Normalized for Mintlify from knowledge-base/neurigraph-memory-architecture/legacy-brain-api-pricing-strategy.mdx.

aiConnected Brain API

Strategic Product Document

Executive Summary

Brain is a persistent memory layer that sits above all AI platforms. Users maintain continuous context across ChatGPT, Claude, Gemini, and any MCP-compatible service. Brain is the foundation of aiConnected’s cognitive operating system for AI and future robotics. The name earns its meaning: aiConnected = All AIs connected through persistent memory.

The Problem

Every person using AI today has fragmented conversations:

ChatGPT knows about your project but Claude doesn’t
You explained your preferences to Gemini last week, now you’re starting over with GPT
Switch models and lose everything
Even within the same platform, context disappears after the conversation ends

The major platforms (OpenAI, Anthropic, Google) will never build cross-platform memory. Walled gardens serve their business interests. That interoperability gap is aiConnected’s entire market.

The Solution

Brain is a persistent memory layer that:

Stores conversation context across all AI platforms
Enables continuous memory regardless of which AI you use
Learns and accumulates knowledge over time
Creates switching costs that compound with usage

Target Markets

Consumer Market

Power users juggling multiple AI platforms
Developers and creators needing continuity
Professionals wanting AI that learns their preferences
Teams needing shared context

B2B Market (via Agency Platform)

Marketing agencies deploying AI for clients
Service businesses using AI-powered chat and voice
Enterprises wanting persistent AI relationships with customers

Pricing Structure

Consumer Tiers

Tier	Base Price	Includes	Overage
Starter	$1/mo	10 free actions	$0.50/action
Personal	$9/mo	100 actions	$0.25/action
Pro	$19/mo	Unlimited actions	-
Teams	$39/seat	Unlimited + shared memory	-

Storage Add-On

Add-On	Price	Includes
Storage expansion	$5/mo	+100,000 memories

Included Storage by Tier

Tier	Included Memories
Starter	1,000
Personal	10,000
Pro	100,000

Conversion Economics

The pricing structure drives natural upgrades:

Monthly Actions	Starter Cost	Personal Cost	Pro Cost
10	$1 (free actions)	$9	$19
15	$3.50	$9	$19
20	$6	$9	$19
25	$8.50	$9 ✓	$19
50	$21	$14	$19 ✓
100	$46	$34	$19 ✓

At 25 actions/month, Personal becomes the better value
At 50 actions/month, Pro becomes the better value
The $1 entry point filters out non-serious users while remaining accessible
The 10 free actions (a $5 value) let users experience the product before paying overage

Revenue Projections

Distribution Assumptions

Segment	Percentage	Average Monthly Revenue
Starter	50%	$4 ($ 1 + ~6 paid actions)
Personal	30%	$11 ($ 9 + overage)
Pro	15%	$19
Teams	5%	$39

Brain Revenue by Scale

Users	Monthly Revenue	Annual Revenue
1,000	$10,100	$121,200
10,000	$101,000	$1,212,000
100,000	$1,010,000	$12,120,000
1,000,000	$10,100,000	$121,200,000

Personas Revenue (Additional)

Average 1.5 personas per user at $9/month average:

Users	Persona Revenue/Month	Persona Revenue/Year
1,000	$13,500	$162,000
10,000	$135,000	$1,620,000
100,000	$1,350,000	$16,200,000
1,000,000	$13,500,000	$162,000,000

Storage Upsell Revenue

Assuming 10% of users purchase average 2 storage add-ons:

Users	Storage Revenue/Month	Storage Revenue/Year
100,000	$100,000	$1,200,000
1,000,000	$1,000,000	$12,000,000

Combined Revenue (Brain + Personas + Storage)

Users	Monthly Revenue	Annual Revenue
1,000	$23,600	$283,200
10,000	$236,000	$2,832,000
100,000	$2,460,000	$29,520,000
1,000,000	$24,600,000	$295,200,000

Infrastructure Costs

Compute Strategy

Self-hosted LLM (Ollama or similar) rather than paying per-token to third-party providers. Rationale:

Summarization and keyword extraction don’t require frontier models
Chinese models (DeepSeek, Qwen) cost ~$0.01 per 1M tokens
Self-hosted reduces this to fixed infrastructure cost
Break-even vs API pricing occurs around 200-400 active businesses

Compute Costs by Scale

Users	Monthly Infrastructure
1,000	$3,000
10,000	$8,000
100,000	$50,000
1,000,000	$150,000

