Vol. 01 — Public launch editionAn autonomous research analystFiled 14.10.2026 Live — reading now

Readseverything.Citesevery line.1

Brief Nº 2291Every line cited

The answer,with receipts.1

Sodium-ion undercuts LFP on cell cost by 20281 and now clears daily grid cycling2. Anode supply caps its 2030 share at 18–24%3.

  1. 1 CATL 10-K · p.43
  2. 2 Nature Energy 9:412
  3. 3 EIA Form 860 · 2026
1 Fig. 0 — one brief, condensed from 4,812 source fragments. Move your cursor to read the ink.

Sable is an autonomous research analyst. Ask it a hard question; it reads thousands of filings, papers and transcripts, and returns a brief where every sentence points to its source.

Scroll to read
§ 01 — ManifestoSable / Launch editionp. 02

Most research is a guess wearing a citation. We skim forty tabs, trust the most confident sentence, and hope nobody asks where it came from. Sable doesn’t skim. It reads every filing, paper, transcript and footnote — then hands back only what it can claim prove.

  1. 012.6MPapers indexed last year. No human team read them all.
  2. 0238Tabs opened per research question, on average. Four get cited.
  3. 0371%Of analyst hours spent finding and checking — not thinking.
§ 02 — How it readsSable / Launch editionp. 03

Four movements.One brief you can defend.

Watch Sable work a real question end to end — from a messy prompt to a two-page brief where every sentence carries its receipt.

01 / 04

Ask.

Ask like you’d ask a sharp colleague.

Type a real question — messy, specific, opinionated. Sable turns it into a research plan you can edit before a single source is read.

  • Plan drafted in ~4 s
  • Editable scope & horizon
  • Private by default
sable / brief 2291 / na-ion-vs-lfpstep 1/4 · planning
Your question

Will sodium-ion displace LFP in grid-scale storage by 2030?

FilingsPapersPatentsEarnings callsTenders2019 → 2026Depth: exhaustive
Research plan · 5 sub-questions · editable
  1. Q1Cost trajectory — cell $/kWh, Na-ion vs LFP, 2019 → 2026
  2. Q2Performance gap — cycle life, density, cold-weather output
  3. Q3Supply chain — hard carbon & cathode precursors
  4. Q4Demand signal — utility tenders & procurement specs
  5. Q5Counter-evidence — who says no, and why
02 / 04

Read.

It reads everything. Actually reads.

Filings, papers, patents, transcripts, tenders, datasets — thousands of them, end to end, footnotes included. Low-provenance sources are discarded, and you can see why.

  • 5,000+ sources per brief
  • 40+ source types
  • Provenance score on each
sable / brief 2291 / na-ion-vs-lfpstep 2/4 · reading
4,812sources read end-to-end
Retrieved
5,140
Discarded
328 · low provenance
Pages
61,207
  • 10-KCATL Annual Report FY2025 — cost of goods, cell segmentp.43
  • PAPERHard-carbon anodes at scale: a techno-economic reviewp.7
  • CALLBYD Q3 2026 earnings call — storage segment Q&A00:41:12
  • PATENTWO2025/118 — Prussian-white cathode synthesis routecl.4
  • TENDERCAISO 2026 RA procurement — technology specifications§3.2
  • PAPERNa₂FePO₄F cathodes: cycle life beyond 6,000 cyclesfig.3
  • DATAEIA Form 860 — planned battery additions, 2026–2030tbl.6
  • NEWSHard-carbon prices ease as new Chinese capacity lands—
  • 10-QNatron Energy Q2 2026 — manufacturing ramp disclosuresp.12
  • PAPERLow-temperature performance of Na-ion vs LFP cellstbl.2
Papers 38%
Filings 22%
Calls 14%
Patents 11%
Tenders 9%
Other 6%
03 / 04

Reason.

Weighs claims, not vibes.

Every claim is built from evidence for and against. Contradictions are surfaced instead of smoothed over — and the ones Sable can’t resolve are flagged for you.

  • Claim ledger, not a summary
  • Contradiction detection
  • Calibrated confidence
sable / brief 2291 / na-ion-vs-lfpstep 3/4 · reasoning
  • C1Na-ion cell cost undercuts LFP by 2028Supported · 0.82
  • C2Cycle life clears daily-cycling grid dutySupported · 0.74
  • C3Hard-carbon supply scales in timeContested · 0.46
  • C4Utilities specify Na-ion in tenders by 2030Weak · 0.31

Contradictions found 17Resolved 15Flagged for you 2

04 / 04

Cite.

