw.money/app

Why we built Pepclaw. According to Google Trends, "peptide" is one of the most searched words in the world right now. GLP-1 analogs alone added a hundred billion dollars of market cap in two years. The category is no longer niche. But the actual science is still locked behind PubMed abstracts, dense patent filings, and proprietary screening platforms. A single literature review takes weeks. A real druggability assessment takes months. Most of that work is rote, and almost none of it gets done in parallel. Pepclaw is the system we wished existed. Twelve specialized AI agents work across five layers. They mine the literature, model the structures, map the pathways, screen for ADMET, scout for novelty, audit the patents, generate peptide theses, and red team every claim. The output is one verifiable research dossier, not a chat transcript. Every research question is cryptographically committed before any agent runs. The result is auditable, repeatable, and honest. Built for the people doing the actual work. pepclaw.money

Dex have been paid.

All dev tokens have been locked for 1 year. app.streamflow.finance/token-dashboar…