Sage 1.5BetaThe AI That Truly Knows You

Sage 1.5 is our active research experiment in stateful AI: an engine that maintains continuous awareness across every conversation, session, and channel, instead of resetting like a chatbot. It is the memory and reasoning layer behind the Intueo platform, and it is still in early beta.

Curious about the difference? Read our in-depth guide on AI employees versus traditional chatbots. Looking for phone automation? Check out our Voice AI solutions.

Persistent Memory

Every conversation builds on the last. Sage remembers your context, preferences, and past interactions.

Natural Conversations

Communicate naturally through text or voice. Sage understands context from weeks or months ago.

Instant Responses

Lightning-fast responses with deep contextual understanding of your needs and preferences.

The Sage 1.5 Experiment

What we are actually building

Sage 1.5 is a research effort to answer a single question: what does an AI feel like when it actually remembers you, persists across channels, and can act with real tools? It is not a polished product. It is a working prototype we run, study, and iterate on every week.

One brain. Multiple specialists. Continuous memory.

Inside the platform you build a roster of agents — a Financial Advisor, a Research Assistant, a Wellness Coach — each with their own persona, knowledge, and toolset. Talk to any of them, switch instantly, and what they learn stays with them.

Every agent runs on Sage 1.5 with three execution depths (Fast, Thinking, Deep), structured tool calls, and a memory system that decides on its own what is worth remembering and what is just noise.

Persistent memory across every session and channel
Three reasoning depths behind a single toggle
Voice and text on the same agent, same memory
Real tool calls, not just chat
Specialised agents instead of one generalist
Three Modes, One Brain

Pick the right depth for the task

Sage 1.5 ships with three execution modes, all available behind a single toggle. Switch instantly, or let the platform handle it for you when you attach images.

Sage 1.5 Fast

Quick, vision-capable responses for everyday messages, image attachments, and casual back-and-forth.

  • Lowest latency
  • Multi-image vision support
  • Best for everyday chat

Sage 1.5 Thinking

Balanced reasoning for moderately complex problems where you want a thoughtful answer without waiting.

  • Mid-depth reasoning
  • Expandable reasoning steps
  • Best for analysis & writing

Sage 1.5 Deep

Heavy reasoning for genuinely hard problems, when you want the model to think before it answers.

  • Maximum reasoning depth
  • Per-message override
  • Best for complex problems

Silent vision-aware model swapping

Attach an image while you are in Thinking or Deep mode and Sage automatically uses Fast for that single message, then immediately returns to your selected mode. The toggle never moves. You never see the swap. The platform hides the technical reality so you do not have to.

Powered By Sage 1.5

Sage is the engine. Intueo is the platform.

Sage 1.5 is the proprietary memory and reasoning layer underneath. The Intueo platform is where you actually use it: a studio for building, running, and managing a team of specialized AI agents.

Each agent has its own persona, persistent memory, knowledge base, and toolset. Switch between agents with a click. Call them by voice. Drag and drop images. Everything you tell one agent stays with that agent, across sessions and channels, for as long as you want it to.

Multiple specialized agents
No-code Agent Workshop
Voice calls with persistent memory
Custom MCP tool integrations
Multi-image multimodal pipeline
Installable PWA on iOS & Android
Request a private demo →

Persistent Memory

Core, archival, and recall memory persist across every session.

Multiple Agents

A roster of specialists instead of one generic assistant.

Voice Calls

Real conversational voice with the same agent and memory.

Stateful Memory

Genuine state, not an extended context window.

Core Capabilities

Everything that makes Sage different from traditional AI

A look inside the prototype

What Sage 1.5 actually looks like today

These are real screens from our internal beta build, captured this week. The design is still changing, but the core ideas are working.

Tool calls, not just chat

Each agent can use a curated toolset. Here, the Financial Advisor runs a structured web search and returns a formatted answer with citations.

A tool workshop, per agent

Hundreds of available tools across categories — core, memory, file, voice, custom MCP servers — wired into the agent that needs them.

Agents that reach out first

Schedule one-off or recurring messages so an agent can check in proactively — morning briefings, follow-ups, reminders, anything you script.

Real voice calls, same memory

Call the same agent you chat with. Voice is just another channel into the same persistent persona — interrupt, change topic, hang up, pick it back up later.

Stateful Intelligence

Every conversation builds on the last. Sage remembers your context, preferences, and past interactions.

Natural Conversations

Communicate naturally through text or voice. Sage understands context from weeks or months ago.

Conversation History

Access your complete interaction history. Search, review, and continue past conversations anytime.

Voice & Text

Seamlessly switch between voice calls and text chat without losing context.

Instant Responses

Lightning-fast responses with deep contextual understanding of your needs.

Privacy First

Your conversations and data are encrypted and completely under your control.

Sage 1.5: Memory System

Sage 1.5 keeps durable memory across sessions and channels, then brings only the relevant pieces back into context when needed.

Each agent has its own constellation of memory blocks — identity, knowledge, working state — that fill in over time.

