CONFIDENTIAL  ·  NOT FOR DISTRIBUTION  ·  FRONTIER LLM, INC.  ·  2026
FRONTIER LLM  //  INVESTOR BRIEF

Pre-seed round.
SAFE. $1.5M.

We're building the infrastructure layer for production AI agents — a compact LLM (EchoNest) and a safety + observability platform (Agentic Rails) designed for teams that can't afford to guess what their models are doing.

YC POST-MONEY SAFE $12M CAP 20% DISCOUNT MFN
INSTRUMENT

What you're signing

A SAFE (Simple Agreement for Future Equity) is not a loan. There is no interest, no repayment schedule, and no maturity date. You give us capital now; it converts into equity when we raise a priced Series A, or when there is a liquidity event (acquisition or IPO). It is the standard early-stage instrument at YC and across Silicon Valley.

This round uses the YC post-money SAFE template — the most investor-friendly and attorney-reviewed version available. Your ownership percentage at conversion is determined by the cap and discount, whichever gives you the better price.

HOW CONVERSION WORKS
01
Valuation cap — $12M
Your SAFE converts as if Frontier was valued at no more than $12M, regardless of what the Series A actually values the company at. If the A is at $40M, you still get $12M pricing.
02
Discount — 20%
Alternatively, your SAFE can convert at 80% of the Series A price per share — whichever gives you the lower (more favorable) conversion price applies.
03
MFN clause
If we issue a subsequent SAFE with better terms before the next priced round, you automatically receive those terms instead.
04
Pro-rata rights (checks ≥ $250K)
Investors writing $250K or more receive the right to participate in the Series A to maintain their ownership percentage.
THE COMPANY

What Frontier LLM is building

Enterprise teams are deploying agents that touch real systems — billing, support queues, customer records. The models they're using are powerful and unpredictable. There is no production-grade platform for controlling, tracing, and trusting what those agents actually do.

We're building two interlocking products:

ECHONEST
Compact instruction-following LLM
102.8M parameter model built on a novel hierarchical LSTM-attention architecture. 512-token context, trained on ~100k instruction-response pairs across general, Python, and multilingual code datasets. Currently epoch 1 of 20.
AGENTIC RAILS
Guardrails, orchestration, observability
Policy enforcement per tool call, deterministic orchestration across models, and full run tracing. Works with any underlying LLM, not just EchoNest.
USE OF FUNDS

Where $1.5M goes

18-month runway to EchoNest v1 launch, Agentic Rails general availability, and first design-partner ARR.

ALLOCATION
AMOUNT
%
Compute — EchoNest training
GPU time to complete training runs through v1
$600K
40%
Engineering headcount
2 senior engineers (infra + ML)
$525K
35%
Agentic Rails product + go-to-market
Design partners, docs, early sales
$225K
15%
Legal, ops, reserve
Entity, IP, buffer
$150K
10%
MILESTONES

18-month targets

Q2 2026 · M+3
EchoNest training complete
All 20 epochs finished. Internal evals pass coding and instruction-following benchmarks. Weights and inference server published.
Q3 2026 · M+6
Agentic Rails private beta
5 design partners live. Guardrails + observability APIs available. First production deployments running on customer stacks.
Q1 2027 · M+12
General availability + first ARR
Public launch of both products. $100K–$250K ARR from 3–5 enterprise design partners converted to paid. Series A process begins.
Q3 2027 · M+18
Series A close
SAFE instruments convert. Target: $8–12M at $40–60M post-money. This is the event at which your investment converts to equity.

Ready to move forward?

We're closing rolling. Request the full data room — cap table, technical architecture, and draft SAFE — and we'll follow up within 48 hours.

Request data room → Schedule a call