Research Lab · Live

The mind doesn't
end at the skull.[1]

[1] "The mental world — the mind, the world of information processing — is not limited by the skin."

Gregory Bateson · Steps to an Ecology of Mind · 1972
Scroll
Sub-200ms voice AI
3-layer memory architecture
Agentic execution
Emotional voice response
Google Verified
Runs while you sleep
Custom model architecture
Sub-200ms voice AI
3-layer memory architecture
Agentic execution
Emotional voice response
Google Verified
Runs while you sleep
Custom model architecture
The Research Thesis

The mind doesn't end at the skull. We're building for what's outside it.

In 1998, philosophers Andy Clark and David Chalmers published The Extended Mind in the journal Analysis. Their argument was not metaphorical: the boundary of the cognitive system is not the body. A notebook that reliably stores memory is memory. A tool that offloads decisions is thinking. Cognition extends into whatever it reliably uses.

"If, as we confront some task, a part of the world functions as a process which, were it done in the head, we would have no hesitation in recognizing as part of the cognitive process — then that part of the world is part of the cognitive process."

Andy Clark & David Chalmers · The Extended Mind · Analysis, Vol. 58, 1998

That paper was about pocket diaries. We are building for what comes after. Not AI you pick up and consult — AI that already lives inside the thought, because it lives inside the data that generated it.

Where cognition actually happens
Biological mind
Intention, creativity, judgment, context
Orbis, extended cognition
Memory, pattern recognition, autonomous execution
Unified workspace data
Tasks, docs, email, calendar, sheets, files

The tool stops being a tool. It becomes infrastructure for thought.

Why We Started

Nobody handed us a permission slip. We didn't ask for one.

"The new always happens against the overwhelming odds of statistical laws and their probability; the new therefore always appears in the guise of a miracle."

Hannah Arendt, The Human Condition, 1958.

We read Clark and Chalmers. We read Heidegger on tools becoming ready-to-hand. We read Edwin Hutchins on distributed cognition in aircraft cockpits. We understood the philosophical grounding. Then we stopped reading and started building, because the insight was clear and nobody in the industry was acting on it.

The separation between person and tool is an artifact of where software started. We are closing it.

What We Built

Planless — one workspace,
one intelligence layer.

Everything connected
Tasks, docs, sheets, email, calendar, forms. One data model. Orbis doesn't integrate with your tools. It is structurally inside them.
Sub-200ms voice AI
Custom architecture on top of the model. Not a wrapper. 200ms end-to-end. Laughs, pushes back, and holds full context while executing work in parallel.
Persistent Orbits
Background automations that run on judgment, not triggers. It watches deadlines, flags silence, builds agendas. They run until you stop them.
3-layer memory
It is built on RAG with three custom layers derived from neuroscience research on hippocampal indexing and memory consolidation. It knows you across time.
Agentic execution
Up to 200 parallel tasks with full thinking capability. It chains complex multi-step reasoning across your entire workspace without you managing the thread.
Google Verified
AES-256 at rest, TLS 1.3 in transit, OAuth 2.0. Real Gmail, Calendar, and Drive through Google's own infrastructure. Not a scraper, not an integration.
200ms
Voice latency
end-to-end
200+
Concurrent tasks
agentic execution
3
Memory layers
above RAG
The Intelligence Layer

Orbis. A brief history
of what changed.

Two versions. Not incremental. A different theory of what AI is for.

O7 · First Release

Full visibility. Real work. On demand.

Complete access to your tasks, docs, emails, calendar, and files. It did real work across all of it. You told it what to do and it did it, reliably and fast. But you had to tell it. It was still a tool in a box.

O7.5 · Current

Acts before you ask. Watches while you work.

It runs in the background. It watches your workspace and handles things before you think to ask. Missed follow-ups. Approaching deadlines. Empty meeting agendas. Stale tasks about to become a problem. It catches them.

Orbits, judgment-based and persistent
The Difference

Most agents live outside your tools. Orbis lives inside the data.