How it works (plain-English pipeline)

1
CaptureSignals arrive from chat/voice, calendar, docs, ERPs/CRMs, and webhooks.
2
AssessA salience + policy check scores importance, recency, uniqueness, and trust.
3
WriteValuable items are saved to the right memory block; throwaway details stay ephemeral.
4
RetrieveFor each task, Sage builds a focused context using hybrid search with recency and role filters.
5
ActTools run with that focused context (e.g., offering non-overlapping calendar slots, citing sources).
6
Summarize & DecayOlder sessions auto-summarize; low-value crumbs expire; long-term memory stays compact.

The five memory layers

1. Working Context (Short-Term)

Live turn-by-turn buffer; not persisted by default.

2. Episodic Store (Sessions & Events)

Time-stamped interactions, decisions, and outcomes.

3. Profile & Preferences (Semantic)

Stable facts about people/orgs/assets with consent gates.

4. Knowledge & Files (Grounding)

SOPs, policies, PDFs, calendars, CRMs—indexed for retrieval.

5. Long-Term Archive (Consolidations)

Periodic rollups of episodes and changes for fast recall.

What Sage 1.5 remembers

  • • Stable preferences and task states
  • • Commitments and verified facts
  • • Pointers to source documents
  • • High-level summaries of past sessions

What it forgets or masks

  • • Raw PII without consent
  • • Transient chit-chat
  • • Secrets in tool outputs
  • • Low-salience fragments

Benefits you feel immediately

Continuity: picks up right where you left off—across channels
Personalization: remembers constraints and preferences
Faster, safer decisions: uses verified memories + citations
Lower cost/noise: summarizes old sessions; only right context loaded
Trust & control: inspect, edit, pin, or delete memories with reasons
Source-aware answers: prioritizes your internal knowledge
A guided tour of the platform

Real screens from the Intueo platform

These are not mockups. Every screen below is captured from the live beta, showing how you build agents, shape their memory, and replay months-old conversations on demand.

Agent Workshop

A roster of stateful beings, not chat sessions

Every agent on the left is a persistent entity with its own memory, tools, voice, and mind. Pick one to inspect, or craft a new one from scratch.

Step 1 — Model & Settings

Pick the brain. Tune the behaviour.

Choose from Sage 1.5 Thinking, Pro, Fast, or any frontier model — Gemini 3, Claude Opus 4.7, GPT-5.5, Kimi K2.5, GLM-5. Then dial in creativity, response length, conversation memory, and long-term memory size.

Step 2 — Personality

Describe the agent. AgentWizard writes the rest.

Type a sentence — "a friendly coding mentor" — and the wizard generates a full persona, communication style, values, approach, and knowledge base as editable memory blocks before the agent is even born.

Conversation History

Replay any conversation. From weeks or months ago.

Unlike ChatGPT or Gemini, you do not start a new session every time. Search history, jump to any afternoon or night from any past day, and pick the thread back up — your agent already remembers it.

Reasoning Effort

Quick. Balanced. Deep. One toggle, per message.

Right inside the composer, choose how hard the agent should think before it replies. Quick for chit-chat, Balanced for everyday work, Deep when you want it to sit with the problem.

Sage vs Traditional Chatbots

FeatureTraditional ChatbotsSage 1.5
Memory
Resets every session
Persistent core, archival & recall memory
Context Window
Limited to current chat
Genuine state across all sessions
Specialization
One generic assistant
Team of specialized agents you build
Reasoning Modes
Single fixed model
Fast / Thinking / Deep, per-message
Vision Handling
Manual model selection
Silent auto-swap on image attach
Voice Support
Text only or separate product
Voice & text on the same persistent agent
Tool Integration
Limited capabilities
Custom tools, MCP servers, knowledge files
Personalization
Generic interactions
Per-agent persona, prompt & memory
← Scroll horizontally to see all features →
Beta & Research

Sage 1.5 is still in the lab

We are deliberately keeping Sage 1.5 in private beta while we test how stateful agents behave in real workflows. Things break. Things change weekly. We learn fast and ship faster.

Active research

We are studying how persistent memory, multi-agent rosters, and tool-using agents change the way people actually work. The product follows the research, not the other way around.

Not generally available

Sage 1.5 is not a product you can sign up for today. Access is limited to design partners, collaborators, and pilot customers we work with directly.

Live demos for the curious

If you want to see what we are building, we are happy to walk you through a private demo and answer questions about the architecture, the roadmap, and the trade-offs.

Want to collaborate, invest, or just see it run?

We are talking to early design partners, technical collaborators, and investors who care about stateful AI. Reach out and we will set up a live walkthrough of the prototype.

Under the Hood: How an AI Employee Works

Stateful & Continuous
Model-Agnostic Swap
Governed & Auditable
User
Short-term Context

AI Employee

LLM Core
Gemini / Claude / etc.
Tool Calls
CRM, ERP, Email...
Recent Memory
Session cache (hours days)
Long-Term Memory
Vector + SQL store
Sleep-time Jobs
Summarise & prune while offline
Persistent Storage
Lives on the Intueo platform, not inside the model.

Curious where stateful AI is heading?

Sage 1.5 is our research into AI that actually remembers — still in beta, still evolving.

Want a closer look? Schedule a private demo and we will show you the prototype live.

Request a private demo →