AI RESEARCH COMPANY

PROJECT 848

RUNE DIGITAL · ACTIVELY RAISING

AI research company building pheromone-inspired memory protocols. Like trails that strengthen with use and fade when abandoned.

$0K
Seed Raise
0
Papers Scanned
0%
Novelty Score
Dec 16
Pitch Submitted
Scroll to explore ↓
01 — THE PROBLEM

AI Forgets Everything

Every conversation starts from zero.

Memory Fragmentation: Data Loss Imminent
Enterprise AI spends billions annually,
but every conversation starts from zero.
Context windows reset. Knowledge vanishes.
The AI you trained yesterday doesn't remember today.
The Gap
Everyone is building memory inside models. No one has built it between them — at the orchestration layer.
02 — THE DISCOVERY

Ants Solved This 100 Million Years Ago

Ant colonies navigate without central coordination using pheromone trails — chemical traces that strengthen with use and decay when neglected. We call these patterns digital remnants. AI conversations leave the same patterns.

The Metaphor
We're not building smarter ants. We're building the trail layer.
03 — THE STACK

SEED-ANT-PRISM Protocol

An orchestration layer that sits between user and model. Works with GPT, Claude, Gemini, Llama — no retraining. Cross-session memory that existing approaches can't touch.

SEED-ANT-PRISM Architecture

SEED compresses context into a tiny formula — like a Minecraft world seed that regenerates reality.

ANT trails replace vector search — paths strengthen with use, decay with neglect.

PRISM refracts output into user-specific wavelengths — same question, different answer.

Development Stage
✓ Research validated (100% novelty, 191 papers)
✓ Working protocols (KINGSLAYER, H-010 demonstrations)
✓ 27.9GB naturalistic AI dataset collected
◐ CMD School beta (Q1 2026)
◐ Enterprise licensing (Q2 2026)
04 — IDLE CONSOLIDATION

Memory That Improves While Idle

Biological memory doesn't stop when you stop thinking. During sleep, your brain consolidates memories, strengthens connections, and prunes the irrelevant. We replicated this process.

PASSIVE DECAY
Unused trails fade over time, preventing irrelevant information from clogging the system.

CROSS-REFERENCE STRENGTHENING
Trails that connect related concepts strengthen each other — even when the system isn't being queried.

DORMANT RECOVERY
Old, faded trails can be reactivated by new context — like remembering a song lyric when you hear the tune.

Why This Matters
Current AI forgets everything between sessions. We remember — and improve — while idle.
05 — VALIDATION

We Proved Our Own Novelty

We scanned for overlaps with ACO-ToT, LLM reasoning, and memory systems. ACO-ToT optimizes reasoning within a session. We persist context across sessions — at the orchestration layer, with any model.

0
Papers Scanned
0
Targeted Queries
0
Cross-Session Matches
The Gap We Found
ACO-ToT = in-session reasoning. We = cross-session persistence. Different category.
06 — TIMING

Google Proved the Category

Google announced Titans in December 2025 — memory for AI is validated. But their approach requires model retraining. Ours works today.

December 2024
Google Titans Paper
Academic research on memory for transformers
December 2025
Google Announces Titans
Requires model retraining, PhD-level expertise, massive compute
December 2025
Project 848 Validated
Works with existing models. Built in 27 days.
Google is building a better engine. We're building better roads.
THE ASK
$500K
Pre-Seed for 6-9 month sprint
60%
Research Team
25%
Compute
15%
Documentation
Milestone
Publishable paper + working implementation + first enterprise pilot
What We Need
Not just capital—partnership. Publishing guidance for first research paper. Portfolio company intros for enterprise pilots. Lilypad talent network for first hire. Founder Academy community for CMD School beta testing.
6-9
Month Sprint
Feb 14
Decision Deadline
Q1 26
Window Closes
M 1-2
Publishing guidance + University intros
M 2-3
Enterprise pilots via portfolio
M 3-4
First hire via Lilypad network
M 6-9
CMD School beta to Founder Academy
07 — THE TEAM

Built by Joel Parkinson

Founder & AI Researcher · Adelaide, South Australia

Background
4 weeks from concept to validated research. Pattern recognition as a superpower — the same thinking that finds gaps others miss.
27.9
GB Dataset Collected
4
Weeks to Validation
11+
Agent Swarm Built
Why I'm Uniquely Placed
Beat Google Titans to practical implementation. Built multi-agent orchestration systems with 11+ specialized agents. Collected 27.9GB of naturalistic AI interaction data that nobody else has.
08 — BUSINESS MODEL

Three Revenue Streams

Pre-revenue. Clear path to monetization across research, education, and enterprise.

01
Research Consulting
Protocol implementation for AI teams
02
CMD School
AI orchestration education platform (Q1 2026)
03
IP Licensing
Enterprise SEED-ANT-PRISM licenses
Target Audience
AI developers wrestling with context limits. Enterprises losing value to session fragmentation. Students learning AI orchestration from first principles.
09 — THE JOURNEY

4 Weeks. 100% Novel.

From non-coder to verified AI research — the full story in under 3 minutes.

RESOURCES

Deep Dive

GitHub (Public)
Open research, protocols & documentation
GitHub (Investor)
Full validation data & proprietary research
Request Pitch Deck
Full 11-slide investor presentation
LinkedIn
Connect with Joel Parkinson, Founder
📊

Request Pitch Deck

Full 11-slide investor presentation with detailed financials, timeline, and technical architecture.

Send Request

Response within 24 hours

🔐

Investor Repository Access

Private GitHub access to full validation data, proprietary research, 27.9GB conversation dataset, and R&D documentation.

Request Access

NDA may be required for full access