Peptome
peptome· WHITEPAPER
v3.0 · APRIL 2026LIVE
Back
PEPTOME · WHITEPAPER · v3.0 · APRIL 2026

the operating system for human biology.

INTELLIGENCE · EXECUTION · CLOSED LOOP

Peptide biology is the most scientifically consequential and informationally fragmented domain in the life sciences. Over four hundred thousand peptide publications sit in PubMed; thousands of trials are active on ClinicalTrials.gov; tens of thousands of patents live in the USPTO queue. No surface grades this corpus. And none of it connects to the body that is supposed to act on it.

Peptome is a two-layer operating system that closes the gap. The intelligence layer indexes, grades, and grounds the peptide corpus. The execution layer applies that intelligence to one individual at a time. Outcomes loop back as evidence. Grades flow forward as protocol. The architecture is the moat.

§01THE PROBLEM

fragmentation as the default condition.

The corpus is enormous. The grading is absent. The body is missing.

PUBMED PEPTIDE REFS
0+
Peer-reviewed and preprint literature, growing hourly.
ACTIVE TRIALS
0+
ClinicalTrials.gov entries with peptide interventions.
PEPTIDE PATENTS
0+
USPTO queryable surface, order-of-magnitude Q1 2026.

A clinician searching human evidence for a GLP-1 dual agonist must cross-reference at minimum five data sources, each with a distinct query syntax and none with a consistent evidence-grade output. Vendor claims and peer-reviewed RCTs are weighted equivalently in a general web search. Preprints and meta-analyses co-mingle. The evidence-weighting burden falls entirely on the reader.

§02THESIS

build the intelligence layer, not another database.

Four architectural commitments. Each one constrains the surface area of the product and simplifies the reasoning about its economics.

COMMITMENT · 01

Ground every answer in source.

No synthesized claim is published without explicit citation. Every card, every Agent response, every protocol proposal carries a chain of reference back to primary literature or to the flagged absence thereof.

COMMITMENT · 02

Grade every source.

No citation is treated equivalently to another. Evidence is ranked by an algorithmic rubric that separates human RCT meta-analyses from rodent case studies from unverified vendor claims. The rubric is public, versioned, and reproducible.

COMMITMENT · 03

Expose one surface twice.

The human application and the agent API are two views on the same grounded knowledge graph. A person on the web and an autonomous agent on HTTP receive the same graded, cited answer. One system of record. Two consumption modes.

COMMITMENT · 04

Price access in the unit of the consumer.

Humans subscribe in dollars. Agents pay per call in stablecoins. The payment primitive adapts to the caller. The underlying intelligence does not.

§03ARCHITECTURE

intelligence above, execution below.

Two layers, one evidence graph. Outcomes from below feed the graph above. Intelligence above routes back into personalization below.

● LAYER 01 · INTELLIGENCE

what is known.

EVIDENCE ENGINECARD GENERATORAGENT COPILOTDISCOVERY LAB
hybrid retrieval · graded corpus · per-claim citation · versioned rubric
EVIDENCE GRAPH · THE SEAM · GRADED · CITED · VERSIONED
● LAYER 02 · EXECUTION

what you do about it.

HOMEAGENTSTACKOUTCOMESBIOLOGY
daily protocol · adaptation engine · outcome capture · vendor verification
§04EVIDENCE MODEL

every answer cites. every claim grades.

The rubric is algorithmic, versioned, and public. It is the platform's most editorially consequential primitive — the single surface on which Peptome's reputation is staked.

RUBRIC
v1.2
6 TIERS
A+
META-ANALYSIS
A
HUMAN RCT
B
PEER REVIEW
C
CASE SERIES
D
ANIMAL
P
PLATFORM OUTCOME
● ABSENCE IS A FIRST-CLASS SIGNAL

the agent never synthesizes upward beyond the graph’s support.

If the strongest human evidence for a claim is Grade C, the Agent returns a Grade-C-qualified answer. It does not assemble Grade C data into a Grade B conclusion. Claims refuted by later evidence are explicitly flagged. Claims reinforced are noted in the card’s update history. Every card carries a freshness timestamp and a changelog.

