Free preview
$0 / no signup
Try the workbench before subscribing.
- 3 chats per day per IP
- Light and Normal mode only
- No Heavy mode access
- No Opus-grade depth
- Resets every UTC midnight
KODIAC LAB · RESEARCH USE ONLY · IN PRIVATE BETA
An analytical workbench for investigators working with research compounds.
Peer-level pharmacology depth. Your private literature, indexed and cited. Compound profiles, stack analysis, COA tracking, vendor records, and daily protocol logging in one connected research environment. Built by Kodiac, for the researchers Kodiac was built for.
Existing Kodiac customers get first access. Use the email on your order to skip the queue.
ACCESS
Three tiers, scaled to the work.
Anonymous visitors get a small daily preview. Subscribers get monthly quotas tied to their account and run Claude Opus 4.8. Heavy mode (Claude Opus 4.8 at maximum reasoning effort) is reserved for the Lab and Heavy tiers.
Bench
$10 / month
Entry tier for casual research questions.
- 100 chats per month
- Light and Normal mode
- Normal mode runs Claude Opus 4.8
- No Heavy mode access
- Compound mechanism, PK, sourcing
- Citations to primary literature
Lab
$25 / month
The standard tier for working researchers.
- 500 chats per month
- Light, Normal, and Heavy mode
- Claude Opus 4.8 on Normal and Heavy
- Up to 50 Heavy-mode chats
- Full stack analysis depth
- Max-effort reasoning on Heavy queries
Heavy
$100 / month
For deep multi-compound research programs.
- 2000 chats per month
- Light, Normal, and Heavy mode
- Claude Opus 4.8 on Normal and Heavy
- Up to 200 Heavy-mode chats
- Priority routing during high load
- First access to new modes
Each tier covers one calendar month of access. Quotas reset on the first of each month UTC.Sign in to keep your usage tied to your account.
Three modes calibrated to the question.
LIGHT
Quick consult.
Concise and direct. Skip mechanism deep-dives unless explicitly asked. One to three short paragraphs or a brief structured answer. Quick reference work and fast consults.
MAX OUTPUT · 1500 TOKENS
NORMAL
Default working mode.
Default working mode. Mechanism and reasoning when it informs the answer. Cites compounds, doses, half-lives, and receptor targets when relevant. The everyday research consult.
MAX OUTPUT · 6000 TOKENS
HEAVY
Exhaustive treatment.
Exhaustive. Full mechanistic treatment with receptor subtype specificity, dose-response data, PK and PD considerations, half-life, and stack interactions. Mines non-English literature where relevant: Russian peptide bioregulator work, Chinese pharmacology, Japanese neuroscience. Surfaces obscure and orphan compounds in the same class. Cites primary sources. Treats the user as a peer-level analyst.
MAX OUTPUT · 16000 TOKENS · MAX EFFORT
02 · YOUR LITERATURE, INDEXED
Every answer cited to your library.
Upload your research literature once. PDFs, DOCX, TXT, Markdown, images. The library ingests, parses, chunks, and embeds every file. Hybrid retrieval combines vector similarity with keyword matching across your entire corpus. Every assistant answer cites the exact source document and page so you can verify the claim against the primary source it came from.
03 · ATTACH PER MESSAGE
Drop a file into the conversation.
Per-message uploads via a paperclip in the input bar. PDF, DOCX, TXT, Markdown, PNG, JPG, JPEG. Multi-file selection. Uploaded files appear as chips above the textarea with filename, size, and a remove control. Files are ephemeral by default. Promote any attachment to your persistent library with one click.
04 · IN-CONTEXT CONTROLS
Switch behavior without leaving the conversation.
A sliders-icon dropdown next to the input bar controls per-message behavior. Switch Research Effort modes. Send the previous assistant turn to a structured note. Apply a saved prompt persona for the next message or the rest of the conversation. Toggle library retrieval on or off for the next message.
05 · STRUCTURED NOTES
Capture the answer, not just the conversation.
Send any assistant turn to Notes. The system extracts a title, a 2 to 4 sentence summary, full content, source conversation link, compound tags, and timestamps. Notes are embedded for retrieval, so they surface in chat the same way library files do. Cross-linked to the compounds they discuss.
06 · COMPOUND PROFILES
An auto-generated reference page for every compound in your library.
Every compound that has ever been tagged in your library or notes gets its own profile page. The system synthesizes a summary from every chunk in your corpus tagged with that compound: mechanism, half-life, literature dose window, receptor targets, common references, sourcing notes. Manual notes section for your own annotations. Linked sources, linked notes, sourcing records, and full COA history. Names are normalized so BPC-157, bpc157, and BPC 157 collapse to one entry.
07 · STACK ANALYSIS
Pick two or more compounds. See the interactions.