Storage Architecture

State	What’s Stored	Size
Active (recent)	Full conversation, uncompressed	~2MB
Archived (after X days)	Full conversation, compressed	~200KB (90% compression)
Search index	Summary + keywords + embeddings (internal only)	~10KB

Storage Costs at 100,000 Users

Assuming 10,000 memories average per user, 90% archived:

Type	Size	Cost/GB/Month	Monthly Cost
Hot (active + index)	210 TB	$0.02	$4,200
Cold (archived)	180 TB	$0.004	$720
Total	390 TB		$4,920

Storage Costs at Scale

Users	Total Storage	Monthly Cost
100,000	~400 TB	$5,000
1,000,000	~4 PB	$50,000

Profit Projections

At 100,000 Users

Item	Monthly
Revenue (Brain + Personas + Storage)	$2,460,000
Compute	$50,000
Storage	$5,000
Total Cost	$55,000
Profit	$2,405,000
Annual Profit	$28,860,000

At 1,000,000 Users

Item	Monthly
Revenue (Brain + Personas + Storage)	$24,600,000
Compute	$150,000
Storage	$50,000
Total Cost	$200,000
Profit	$24,400,000
Annual Profit	$292,800,000

Gross Margins

Scale	Gross Margin
1,000 users	~70%
100,000 users	~98%
1,000,000 users	~99%

Technical Architecture

Integration Model

Brain operates as an MCP (Model Context Protocol) server:

Integrates the same way as Google Calendar, Gmail, GitHub MCPs
Works with Claude natively
Other platforms integrate as MCP adoption grows
No proprietary protocol needed; uses existing standard

Memory Operations

Operation	What Happens
Store	Full conversation saved, compressed after X days, summary and keywords generated internally for search
Search	Internal summaries and keywords searched, relevant memories identified
Retrieve	Full conversation decompressed and returned to active context

What Users See (Metadata Only)

Memory exists
Token count
Attachments
Timestamp
Source conversation

What Stays Hidden (Trade Secret)

Summaries (internal indexing only)
Keywords (internal indexing only)
Ranking/relevance algorithm
Compression method
Retrieval logic
Storage architecture

Competitive Protection

Strategy: Trade Secret Over Patent

Approach	Trade-Off
Patent	Requires full public disclosure; 20-year protection in theory
Trade Secret	No disclosure; protection lasts as long as secret is kept

Patents are not pursued because:

Enforcement against well-funded competitors costs millions
International coverage is limited (China doesn’t honor US patents)
Competitors can design around patents
Software patents are increasingly weak in courts
Filing requires disclosing exactly what we’re protecting

Protection Layers

Component	Strategy
Core memory mechanism	Trade secret
Brand (aiConnected, Brain)	Trademark
Code	Copyright (automatic)
UI/UX innovations	Possibly patent if unique enough

What Protects Us Over Time

Timeframe	Primary Moat
Year 1	Secrecy + speed to market
Year 2	User data gravity (memories accumulate)
Year 3+	Ecosystem lock-in (personas, integrations, agencies)

Why Competitors Can’t Easily Replicate

Factor	Protection
No public spec	6-12 months of guessing for competitors
Hidden summarization logic	They build inferior version first
No insight into search quality	Trial and error required
Data gravity	Even if replicated, user memories don’t transfer

Strategic Position

The Gap We Fill

Platform	Their Memory	The Limitation
OpenAI	ChatGPT memory	Only works in ChatGPT
Anthropic	Claude memory	Only works in Claude
Google	Gemini memory	Only works in Gemini
Microsoft	Copilot memory	Only works in Microsoft ecosystem
aiConnected	Brain	Works everywhere

The major platforms have zero incentive to make memory portable. Their business model depends on keeping users inside their ecosystem. This interoperability gap is permanent and is our entire market.

Brain Within aiConnected

Brain is not a standalone business. Brain is the foundation of the cognitive operating system.

aiConnected Cognitive Operating System
    │
    ├── Brain (memory, continuity, learning)
    │
    ├── Knowledge (retrieval, expertise)
    │
    ├── Voice (verbal communication)
    │
    ├── Chat (text communication)
    │
    ├── Personas (identity, behavior)
    │
    └── Future: Vision, Motor Control (robotics)

Long-Term Vision

aiConnected is building the cognitive operating system for the physical AI era. Today’s products (Brain, Knowledge, Voice, Chat) generate revenue and establish the memory layer. Tomorrow, this same infrastructure powers robotics:

Component	Role in Robotics
Brain	Remembers tasks, learns from experience, maintains continuity
Knowledge	Accesses manuals, procedures, domain expertise
Voice	Verbal interaction with humans
Chat	Text-based commands, logging, reporting
Persona	Consistent personality, appropriate behavior per context

Any robotics company can build a body. Any robotics company can license an LLM. Nobody else is building the integrated cognitive stack that connects persistent memory, retrievable knowledge, multi-modal communication, and consistent identity.