P-TIER · PLATFORM OUTCOME · BELOW CLINICAL · ABOVE ANECDOTE · ADMITTED AFTER n≥50 COMPLETED PROTOCOLS
§05THE EXECUTION LAYER

five screens. one operating loop.

Decide. Ask. Commit. Observe. Learn. Nothing in the surface is decorative.

SCREEN · 01
Home

Today's protocol — what to take, dose, route, timing. Adherence streak. Next outcome check.

SCREEN · 02
Agent

Conversational Agent grounded in your stack and the evidence graph. Every answer cites. Every refusal explains.

SCREEN · 03
Stack

Active peptides and supporting compounds. Interactions checked against vendor verification and the graph.

SCREEN · 04
Outcomes

Structured logs: sleep, energy, libido, mood, biomarkers, side effects. Every entry timestamped and linked to dose.

SCREEN · 05
Biology

Personal evidence layer — bloodwork, hormones, body composition, training load. The substrate the Agent reasons over.

● ADAPTATION ENGINE

the protocol moves with the person.

INPUTS
  • Outcome logs · 14d window
  • Adherence rate
  • Biomarker deltas
  • Oura / training load
RULES
  • Evidence-cited rule
  • Stop-condition first
  • Explicit threshold
  • Cited tier on every rule
OUTPUTS
  • Proposed change + diff
  • Expected effect window
  • Evidence chip
  • Explicit decline path
● GUARDRAILS · NON-NEGOTIABLE
  • NO · 01No prescribing — drafts gated behind a clinical pathway.
  • NO · 02No ungraded claims — tier chip on every rendered statement.
  • NO · 03No silent changes — diff, reasoning, and decline path always visible.
  • NO · 04No scheduled-substance workflows — region-aware gate, default hidden.
  • NO · 05No vendor endorsement without third-party CoA verification.
  • NO · 06No cohort inclusion without opt-in consent per stack, revocable.
§06THE CLOSED LOOP

the moat is a circuit, not a wall.

A research API on its own is a content product. A consumer Agent on its own is a wellness app. Together — and only together — they form a circuit in which every individual outcome makes the underlying intelligence more specific.

EVIDENCE
GRAPH
AGENT
PROTOCOL
OUTCOME
EVENT
P-TIER
CLAIM
● OUTCOME EVENT SCHEMA · v1

Every outcome captured in the Agent is normalized into a narrow event schema before it is admitted to the cohort layer. Structure beats anecdote. Events outside the schema do not feed the cohort.

event_id · user_pseudonym · protocol_id · compound · route ·
administered_at · day_of_protocol · outcome_metric · value ·
side_effect · adherence_window · consent_cohort
§07$PEPT · ECOSYSTEM TOKEN

the token at the core of the ecosystem.

$PEPT powers the Peptome subscription and will share in protocol revenue. Terms to be announced.

$PEPT
ECOSYSTEM
— FIXED SUPPLY
0
ONE BILLION $PEPT · IMMUTABLE
LAUNCHFAIRLAUNCH
MODELSUBSCRIPTION
REVENUETBA
§08ROADMAP

directional, not committed.

Specific dates and features are subject to revision. Canonical values live in the platform's live documentation.

Q2 2026

Human interface v1.

Search, Card Generator, Protocol Agent, Discovery Lab alpha. Closed cohort.

Q3 2026

Institutional onboarding.

Agent-native API to external operators. Medical advisor governance council seated.

Q4 2026

Validator network + P-tier opens.

External operators post stake. Outcome-submission endpoint opens. Discovery Lab admits verified researchers.

Q1 2027

$PEPT TGE.

Fairlaunch, audits published, governance handoff to on-chain voting. Preconditions in the architecture section must be met first.

BEYOND

Federated graph.

Peptome as a read surface inside third-party clinical research tools, and federated graph integrations with academic consortia.

§09CONCLUSION

a reference implementation for the decade ahead.

Scientific information consumption is migrating from human readers to autonomous agents. Evidence grading is becoming infrastructure rather than editorial work. And personal biology is finally structured enough to close the loop between what’s known and what an individual actually does.

Peptome is the reference implementation of that intersection. The platform’s ambition is not to own peptide biology. It is to be the surface peptide biology, in its next phase, is reasoned over from.

PEPTOME · v3.0 · INTELLIGENCE · EXECUTION · CLOSED LOOP