The autonomous AI company
isn’t a deck.
VoidLens is live infrastructure with seven departments, twenty-nine agents, forty-nine thousand audit rows, and the safety rails you'd want before signing the check. We're raising a $2.5M pre-seed to expand the agent pool and open the founder pilot program. Below is what you need to evaluate that.
Advisory AI is over. Agentic AI ships work.
The 2025 wave gave the market chat tools that answer questions. The 2026 wave is AI that senses, decides, and securely acts. That requires a kernel: a real org-shaped runtime with capability gating, audit, budgets, and ladders. VoidLens is that kernel, and VoidLens is the first product on top.
The five engineering pillars
Rigor
Detect → heal → prevent. Every incident is a fingerprint, every fingerprint is canonical, every canon entry is a CI gate.
Creative leverage
The sharpest fix, not the obvious one. Pattern-notice across silos. The hard refactor that makes ten future fixes trivial.
Staff-engineer ship
Would they actually ship this? Spec, eval set, capability envelope, autonomy declaration. No magic, no vibes.
Genius + tenacity
No problem stays unsolved. Reframe, create, persist. Three strikes ⇒ deep-dive. Deep-dive ⇒ peer-review. Peer-review ⇒ canon.
Compounding stewardship
Every session is a deposit. Canon is the deliverable. The system gets sharper every week, not just every release.
Agentic AI: $53B in 2026, $471B by 2030.
Source: Allied Market Research, ServiceNow Agentic Index, Sequoia 2025 AI Market Map. Even the bear cases price agentic-AI at 4-7x the 2025 chatbot category by 2030.
Why now
Three things converged in 2026 that weren't true a year ago. One, frontier models (Claude Opus 4.6, Sonnet 4.6, Haiku 4.5) finally tool-call reliably enough to run a production agent loop without a human babysitter on every turn. Two, the MCP standard (Model Context Protocol) gave every SaaS a common surface — connectors no longer have to be bespoke integrations. Three, the buyer is ready. Compliance teams stopped saying "no AI" in 2025 and started saying "show me the audit trail" in 2026. VoidLens is what they get back.
Six weeks. From spec to live system.
We don't have ARR yet — this is a pre-seed. What we have is engineering momentum that's measurable. The audit log is the receipt.
What's next (the pilot path)
Founder pilot opens at 80% pilot-meter (we're at 60%). That milestone requires: revenue-stream first non-dry-run fire, customer-ops runner shipped (Board MUST-consult queued), and three operational pickups (V2 smoke-test, dignity-lint calibration, V1 cohort push). Realistic window: 2026-Q3 open, first paying pilot Q3-Q4.
The kernel is the moat.
Anyone can wire Claude or GPT into a Slack channel. The hard part is making that wire safe enough to ship, audit, and scale. Our kernel is six primitives that compose. Each one took weeks. Together they're a moat.
Guardian capability gate
Every external action proves itself against an allowlist before it fires. Forbidden list (payments, hard-deletes, unapproved publishes) is immutable. No prompt can override it.
Audit-every-action
Canonical JSONL schema, append-only, cross-OS-locked. Every row carries cost, duration, outcome, reason, run_id. The audit reader is part of VoidLens.
Autonomy ladder L0 → L3
Read-only → propose → act-with-review → act. Every climb requires a clean 48h shadow window. Any breach resets the clock. Mechanical, not vibes.
Spec-before-code
spec.json, worker.md, playbook.md, evals/eval-set.yaml. The contract is the design. The code is the receipt. Reviewable before, during, and after the ship.
Budget & rate discipline
Per-agent daily cost caps. Per-action rate limits. Fan-out budgets. The system halts cleanly. YELLOW before RED, every time.
Eval-driven deployment
Offline eval set + online sample. Pass-rate below threshold blocks the promotion. Behavior change is gated, not declared.
Five years. Conservative on revenue, honest on cost.
Pre-seed projections. Every number below is grounded in a stated assumption you can audit. Numbers in $K unless noted.
| Year | Paying teams | Blended ARPU/yr | Revenue | Gross margin | Burn | Headcount |
|---|---|---|---|---|---|---|
| 2026 (Y0 · pre-seed) | 3 design partners (free) | — | $0 | — | $420K | 2 |
| 2027 (Y1) | 200 | $6K | $1.2M | 72% | $1.8M | 6 |
| 2028 (Y2) | 1,500 | $9K | $13.5M | 76% | $2.4M | 14 |
| 2029 (Y3) | 7,000 | $12K | $84M | 78% | — | 32 |
| 2030 (Y4) | 22,000 | $14K | $308M | 80% | — | 78 |
Assumptions: blended ARPU grows as we move up-market and add Enterprise tier. Gross margin reflects inference cost reduction (frontier models trending ~30%/yr cheaper on the same capability tier) plus capacity-share with customers via bring-your-own-key. Burn excludes inference variable cost (passed through to customers post-Operator tier). Forecast updated 2026-05-12.
Founder-led. Engineering-heavy.
Solo today. Pre-seed dilution funds the first engineering hire and a design partner program. No sales leader until $1M ARR.
Jeffry Hicks · Founder & CEO
Logistics operator first, builder second. Nine years in the Army across two combat tours, specializing in convoy operations and logistics management at scale; then a decade running civilian-side logistics — dispatch, fleet, freight, the daily mechanics of moving real things through real systems on real deadlines. VoidLens shipped fast because I already knew what I needed from an AI workforce. The autonomous-AI-company thesis is a 10-year compounding bet; the kernel is the IP.
Open · Founding engineer
First post-pre-seed hire. Profile: 5-8 yrs platform/infra experience, strong on capability gating, audit trails, and live systems. Will own the void-orchestrator + scheduler v2 plus the V2-clone customer-instance path. Bring your worker.md.
$2.5M pre-seed. SAFE. 2026 cap.
Capital fuels twelve months of runway and the founder pilot program. Below is what the dollars buy.
Use of funds
Twelve months of operating runway and a deliberate, measured expansion plan. Capital efficiency is part of the thesis — we run the company with the kernel.
What we want from the round
Lead investor with conviction on the agentic-AI-as-org-runtime thesis. We'd rather have one investor who's read the audit log than five who've only seen the deck. Founder-friendly terms (SAFE preferred). Strategic intros into mid-market & F500 design-partner candidates. No board seat at pre-seed.