Every sentence, accountable.

The brief reads like your best analyst wrote it. Every sentence links to the exact passage, page and source it stands on — click, and you are there.

  • Passage-level citations
  • Export to PDF, Docs, Notion
  • Audit trail for every edit
sable / brief 2291 / na-ion-vs-lfpstep 4/4 · writing
Brief Nº 2291Confidence · moderate-high

Sodium-ion takes a real share of new grid storage by 2030 — but it won’t displace LFP.

Sodium-ion cell costs are on track to undercut LFP by 20281, and cycle life now clears daily-cycling duty2. The binding constraint is hard-carbon anode supply3, which caps a realistic 2030 share at 18–24% of new deployments4.

  1. 1CATL, Annual Report FY2025, p.43 — “cell-level cost, sodium line…”
  2. 2Zhao et al., Nature Energy 9:412 (2025), fig. 3
  3. 3BNEF, Anode Materials Outlook, Aug 2026, §2
  4. 4Sable synthesis of 212 utility tenders, EIA-860 + CAISO RA
§ 03 — ProofSable / Launch editionp. 04

Evidence, not adjectives.

Private beta · 1,200 briefs · 140 research teams

  • 11 minMedian time from question to cited brief
  • 4,812Sources read end-to-end, per brief (median)
  • 98.7%Citation precision, human-audited sample
  • 0Sentences shipped without a source

“It found the footnote that killed our thesis — on page 211 of a filing the whole team swore they’d read.”

Head of Research, Northwind Capital — private beta
  • SEC EDGAR
  • arXiv
  • PubMed
  • USPTO & EPO
  • Earnings calls
  • Court dockets
  • Central-bank minutes
  • Clinical trials
  • Utility tenders
  • Standards bodies
  • Your own PDFs
§ 04 — PricingSable / Launch editionp. 05

Cheaper thanbeing wrong.

Billed yearly — about two months on us.

  • 01

    Reader

    For the curious.

    $0forever
    • 10 briefs a month
    • Up to 300 sources per brief
    • Open web, arXiv & PubMed
    • Markdown export
    Start reading →
  • 03

    Desk

    For research teams.

    $143per seat / month
    • Unlimited briefs
    • Private corpora — your PDFs & data rooms
    • Shared citation library
    • SSO, audit trail, retention controls
    • A named research engineer
    Talk to the desk →

Prices in USD, excl. tax. Education & nonprofit desks: 50% off.

§ 05 — FAQSable / Launch editionp. 06

Fairquestions.

Anything else? Sable would read the docs for you — but here are the ones people ask first.

  • Anything with provenance: regulatory filings, peer-reviewed papers and preprints, patents, earnings-call transcripts, court dockets, procurement tenders, public datasets — plus the private documents you connect. It reads them end to end, including tables and footnotes, and scores every source for provenance before it is allowed to support a claim.

  • Every sentence in a brief is linked to one or more exact passages — document, page, paragraph, and the quoted span. Click a citation and the source opens at that line. If Sable can’t point to a passage, the sentence doesn’t ship.

  • Disagreement is information, so Sable keeps it. Each claim carries evidence for and against, a calibrated confidence, and a note on why the strongest counter-evidence was or wasn’t outweighed. Contradictions it can’t resolve are flagged for you instead of averaged away.

  • Yes. Desk plans connect data rooms, drives and PDF archives as private corpora. Your documents are cited exactly like public sources and are never used to train shared models.

  • Median eleven minutes for an exhaustive brief of around five thousand sources. You can watch it read in real time, steer the plan mid-run, or close the tab and get an email when it’s done.

  • No — Sable is a design and engineering concept by Cristhian Tilleria, exploring how an AI research product could look and feel at public launch. All names, figures and quotes are fictional, and the waitlist form is a demo that sends nothing anywhere.

§ 06 — WaitlistSable / Launch editionp. 07

Get in linebefore the inkdries.

Public access opens in waves. Leave an email and the first question you’d hand to Sable — we read every one.

Demo only — nothing is sent or stored.