How Eluent works

Introduction

Eluent helps users plan, refine, and run peptide research cycles. It is not a marketplace, vendor directory, or medical service. Outputs are educational simulations.

The product is built around one core experience: describe your goal in plain language → get a sourced, scored cycle plan → refine it iteratively in chat → run it with reconstitution math on every dose → log bloodwork over time.

Getting started

1. Open the planner on the home page
2. Describe your goal: "first time GLP-1, 95kg, want fat loss over 16 weeks"
3. The agent searches the protocol library, picks the best fit, adapts to your weight + experience, runs the interactions check, scores the protocol, and returns a cycle
4. Refine it in chat: "lower the GH dose 20%", "swap CJC for Mod GRF", "add a deload at week 6"
5. Click any dose to see the reconstitution math (vial → bac water → syringe units)

Free tier: 3 cycle generations per ISO week. No account required to start. Sign in to save cycles and unlock the rest.

The planner

The planner is a chat interface backed by the Eluent agent (Claude Haiku 4.5 by default with Sonnet 4.5 fallback for deep reasoning).

Workflow

• You describe your goal in plain language
• Agent calls listProtocolTemplates(goal, experience_level) to find candidates
• Picks best fit, calls getProtocolTemplate(slug), adapts to your stated weight + experience
• Calls checkInteractions on the resulting stack
• Calls scoreProtocol — refuses to mark "ready" below 60
• Returns formatted cycle with citations + educational disclaimer

Refinement

Refinements are conversational. The agent updates only what changed and re-scores. Examples:

• "Lower the GH dose 20%"
• "Add a 2-week deload at week 6"
• "Swap CJC for Mod GRF 1-29"
• "I'm getting nausea — should I pause?"
• "What does PubMed say about this combination?"

Protocol library

Eight curated templates covering the most common goals:

Each template is sourced, version-controlled, and contraindication-tagged. Click any in the library to ask the agent to adapt it to your situation.

Peptide catalogue

The catalogue holds 255 peptides across 12 categories: metabolic, recovery, GH-secretagogue, cognitive, longevity, cosmetic, sleep, immune, reproductive, mitochondrial, pain, and other.

Each entry stores:

• Slug, common name, aliases
• Category and receptor targets (e.g. GLP-1R, GHRH-R, GHS-R1a)
• Mechanism summary (1–2 sentence pharmacology)
• Half-life in hours
• Typical dose range (low / high in mcg)
• Typical frequency (daily, weekly, twice-weekly, etc.)
• Administration route (SC, IM, intranasal, oral, topical)
• References (PubMed when available)

Browse the full catalogue on the home page with full-text search and category filters.

Confidence score

Every cycle is scored 0–100 from six positive components and one penalty band. The formula is published, versioned, and never changed silently.

Below 60, the agent will not mark a cycle as "ready". Below 40, it refuses to generate at all and instead explains what's missing.

Reconstitution math

Built into every dose. Given vial size, bac water volume, and target dose, Eluent computes:

concentration_mcg_per_ml  = (vial_size_mg × 1000) ÷ bac_water_ml
volume_ml_per_dose        = target_dose_mcg ÷ concentration_mcg_per_ml
units_to_draw_u100        = volume_ml_per_dose × 100
doses_per_vial            = floor((vial_size_mg × 1000) ÷ target_dose_mcg)

Example

10 mg Tirzepatide vial reconstituted with 3 mL bacteriostatic water, target dose 1.25 mg:

concentration  = (10 × 1000) ÷ 3       = 3,333 mcg/mL
volume_per_dose = 1250 ÷ 3333          = 0.375 mL
units (U-100)   = 0.375 × 100          = 37.5 units
doses_per_vial  = floor(10000 ÷ 1250)  = 8 doses

Try it in the planner: "How many units do I draw from a 10mg Tirzepatide vial reconstituted with 3mL bac water for a 1.25mg dose?"

Bloodwork tracking

Every protocol specifies which markers to monitor. Examples:

• GLP-1 cycles → A1c, lipase, ALT, AST, creatinine, TSH
• GH-secretagogue cycles → IGF-1, fasting glucose, A1c
• BPC-157 protocols → CBC baseline

You log panels over time and the agent reads trends:

Stop thresholds are pre-defined per template; the agent enforces them mid-cycle and offers protocol changes when triggered.

PubMed integration

The agent has a searchPubMed tool that hits the NCBI E-utilities API live. When you ask "what does the literature say about Retatrutide?" or "any human trials for BPC-157?":

1. Calls esearch.fcgi with your query → PMIDs ranked by relevance
2. Calls esummary.fcgi → titles, authors, journal, year
3. Returns clickable links to pubmed.ncbi.nlm.nih.gov/PMID

No mock data. No hallucinated citations. Every paper is real and linkable.

Agent tools

The agent has nine deterministic tools backed by the dataset and PubMed:

Free tier & limits

Free anonymous use is rate-limited to 3 chat calls per ISO week. The limit resets every Monday 00:00 UTC.

The counter is visible above the chat input. When you run out, the upgrade modal opens.

Tiers

Privacy & IP tracking

To enforce the free-tier limit fairly we track usage by IP address. Specifically:

• We compute SHA-256(ip + salt) and store only the first 16 hex characters of the digest
• Raw IP addresses are never written to disk
• The hash + ISO week + count is stored in .usage.json on the server
• Records older than 4 weeks are automatically pruned
• You cannot bypass the limit by signing up multiple times — the limit is per source IP, not per account

If you share an IP with other users (e.g. office, university), you'll all share the 3-per-week limit. Sign in to a paid plan to remove this.

What we store at sign-in

• Email (Google sign-in) or wallet address
• Cycles you generate
• Bloodwork you choose to log
• Standard server logs (truncated IP, user agent, timestamps)

We never sell or share your data. Bloodwork is encrypted at rest. You can export and delete your data at any time.

FAQ

Do I need an account?

No — you can use the planner anonymously up to 3 times per ISO week. Sign in to save cycles, log bloodwork, or upgrade for unlimited.

Why ISO weeks instead of "rolling 7 days"?

Predictable reset (every Monday 00:00 UTC) is easier to communicate and harder to game.

What models does the agent use?

Claude Haiku 4.5 by default for cost efficiency, with Sonnet 4.5 as fallback for complex reasoning. System prompt is cached server-side for ~90% input-token savings.

Where do peptide doses come from?

Published clinical and animal literature. Every range cites the relevant trial or paper when available. Beginner mode uses the lower half of the published range.

Why no vendor recommendations?

Strategic choice. Recommending vendors carries legal exposure (defamation risk, payment-processor friction) and pulls us away from being a research tool. We point users to bloodwork their own purchases instead.

Can I bring my own protocol?

Yes — paste it into the planner. The agent will validate doses, check interactions, run the Confidence Score, and suggest fixes if anything is off.

Disclaimer & legal

By using Eluent you agree:

• You are 18+ and using this for personal research
• You will consult a licensed clinician before any self-experimentation
• We provide no warranty on protocol accuracy
• You release Eluent from liability for self-administered cycles