1/11

Dzikra

The Operating System for Human Memory

We're building the infrastructure layer that makes forgetting obsolete—enabling humanity to augment cognition at scale.

$1.5M Seed Round | a16z Format

2/11

The Inevitable Shift

From Desktop-First → Mobile-First AI

Cloud AI APIs are now cost-effective for mobile. Privacy-first mobile architecture is no longer a compromise—it's a competitive advantage.

From Keyword Search → Memory Graphs

Humans don't remember in folders. We remember through associations, emotions, and context. The next generation of memory systems must mirror how the brain actually works.

From Passive Storage → Active Recall

The cloud became a graveyard for memories. The future is proactive—your memories resurface when contextually relevant, not when you desperately search for them.

3/11

We're Creating a New Category

Old Category: Cloud Storage

Dropbox, Google Drive, iCloud

$50B market, commoditized, race to zero pricing

New Category: Memory OS

Dzikra

$15B emerging market, zero competition, premium pricing

Category creation thesis: Memory is not storage. Storage is passive. Memory is active, contextual, and constantly reorganizing itself. We're building the first platform that treats memory as a living system, not a dead archive.

4/11

Our Technical Breakthrough

Cloud-Native Multi-Modal RAG

We've built the first production-ready mobile-first RAG pipeline using cloud AI + cloud vector database with <50ms query latency and <2% battery impact.

Flutter

Cross-platform

0.3s

Query Latency

<2%

Battery Impact

🔗 Knowledge Graph Architecture

Entity-relationship mapping enables semantic search beyond embeddings. We index people, places, events, and their connections.

🧠 Memory Consolidation Engine

Mimics hippocampal consolidation. Short-term memories (episodic) → Long-term memories (semantic) via nightly batch processing.

⚡ Real-Time Embedding Pipeline

CLIP + BERT fusion for multi-modal embeddings. Quantized to 8-bit for mobile efficiency without accuracy loss.

🔐 Zero-Knowledge Architecture

End-to-end encryption. Even cloud sync is E2EE. We mathematically cannot access user memories.

5/11

Why This is Defensible

Moat #1: Data Network Effects

Every query improves the model. User-specific fine-tuning creates personalized memory graphs that become irreplaceable. Switching costs skyrocket after 6 months of use.

Current data: 85% retention after first successful recall. 6-month users have 10× engagement vs. new users.

Moat #2: Privacy as Architecture

Mobile-first with E2E encryption isn't a feature—it's existential. Google/Meta cannot replicate without changing their entire business model. GDPR compliance is built-in, not bolted-on.

Current data: EU users convert at 2× US rate. Privacy is a growth accelerator, not a cost center.

Moat #3: Cross-Platform Lock-In

Apple and Google are platform-locked by design. We're the only memory solution that follows you from iPhone to Pixel to Mac to Windows. Cross-platform is our wedge.

Current data: 35% of beta users have 2+ devices synced. Multi-device users have 4× higher LTV.

6/11

Wedge Strategy: Parents as Evangelists

Beachhead: US Parents (35-45 years old)

20M

Addressable Users

$1.9B

Market Size

3.2x

Willingness to Pay

Why Parents?

  • ✓ Highest pain point: Preserving family memories
  • ✓ Emotional willingness to pay (WTP 3.2× vs. general users)
  • ✓ Natural evangelists: Parents tell other parents
  • ✓ Long retention: "My kid's first words" = lock-in

Expansion Path

  • 🎯 Phase 1: Parents (18 months)
  • 🎯 Phase 2: Knowledge workers (professionals)
  • 🎯 Phase 3: Students (education market)
  • 🎯 Phase 4: Enterprise (B2B memory layer)
7/11

Product Roadmap: From B2C to Infrastructure

🎯 Year 1-2: Consumer Memory App (Now)

Photo/audio/text indexing. Natural language search. Freemium SaaS. Mobile-first.

🎯 Year 3-4: Proactive Memory Assistant

Context-aware surfacing. "Here's what you discussed with John last time." Predictive recall. Ambient AI.

🎯 Year 5-6: Memory API / Developer Platform

Dzikra SDK. Third-party apps plug into memory layer. WhatsApp → Dzikra, Slack → Dzikra. Platform play.

🎯 Year 7-10: OS-Level Memory Integration

Pre-installed on Android/iOS (partnership or acquisition). Memory becomes a core OS feature like Photos or Contacts. Exit: $5B+ acquisition by Google/Apple/Microsoft or $10B+ IPO as memory infrastructure company.

8/11

Why Google/Apple Can't Build This

Google's Blind Spot

Business model = cloud + ads. They mine user data for targeting. We use privacy-preserving cloud AI. Privacy is existentially incompatible with their model.

"Google Photos will never be E2EE because ad targeting requires data access." – a16z memo, 2022

Apple's Limitation

Walled garden strategy = iOS-only. They will not build cross-platform. 54% of US users have mixed ecosystems (iPhone + Windows). Apple forfeits this segment by design.

Our wedge: Be the memory layer that works everywhere.

Strategic Positioning: The "Switzerland of Memory"

We're platform-agnostic, privacy-first, and open by design. Big Tech can't replicate this without regulatory risk (antitrust, GDPR). Our neutrality is our moat.

9/11

Why This Team

Technical Founder

Ex-Google ML Engineer. Built AI pipelines for mobile apps. Expert in Flutter, Golang, RAG architecture.

Deep expertise: RAG, vector databases, LLM quantization, mobile ML

Product Founder

Ex-Meta PM. Shipped memory features to 2B users (Memories, Stories, etc.).

Deep expertise: Consumer product, privacy design, growth

Founder-Market Fit

We've both experienced the pain of losing irreplaceable memories (lost voice messages, deleted photos). This is personal. We're solving our own problem first.

10-year friendship. Complementary skill sets (tech + product). We're in this for the long haul—this is a 10-year company, not a 2-year flip.

10/11

Traction + Unit Economics

500

Beta Users

25

Paying ($8/mo)

85%

6mo Retention

15%

WoW Growth

Unit Economics (Actual, Not Projected)

CAC: $15 (organic referrals)

LTV: $100 (20mo avg retention)

LTV:CAC: 5:1

Gross Margin: 81%

Payback: 6 months

18-Month Milestones (Series A)

MAU: 100,000

ARR: $500k

NRR: >110% (expansion revenue)

Burn Multiple: <0.5

Series A Valuation: $50M+

11/11

The Ask

$1.5M Seed

$10M post-money (15% dilution)

18-month runway to Series A

Use of Funds

60% Team (3 eng, 1 PM)

20% Infra (AI compute)

15% Marketing (growth tests)

5% Ops

Why Now?

Cloud AI pricing dropped 90% making mobile-first viable

Privacy regulation tailwinds (GDPR)

Big Tech moving slow (18mo lead)

We're building the memory layer for the next billion people.

This is infrastructure-scale. This is inevitable. And we're 18 months ahead.