Select compounds from your library. The system pulls relevant retrieval chunks for each one, runs an interaction analysis across them, and surfaces synergies, redundancies, contraindications, and pharmacokinetic conflicts. Every claim cites the source document the conclusion came from.
08 · COA TRACKING
Upload a COA. Get structured comparison.
Upload a Certificate of Analysis PDF. The system extracts purity percentage, identity confirmation, flags, analyst, date, and batch identifier. Linked to the relevant compound. Builds a COA history per compound so you can compare batches over time. The file itself goes through the standard library ingestion pipeline so it is also retrievable in chat.
09 · VENDOR RECORDS
Source intelligence in one ledger.
Structured records for raw-powder, packaging, equipment, and analytical vendors. Per-vendor metadata: website, contact methods, lead time, payment methods, quality tier. Quote history. Order history with QC outcomes. Attached COA records. Free-text notes. The institutional memory of your supply chain.
10 · DAILY LOGS
Log it once. Query it forever.
Calendar-view daily logs with structured fields: compounds taken with dose, unit, and timing. Subjective notes. Mood and sleep quality. Training load. Free text. Every entry is embedded on save. A natural-language query bar lets you ask questions like "what was my mood the week I added 9-Me-BC" and the system retrieves relevant entries by semantic similarity and date range.
11 · PERSONAS
Save the prompts you keep rewriting.
Saved prompt personas with optional default Research Effort, default library setting, and default compound tags. Apply a persona for a single message or the rest of the conversation. Build the prompts that match the kind of work you do, and stop typing them again.
12 · DIFF VIEW
Two of anything, side by side.
Compare any two of: compounds, stacks, vendor quotes, notes, protocol log entries. Side-by-side rendering with an LLM-generated key-differences summary at the top. Useful when you cannot remember why one stack worked and the other did not.
WHAT IT LOOKS LIKE IN USE
An actual exchange.
HEAVY MODE
YOU
Mechanism and PK comparison of phenylpiracetam vs nefiracetam, with non-English literature where relevant.
KODIAC LAB
Mechanism.
Phenylpiracetam (carphedon) is a phenylated derivative of piracetam with documented activity at NMDA, AMPA, GABA, and nicotinic acetylcholine receptors, plus dopamine and norepinephrine transporter modulation [1]. Nefiracetam acts more selectively on AMPA-mediated long-term potentiation and shows nicotinic acetylcholine receptor sensitization at low doses [2]. Both modulate glutamatergic signaling, but the receptor breadth diverges materially.
Pharmacokinetics.
Phenylpiracetam plasma half-life is reported at 3 to 5 hours with rapid absorption [3]. Nefiracetam shows a longer functional half-life of approximately 5 to 8 hours and slower onset, with cumulative effects reported across multi-week dosing [2].
Receptor targets.
Phenylpiracetam: NMDA, AMPA, GABA-A, nAChR, DAT, NET. Nefiracetam: AMPA (positive allosteric modulation), nAChR sensitization, weaker GABA-A modulation. The phenyl substitution on phenylpiracetam adds the monoamine transporter activity that distinguishes the stimulant-leaning profile.
Notable differences.
Phenylpiracetam reads as a stimulant-adjacent nootropic with measurable acute output: faster reaction time, increased motor activation, modest dopaminergic load. Nefiracetam reads as a slower cognitive-load tool with cumulative working-memory effects and a cleaner side-effect profile in published studies. Tolerance develops faster on phenylpiracetam.
Stack considerations.
Pairing the two is documented in non-English nootropic literature but rarely with rigorous controls. The shared AMPA modulation creates redundancy. The unique-to-phenylpiracetam monoamine activity creates an additive risk if combined with other dopaminergic compounds.
Sources
- [1] phenylpiracetam_carphedon_review_2019.pdf · p.7
- [2] nefiracetam_ampa_ltp_2003.pdf · p.12
- [3] phenylpiracetam_pk_human_2018.pdf · p.4
13 · NOT A GENERAL CHATBOT
Built for the work, not the lowest common denominator.
| General AI | Kodiac Lab | |
|---|---|---|
| Mechanism depth | Surface-level summaries | Receptor subtype, dose-response, PK/PD |
| Citations | Rare or none | Every claim cited to your library |
| Posture | Patient-directed | Investigator-directed (RUO) |
| Library | None | Your private indexed corpus |
| Compound profiles | None | Auto-generated, editable |
| Non-English literature | Rarely surfaced | Mined when relevant |
| Vendor and protocol records | None | Built in |
| Compare tool | None | Side by side with diff summary |
INTERNAL TODAY · PUBLISHED LATER
Every file, note, and compound has a visibility flag: private or published. The platform is built first for private research workflows. A published-research layer is planned for verified Kodiac customers. Your private corpus stays private.
Get early access.
Existing Kodiac customers are prioritized. Use the email on your most recent Kodiac order to skip the queue. New researchers welcome.