Adopt AI without getting sanctioned, sued, or suspended.
LEXICORE is the trust-first legal operating system for attorneys and small firms. Every AI output is source-linked, human-approved, and audit-trailed. Built on CORTEX for tamper-evident firm memory.
Attorneys must adopt AI and are afraid of it. The fear is rational.
The Charlotin database tracks AI-hallucination sanctions cases worldwide. The count has crossed 1,227, roughly 90 percent of them in 2025, with new cases landing weekly into 2026. The pressure to adopt is real. The market has not built the answer.
Sanctions climbed from $500 admonishments to routine five figures to $30,000-plus this March (Sixth Circuit, Whiting v. City of Athens). Bar suspensions and pro hac vice revocations are now in the mix, not just fines. Sanctioned filings have included briefs produced with funded incumbents like CoCounsel and Westlaw, so capability alone has not solved trust.
Reasoning is a commodity. Trust is the only moat a new entrant can still build.
Frontier models have made legal reasoning a commodity. What still defends a position in this market is distribution, content, workflow depth, and trust. A new entrant cannot beat the giants on the first three. Trust is the one moat a focused team can build without a billion dollars or a century of case law, and it is the moat the market most visibly lacks. LEXICORE occupies that seam, deliberately.
Five rules the product enforces. Not marketing copy. Defaults.
These are the defaults enforced by the runtime. Disabling one is loud, visible, and logged. An attorney can show a judge or a bar investigator exactly how every output was produced and exactly who approved it.
Observe first. Never act first.
No action without explicit human confirmation.
All outputs are attributable, reviewable, and reversible.
No silent automation. Ever.
Failure states are visible, not hidden.
The moat is the trust architecture, plus CORTEX, the auditable firm memory.
The four pillars compound. Verifiability without an audit trail is a marketing claim. Approval gates without privilege protection are theater. The pillars only work together, which is why LEXICORE is the runtime, not a plug-in.
Verifiable by construction
Every AI assertion links to a real, checkable source: a document the firm owns, a public record, or a licensed corpus. Nothing surfaces without a traceable citation. This is the answer to hallucination, built into the data layer, not bolted onto the chat.
Human-controlled by design
Approval gates on anything that leaves the building. The attorney is always in the loop, always on the record, always the responsible party. The product makes the gate easy rather than burdensome. Observe first, never act first.
Auditable end to end
Every action, source, edit, and approval is logged in a tamper-evident trail an attorney can produce for a judge, a bar inquiry, or a malpractice carrier. CORTEX is the substrate: per-firm auditable memory of how every output was produced.
Confidential and private
Privilege preserved by default. Private cloud and on-prem deployments for firms that will not put privileged data on a shared tenant. SOC2 is in the use of funds, not a future promise. The ATERNA on-prem practice applied to the most regulated workload.
Every source, edit, and approval is shelved the moment it happens and stays on the record. A firm on CORTEX can produce the whole chain for a judge, a bar inquiry, or a malpractice carrier.
Not a Clio killer. The seam the giants leave open.
Solo and small firms of 1 to 20 attorneys carry the personal exposure to sanctions and malpractice that the funded incumbents pitch around.
Who they pitch
Am Law 100 and global enterprise
Solo and small firms (1 to 20 attorneys)
Annual cost per attorney
$10K to $30K+ on the funded incumbents
Per-seat priced so a 5-attorney firm can actually buy it
What the buyer fears
Falling behind their peer firms
A sanctioned brief in the next filing
Verification surface
Available, often disabled by default for speed
On by default. The product breaks if it is off.
Audit trail
Some logging, vendor-controlled retention
Tamper-evident, firm-owned, bar-inquiry-ready
On-prem option
Rare, enterprise-only, custom-priced
A productized tier in the price list
Sources: 2026 Harvey, Legora, and Clio funding announcements (GIC and Sequoia, NEA), the Charlotin sanctions database, Sixth Circuit Whiting v. City of Athens (March 2026), Lacey v. State Farm.
Priced for the firm that buys it. Not the firm Harvey pitches.
Pricing reflects who can actually buy the product, not who Clio and Harvey put on their case studies. At $249 a seat a month, a five-attorney firm pays $14,940 a year for all five seats, against $10K to $30K per attorney on the funded incumbents, and gets a defensible AI workflow with the audit trail on by default.
Practice Seat
Per-attorney access to the trust-first legal OS: research with source-linked citations, case management, document intake and analysis, approval-gated drafting. Includes audit-log retention and citation-verification catch rate dashboards.
Matter Workflow
Per-matter charge for heavy workflows: discovery analysis, deposition prep, expert-witness vulnerability analysis, motion drafting with full chain-of-citation. Pay only for matters where the verification surface is the product.
On-Prem Deployment
Private or on-prem install for firms that will not put privileged data in a shared cloud. Includes a dedicated CORTEX instance, SOC2-aligned controls, and the firm-owned audit trail. The tier the giants do not ship.
$2.5M open at $12M pre-money.
One lead partner and a small syndicate. $14.5M post-money at close.
Open at $12M pre-money
Approximately 17 percent post-money. SAFE or priced. Single lead writing $1M to $1.5M plus a syndicate for the remainder. Pro rata rights for the lead. Sized to fund the trust architecture, the on-prem build, the bar-compliance work, and the first 30 paying firms.
Paying small firms in year one
Roughly $100K-plus MRR, SOC2 Type 1 attestation, on-prem GA, and the verification catch rate published as hard public data. Ready for a Series A at a $40M to $60M step-up on real trust metrics.
- Pre-money
- $12M. The deck and the operating model go to lead partners on request.
- At close
- $2.5M raised against a $14.5M post-money, about 17 percent.
We do not pitch LEXICORE as a Clio killer or a Harvey killer. We pitch it as the trust layer the funded giants leave open, productized for the firms most personally exposed to sanctions.
Investors who fund the moat the market is missing, not the slide that overclaims it, are the right partners for this round.
Twenty-four pages, sourced, ready for diligence.
The deck includes the sourced trust-crisis data, the wedge framing against Harvey, Clio, Lexis, and Thomson Reuters, the four-pillar trust architecture, the beachhead and ICP, the CORTEX moat, monetization, use of funds, milestones, and the term sheet shape.
The sourced sanctions data: the Charlotin database, the Sixth Circuit, the bar guidance count.
The explicit framing against Harvey, Clio, Lexis, and Thomson Reuters, and the seam they leave open.
The four-pillar trust architecture and the five defaults the runtime enforces.
The ideal customer: solo and small firms of 1 to 20 attorneys, and why they buy first.
The auditable firm memory underneath every output, and why it is a moat.
Practice Seat, Matter Workflow, and On-Prem, priced for the firm that buys.
Where $2.5M goes over twelve months.
60-plus firms, SOC2 Type 1, on-prem GA, the published catch rate.
The term sheet shape: $12M pre, SAFE or priced, one lead plus a syndicate.
Build the trust layer with us.
We are running this raise out of ATERNA Labs. We prefer a single lead partner of $1M to $1.5M plus a small syndicate for the remainder. On the call we walk the term sheet, the beachhead rationale, the CORTEX integration depth, and the bar-compliance model.
- Investor inquiries
- labs@aterna.ai
- Product
- lexicore.io
- On request
- Term sheet, beachhead rationale, CORTEX integration depth, bar-compliance model.
