alignment

Capabilities


01

Persistent agents

Give every person an agent that carries relevant context forward instead of starting each conversation from zero.

02

Governed knowledge

Assign approved knowledge to the people and agents permitted to use it.

03

Connected agents

Preserve shared decisions and project context as teams work and priorities evolve.

AB
04

Secure data infrastructure

Your intelligence is stored only with database providers holding SOC 2 Type II attestations or ISO/IEC 27001 certification.

How Iridium works

Every operation.
One queryable intelligence.

Intelligence boundariesGoverned
Personal
Private working memory
Owner controlled
Team
Governed shared knowledge
Permissioned
Leadership
Approved organisational signals
Read only
Knowledge
Company and assigned sources
Curated
Evidence
Sources, timing and decisions
Traceable
Providers
ChatGPT, Claude, Local and what comes next
Portable
Why Iridium

Your business intelligence.
Independent of the model.

Work completed inside an AI platform can leave valuable context tied to that provider. Iridium stores your organisation’s memory, knowledge and decisions independently—keeping them persistent, portable and under your control.

Preview|--:--:--
0
Start from as few as 10 agents
0
Memory reads / writes per month
0+
Compatible with hundreds of LLMs
0.00%
Uptime this quarter
connected context

Persistent intelligenceacross any model or harness.

Bring the right memory, knowledge and evidence into every authorised agent—without repeatedly loading the whole archive.

ChatGPT
AI interface
Claude
AI interface
Documents
Knowledge
Conversations
Memory
Decisions
Evidence
Customers
Business context
Projects
Team context
Policies
Governance
Feedback
Learning signal
Operations
Business signal
Personal agents
Private intelligence
Team agents
Shared intelligence
ChatGPT
AI interface
Claude
AI interface
Documents
Knowledge
Conversations
Memory
Decisions
Evidence
Customers
Business context
Projects
Team context
Policies
Governance
Feedback
Learning signal
Operations
Business signal
Personal agents
Private intelligence
Team agents
Shared intelligence
Team agents
Shared intelligence
Personal agents
Private intelligence
Operations
Business signal
Feedback
Learning signal
Policies
Governance
Projects
Team context
Customers
Business context
Decisions
Evidence
Conversations
Memory
Documents
Knowledge
Claude
AI interface
ChatGPT
AI interface
Team agents
Shared intelligence
Personal agents
Private intelligence
Operations
Business signal
Feedback
Learning signal
Policies
Governance
Projects
Team context
Customers
Business context
Decisions
Evidence
Conversations
Memory
Documents
Knowledge
Claude
AI interface
ChatGPT
AI interface
Token-efficient retrieval

Retrieve the answer.
Not the archive.

Iridium combines semantic RAG with graph retrieval to find relevant passages and connected evidence, then sends only the smallest useful context to the model.

Semantically matchedGraph connectedQuery routedContext boundedEvidence linked

Semantic RAG

Finds the passages most relevant to the question without sending entire documents to the model.

Graph retrieval

Connects evidence across people, projects and decisions when an answer spans multiple sources.

Adaptive routing

Selects the right retrieval method for each question instead of searching everything.

Compact context

Reranks and deduplicates results before giving the model the strongest source-linked evidence.

Two people aligned through a shared flow of intelligence.
For platform teams


const agent = {
  owner: 'authorised member',
  scope: ['personal memory', 'assigned knowledge'],
  models: ['ChatGPT', 'Claude']
}
94%performance benchmarkLONGMEMEVAL
20xtoken reductionVS AGENTIC SEARCH
<1hrrapid deploymentAVERAGE ONBOARDING
91%Lower P95 latencyVS VECTOR RAG
94%performance benchmarkLONGMEMEVAL
20xtoken reductionVS AGENTIC SEARCH
<1hrrapid deploymentAVERAGE ONBOARDING
91%Lower P95 latencyVS VECTOR RAG
94%performance benchmarkLONGMEMEVAL
20xtoken reductionVS AGENTIC SEARCH
<1hrrapid deploymentAVERAGE ONBOARDING
91%Lower P95 latencyVS VECTOR RAG
94%performance benchmarkLONGMEMEVAL
20xtoken reductionVS AGENTIC SEARCH
<1hrrapid deploymentAVERAGE ONBOARDING
91%Lower P95 latencyVS VECTOR RAG
94%performance benchmarkLONGMEMEVAL
20xtoken reductionVS AGENTIC SEARCH
<1hrrapid deploymentAVERAGE ONBOARDING
91%Lower P95 latencyVS VECTOR RAG
94%performance benchmarkLONGMEMEVAL
20xtoken reductionVS AGENTIC SEARCH
<1hrrapid deploymentAVERAGE ONBOARDING
91%Lower P95 latencyVS VECTOR RAG
What effective AI teams need
01 / 04

O

Operations

Team requirement, Daily workflows

Key Result

Context that carries forward

Built around durable organisational intelligence

Persistent memorySemantic retrievalGraph retrievalSource-linked evidencePersonal boundariesGoverned team knowledgeRead-only leadershipModel portability
Persistent memorySemantic retrievalGraph retrievalSource-linked evidencePersonal boundariesGoverned team knowledgeRead-only leadershipModel portability
Pricing

Simple, transparent
pricing

From start-up focused packs of 10 agents, through to enterprise scale deployment, our packages are designed to scale with you.

MonthlyAnnual
Most Popular
01

Starter

For one organisation starting with persistent agents

$100/agent/month

Based on 10 agents

  • 10 personal or team agent slots
  • 1 leadership reporting agent
  • Isolated memory and knowledge
  • ChatGPT and Claude connections
  • Organisation administration
Choose Starter
02

Scale

For larger teams and broader knowledge estates

$80/agent/month

Based on 20 agents and 5 reporting agents

  • Expanded agent allocations
  • Multiple leadership agents
  • Additional governed sources
  • Advanced team permissions
  • Priority onboarding
  • Organisation intelligence views
Talk to us
03

Enterprise

For complex, regulated or multi-entity organisations

Custom
  • Everything in Scale
  • Custom intelligence boundaries
  • Dedicated implementation support
  • Custom connection architecture
  • Contractual service levels
  • Security and governance review
  • Custom commercial terms
Contact sales

Starter and Scale pricing is per agent and based on the minimum allocations shown. Annual and Enterprise commercial terms are illustrative until confirmed. View plan details

See Iridium workwith your knowledge.

Bring a real workflow and we’ll show how Iridium preserves its memory, retrieves the right evidence and carries that intelligence across your AI tools.

Rent the model. Keep the intelligence.