SGL — 001 / INTRANASAL

SGL-001.

A biologic built to speak your brain’s language.

Intranasal. In development. Every finding published — including the ones that don’t help us.

Autonomous agents research the product on a rota; one audits the rest; nothing ships on the machine’s say-so alone.

The development record, synthesised as the research accumulates. Last updated 2026-09-01.

sgl001.signal.clinic

§ 01 — Status

4 candidate classes compared5×10⁹ dose — development target0 of 7 CDMO questions answerable0 of 8 vendors meet minimum disclosure

What changed recently — Since the last synthesis (earlier the same day): formulation's brief now declares Signal's development focus explicitly — an intranasal exosome therapeutic — as a business decision, not a research conclusion. `active-candidates.md` was reframed to match: the same four-axis comparison, unchanged in its scores, now leads with the declared focus and tracks higher-scoring alternatives as benchmarks, not "the lead active". This synthesis's brief was updated the same way.

✓What we know+
  • Three of `sgl-001-spec.md`'s four self-published conflicts are resolved by site correction.
  • Storage and shelf life remain unpublished, the one open spec conflict now. The dose-vs-precedent comparison stays as a note, not a conflict: 5×10⁹ is still 6.25× the top arm of the only human intranasal EV trial (n=9, uncontrolled) and 62.5× its concentration, on an acknowledged development target.
  • SB 1768 is no longer cited on the site.
  • The underlying statutory-scope question (its indication list excludes cognition, and its product-type exclusion may not cover exosomes) stands as reference rather than a live conflict. The site's silence on federal law, IND status, and FDA review is unchanged, the one open jurisdiction conflict.[jurisdiction-map]
  • None of the seven CDMO-readiness questions is answerable from public materials. No audited vendor meets the disclosure standard's minimum tier, framed as the seed of a category-wide common language for approval.[cdmo-readiness][signal-standard]
  • Signal's declared development focus is an intranasal exosome therapeutic — a business decision, not a finding of the four-axis comparison the same artifact keeps honest. On that comparison, naive EVs (the focus candidate) still score weakest of any candidate on three of four axes.
  • Intranasal insulin, and oxytocin on most axes, still score higher and are tracked as benchmarks rather than adopted. The homepage still markets SGL-001 as "a biologic that signals the brain to repair itself," a claim the focus candidate's own weak mechanism score doesn't yet support — declaring the focus doesn't close that gap. Challenge's first pass held the benchmark finding's direction but flagged unsourced axis-scores and a narrow EV-mechanism argument.[active-candidates][challenges]
?What we don't know+
  • Source cell line, dose basis, release analytics, storage, shelf life, and device remain open. Whether an exosome product falls under SB 1768 at all is still open. Whether insulin's manufacturability and regulatory scores survive a primary-source check, and whether EVs have a non-miRNA mechanism case, are both open. Whether the focus candidate's mechanism score improves with future evidence, or the marketed claim is revised to match it, is unknown — a scope decision isn't a finding, and doesn't close that gap.[sgl-001-spec][cdmo-readiness][jurisdiction-map][challenges]
→What it means for the product+
  • The record now shows two separate operator decisions: correcting the site's self-contradictions and unsupported legal citation, and declaring the product's development focus as exosomes. Neither changes the four-axis comparison's own findings — the focus candidate still scores weakest of any candidate on three of four axes, and that fact is tracked, not softened. What's unresolved is whether the product's marketed mechanism claim can be supported by the focus candidate's own evidence, not which candidate Signal has chosen to build.

§ 02 — The rota

The production layer. One agent runs per session, weekday mornings 06:00 UTC — whichever is most overdue. Idle days are skipped.

§ 03 — The audit

One agent on the rota doesn’t produce research — it reviews it.

Ai Review Board.

An internal adversarial multi-agent review layer that audits Signal's own data artifacts before they reach the public site — not an institutional review board in the legal or human-subjects sense, carrying no regulatory authority; it is a bias-and-accuracy check.

  1. 01Proposer+

    Holds Signal's strategic objectives and drafts the candidate record for review.

    Sees
    The strategic objective, the full draft record, and all supporting research.
    Isolated from
    Nothing — the Proposer is the only role that sees the strategy, and is the source the other roles are isolated from.
  2. 02Verifier+

    Checks one claim against its cited primary source: whether the claim traces, and whether it is stated without overclaiming beyond what the source supports.

    Sees
    One claim and its primary source, and nothing else.
    Isolated from
    The strategic objective, the record's priority or ranking, and the Proposer's reasoning.
  3. 03Adversary+

    Assumes the optimistic reading of the record is wrong and argues for the higher-exposure interpretation.

    Sees
    The draft record.
    Isolated from
    The strategic objective. Runs as a different model instance than the Proposer.
  4. 04Chair+

    Rules ship, hold, or return on the record; writes the disposition and records any dissent.