Revenue Comparison

Brain vs Agency Platform

Business	Revenue at Scale
Agency platform (100 agencies × $500/mo)	$600,000/year
Brain (100,000 users)	$29,520,000/year

Brain is 49x larger at modest scale. However, Brain remains one product within aiConnected because the long-term goal requires the complete cognitive stack.

Revenue Layers

Layer	Role
Brain	Foundation, entry point, highest retention
Personas	Monetization layer on top of Brain
Agency Platform	Cash flow engine, funds Brain development
Knowledge/Voice/Chat APIs	Developer and platform revenue
Robotics Cognitive Layer	Future enterprise and manufacturer revenue

The Moat

Data Gravity: User memories accumulate in Brain. Switching means losing them.
Persona Relationships: A Writing Partner with 8 months of context on your book can’t be replicated elsewhere.
Cross-Platform Freedom: Ironically, Brain lets users use any LLM they want, but Brain itself becomes irreplaceable.
Network Effects: Teams tier means shared memories. A company’s institutional knowledge lives in Brain.
Ecosystem: Agencies building on aiConnected infrastructure aren’t switching because OpenAI added memory to ChatGPT.

Summary

Brain is the persistent memory layer for all AI. It solves the fragmentation problem that major platforms will never solve because doing so conflicts with their business model. At scale, Brain alone generates $290M+ annual profit. Combined with Personas and the broader aiConnected product suite, the business supports the long-term vision of becoming the cognitive operating system for AI and robotics. The protection strategy is trade secret over patent. Secrecy buys time; data gravity makes the moat permanent. By the time anyone reverse-engineers a comparable system, users have years of memories they won’t abandon. aiConnected: All AIs connected through persistent memory.

Document Version: 1.0 Last Updated: April 2026

​aiConnected Brain API

​Strategic Product Document

​Executive Summary

​The Problem

​The Solution

​Target Markets

​Consumer Market

​B2B Market (via Agency Platform)

​Pricing Structure

​Consumer Tiers

​Storage Add-On

​Included Storage by Tier

​Conversion Economics

​Revenue Projections

​Distribution Assumptions

​Brain Revenue by Scale

​Personas Revenue (Additional)

​Storage Upsell Revenue

​Combined Revenue (Brain + Personas + Storage)

​Infrastructure Costs

​Compute Strategy

​Compute Costs by Scale

​Storage Architecture

​Storage Costs at 100,000 Users

​Storage Costs at Scale

​Profit Projections

​At 100,000 Users

​At 1,000,000 Users

​Gross Margins

​Technical Architecture

​Integration Model

​Memory Operations

​What Users See (Metadata Only)

​What Stays Hidden (Trade Secret)

​Competitive Protection

​Strategy: Trade Secret Over Patent

​Protection Layers

​What Protects Us Over Time

​Why Competitors Can’t Easily Replicate

​Strategic Position

​The Gap We Fill

​Brain Within aiConnected

​Long-Term Vision

​Revenue Comparison

​Brain vs Agency Platform

​Revenue Layers

​The Moat

​Summary

aiConnected Brain API

Strategic Product Document

Executive Summary

The Problem

The Solution

Target Markets

Consumer Market

B2B Market (via Agency Platform)

Pricing Structure

Consumer Tiers

Storage Add-On

Included Storage by Tier

Conversion Economics

Revenue Projections

Distribution Assumptions

Brain Revenue by Scale

Personas Revenue (Additional)

Storage Upsell Revenue

Combined Revenue (Brain + Personas + Storage)

Infrastructure Costs

Compute Strategy

Compute Costs by Scale

Storage Architecture

Storage Costs at 100,000 Users

Storage Costs at Scale

Profit Projections

At 100,000 Users

At 1,000,000 Users

Gross Margins

Technical Architecture

Integration Model

Memory Operations

What Users See (Metadata Only)

What Stays Hidden (Trade Secret)

Competitive Protection

Strategy: Trade Secret Over Patent

Protection Layers

What Protects Us Over Time

Why Competitors Can’t Easily Replicate

Strategic Position

The Gap We Fill

Brain Within aiConnected

Long-Term Vision

Revenue Comparison

Brain vs Agency Platform

Revenue Layers

The Moat

Summary