●  COHORT 01 — APPLICATIONS OPEN

Apply to the first
CTI cohort.

250 spots. Hand-picked. Zero servers.
The voice layer that lives above every AI — and measures your DPC.

Claude · ChatGPT · Gemini · Perplexity · Grok · 3 more

▶ Apply to the cohort See how it works →
0/250
Spots taken
0
CTI innovations
0
AI platforms
$0
Server — local-first
IRIS Memory Engine
v2.2 · CTI
847
MSGS
64
SESSIONS
3
PLATS
🎤 Voice query
Claude AI 94%
…quantum entanglement can't transmit information faster than light because of the no-cloning theorem…
Gemini AI 87%
…a quantum computer with ~4000 logical qubits could factor RSA-2048 in hours. NIST standardized CRYSTALS-Kyber…
ChatGPT AI 61%
…post-quantum cryptography uses lattices and stays secure even against quantum adversaries…
IRIS responds
—
Claude ChatGPT Gemini Perplexity Grok Copilot AI Studio + any AI chat Claude ChatGPT Gemini Perplexity Grok Copilot AI Studio + any AI chat
The problem

Your knowledge dies in the scroll.

Every AI platform locks you inside its walls. Claude can't see your ChatGPT threads. None of them can see your Gemini ones. They have no incentive to connect any of it — but you do.

94%
of insights generated inside AI chats are never revisited
4.2h
of AI chat per day — most of that knowledge evaporates
3.7×
AI platforms used by the average power user — each one in its silo
$0
tools that recover that knowledge across platforms — until IRIS
How it works

Three actions.
One voice.

01 —
Auto-indexes — zero clicks
Every time you open a chat in Claude, ChatGPT or Gemini, IRIS captures it in the background. Differential indexing v2.2: only new messages get sent. Zero friction, 90% less traffic.
CTI-06 · Differential Indexing · MutationObserver · SHA-256 dedup
02 —
Ask out loud — semantic search
"Hey IRIS — what did I discuss about quantum physics?" IRIS listens, parses your intent, and searches every chat with TF-IDF + cosine similarity. The Context Window Manager ranks by recency and action weight.
CTI-03 · CWM · Web Speech API · ElevenLabs Flash v2.5
03 —
IRIS responds — chunked audio
It pulls cross-platform fragments and reads them back in 20-second audio chunks. Surfaces connections between concepts. You keep doing what you were doing. Local-first: your memory never leaves the browser.
CTI-08 · Semantic Chunking · Local-first · $0 infra cost
Cognitive Throughput Infrastructure

The first system that measures
how you think with AI.

CTI isn't a feature. It's a new category. IRIS introduces DPC — Decisions Per Cycle, the first metric that quantifies cognitive throughput. Most power users run at 1.5 DPC. Top performers break past 4.

Your DPC this week — beta v2.2 average
0.0
decisions / cycle
LOW · 1.5 HEALTHY · 3.0 PEAK · 5.0
CTI-01
DPC Tracker
The first metric for AI-augmented thinking — decisions per cycle.
CTI-02
Cognitive Load Balancer
IRIS reads complex text slower, simple text faster. Adaptive throughput.
CTI-03
Context Window Manager
Recent and revisited content ranks higher. Your memory has weight.
CTI-04
Knowledge Graph
IRIS surfaces concepts crossing platforms — connections you didn't know existed.
CTI-05
Briefing Mode
90 seconds of morning audio: your threads, decisions, patterns.
CTI-06
Differential Indexing
Only new messages get sent. 90% less traffic on long chats.
CTI-07
Voice Autocomplete
IRIS pre-computes results while you speak. Zero search latency.
CTI-08
Semantic Chunking
Long content split into 20-second units. Higher information retention.
CTI-09
Throughput Analytics
Your DPC dashboard. The first time you see how you actually think.
Product roadmap

Three layers.
One system.

IRIS evolves in layers. Each builds on the last. Each widens the moat.

01
Voice Reader
Reads chats out loud. Multi-platform. Voice commands. ElevenLabs Flash v2.5.
LIVE
02
Memory Engine — v2.2 CTI
Indexes your full AI history. Cross-platform semantic search. Voice queries with CWM. Diff engine active.
LIVE BETA
03
Active Intelligence
Morning briefings. Pattern detection. Proactive cross-platform connections. Your IRIS speaks first.
Q4 '25
Live demo · 90 seconds

Watch IRIS
operate.

One take. Real voice. Three platforms. Zero servers. The cohort sees it before anyone else.

DEMO · DROPPING WITH COHORT 01
CAPTURED LIVE · NO POST-PROCESSING · v2.2 CTI
"No AI platform can connect
your conversations to each other —
because each one needs you to stay inside."
— The IRIS thesis
0/250 · COHORT 01 · APPLICATIONS OPEN

Apply to the cohort.

Not a waitlist. Selection.
We're picking 250 AI power users willing to co-define CTI from day one.

Why we select
IRIS is only useful for someone already living in 2+ AIs every week. If you join and don't use it, you don't learn — and neither do we. Cohort 01 is co-research: your DPC becomes the first public dataset of AI-augmented cognitive productivity.

No spam · Zero data on servers · Local-first · Reply < 48h

◉ ACCESS
Cohort 01 ships with the full CTI engine: DPC tracker, CWM, differential indexing v2.2.
◎ VOICE
ElevenLabs Flash v2.5 unlimited throughout the cohort. The full IRIS voice.
◈ LOCKED
Cohort price locked. $7/month forever — frozen the moment 250 spots close.
✓
Application received.
You'll hear back in less than 48h.
In the meantime — share IRIS with a power user you know. The quality of the cohort is defined by the network.
→ Share on X → Copy link