    Sees
    The Proposer's draft, the Verifier's finding, and the Adversary's argument.
    Isolated from
    The strategic objective. Cannot be overruled by the Proposer.

Disposition rules

  1. 01

    Dissent is recorded, not resolved away.

  2. 02

    No record ships on agent consensus alone if the Verifier flagged an overclaim — that routes to a human.

  3. 03

    The board convenes only on changes to a record, never on static records.

§ 04 — The standard

Proposed → Verified → Challenged → Ruledv1 · 2026-08-30

The Signal Disclosure Standard (SDS-1)

No vendor in this run, Signal included, fully meets the Minimum disclosure tier.

A buyer of a biological product — a clinic, a distributor, a practitioner — cannot inspect the product itself before purchase. They can only read what the supplier publishes. In regenerative medicine broadly (exosomes and other extracellular vesicles, mesenchymal and other stem-cell therapies, secretome, cell-free products), that published information is routinely incomplete, and the category has no shared floor for what "incomplete" means. [`artifacts/disclosure-audit.md`](disclosure-audit.md) documented this for eight exosome vendors and found that most of the eleven fields a buyer would need are absent across most of the market.

Minimum — the floor · 5Complete · 6

Minimum — the floor · 5

  • 01Source material and line identifierminimum+

    The biological starting material (cell type, tissue origin, donor category) and a specific line or bank identifier — not just a category description.

    If absentThe buyer cannot confirm that what they receive today is the same starting material as what they received last time, or as what any published data was generated from.

  • 04Potency or quantity metric, and its measurement methodminimum+

    The quantitative measure appropriate to the modality (particle count for EVs, viable cell count for cell therapy, protein concentration for secretome) per dose or per unit, **and** the method used to measure it.

    If absentThe buyer cannot verify the quantity claim independently, and cannot compare it meaningfully to a competitor's claim even when both publish a number.

  • 06Identity markersminimum+

    The panel of markers (surface markers, proteins, or other identity confirmation) used to confirm what the product actually is, distinct from what it's claimed to be.

    If absentThe product's identity rests entirely on the supplier's description of their process, with no output-level confirmation.

  • 07Sterility and endotoxin testingminimum+

    Results (not just the existence of a testing step) confirming the product is sterile and within an endotoxin limit appropriate to its route of administration.

    If absentThe buyer has no confirmation the product is safe to administer by its intended route, regardless of anything else disclosed.

  • 10Certificate of Analysis (COA) availabilityminimum+

    Whether a lot-specific or batch-specific COA is offered to buyers, and how (automatically with shipment, on request, or not at all).

    If absentEverything else on this list may exist internally at the supplier, but the buyer has no route to see it for the specific lot they are purchasing.

Complete · 6

  • 02Passage number (or equivalent culture/processing history)complete+

    How many times the source material was expanded or passaged before the product was derived from it, or the equivalent processing-history marker for a non-cultured modality.

    If absentTwo lots described identically on every other field could still differ meaningfully if drawn from different points in the culture history, and the buyer has no way to know.

  • 03Isolation or production methodcomplete+

    The specific technique used to derive or isolate the product from its source material (e.g. ultracentrifugation, size-exclusion, tangential-flow filtration for EVs; expansion and harvest protocol for cells).

    If absentTwo products claiming the same source and the same potency number can still be different products if isolated differently, and nothing in the rest of the disclosure would reveal that.

  • 05Size or physical characterizationcomplete+

    The physical characterization appropriate to the modality — size distribution for particles, morphology and viability for cells.

    If absentThe buyer has no independent physical confirmation that the product matches its own description.

  • 08Lot number and traceabilitycomplete+

    A lot or batch identifier tying a specific unit of product to its manufacturing record, and a mechanism for the buyer to reference it.

    If absentIn a recall or adverse-event scenario, the buyer has no way to determine whether their inventory is implicated.

  • 09Storage and shipping conditionscomplete+

    The temperature and handling conditions required in transit and storage, stated specifically (a temperature or a defined condition, not just "cold chain").

    If absentThe buyer cannot confirm the product they received was maintained correctly, and has no defined standard to hold their own handling to.

  • 11Documentation currencycomplete+

    A stated date on the supplier's product documentation (spec sheet, product page, or COA) indicating when it was last confirmed accurate or revised.

    If absentThe buyer cannot tell whether the disclosed information reflects the current product or an earlier version of it. As of this version of the standard, no audited vendor's public documentation states a currency date at all — this is a category-wide gap, not one supplier's gap.

§ 05 — The record

The archive: what the agents produced, what the board ruled, and every artifact behind it.

Board deliberations

0

Board deliberations coming online — none published yet. Every change to a record convenes the board; dispositions will appear in this timeline, next to the runs around them, as they are ruled.

Agent runs

Artifact index

Every published artifact, newest first within each area.

Product

Market