Other AI agents authenticate, scrape, and infer context from what they can read. Orbis doesn't infer. It knows. It was there when the task was created. It's been watching the deadline move. It read the email the moment it arrived. The difference is not speed. It is position.

Orbis Workspace — Live
Follow-up watcher · Andreessen
4 days no reply. Drafting follow-up
Running
Board deck · Q2 metrics pulled
Slides updated from live sheets
Done
Deadline watcher · Sprint end
3 tasks approaching. Flagging now
Running
Investor update · 12 VCs
Monday 9am · all sent
Done
Weekly agenda builder
Runs Sunday 8pm
Idle
7 Orbits running · 3 completed today
Voice AI

Sub-200ms. Emotional. Agentic. The first voice AI that acts while you talk.

We didn't bolt voice onto a product. We built a custom architecture directly on top of the model. Sub-200ms end-to-end latency. Orbis can laugh, express genuine uncertainty, push back on bad ideas, and simultaneously execute complex multi-step work in your workspace while the call is still live.

It has persistent memory. It remembers the conversation from last week. It knows your patterns. It gets sharper as your context accumulates.

Sub-200ms latency

Custom architecture, not a wrapper. The response feels like talking to a person, not a server.

Emotion + personality

Laughs, disagrees, and expresses uncertainty. Has opinions. Not afraid to use them.

Acts while you talk

Executes complex agentic tasks in your workspace while the conversation is still happening.

Persistent memory

Remembers every session. Learns how you specifically work. Gets sharper over time.

Architecture complete. Tuning in progress. Full release coming.
200ms
Latency
200+
Tasks/call
Memory
You: "The pitch was a mess. We need to rethink the deck completely."
Orbis: "Agree. Slide 3 buries the lead and you lose them before you even reach the product. I've seen the retention metrics. Let me restructure from there and pull the Q2 chart while we keep talking. Give me 40 seconds."
Under the Hood

We built our own architecture.
Not a wrapper. Not a prompt chain.

After a deep study of how biological memory actually works, we built three custom layers on top of RAG. The result: a system that remembers like a person, not like a database lookup.

01
RAG Foundation

Standard retrieval-augmented generation as the base layer. Fast semantic search across all workspace content. The floor, not the ceiling.

02
Associative Layer

Modeled on hippocampal indexing theory. Automatically connects last week's call to this morning's task by building associations across episodic memories.

03
Pattern Consolidation

Slow consolidation of behavioral patterns into persistent priors. Learns how you specifically work. Not a generic user model. A model of you. It runs offline, continuously.

04
Workspace Attention

A lightweight observation layer monitoring the live workspace for signals. Not polling, not if-then. Judgment-based prioritization modeled on attentional salience research.

05
Agentic Planner

Multi-step task decomposition with full thinking capability: up to 200 parallel subtasks. Thinking is available for text today. Voice reasoning pipeline is in active development.

06
Research Publication

Full architecture documentation, methodology, ablation results, and failure cases will be published Q4 2026. The full technical record, open.

The Market Just Proved the Thesis

OpenClaw (formerly Clawdbot) built the same idea. 145,000 GitHub stars in eight weeks. One problem: it lives outside your data.

In early 2026, an open-source project called Clawdbot went viral. It was renamed Moltbot, then OpenClaw. Peter Steinberger built an AI agent that does things: sends emails, books meetings, monitors tasks, acts without being asked. Andrej Karpathy called it "the most incredible sci-fi takeoff-adjacent thing" he'd seen. CNBC ran the story. IBM wrote analysis pieces. The world finally understood what we had been building toward.

But OpenClaw proved something else too: persistent agents built outside your data are a security disaster waiting to happen. Cisco's security team found prompt injection and data exfiltration in third-party skills. One maintainer warned users that "if you can't understand how to run a command line, this is far too dangerous." API keys dumped in plaintext. Credentials exposed. Agents sending emails nobody authorized.

The architectural difference

OpenClaw / External Agents
Authenticates into your tools from outside
Infers context by scraping what it can read
Holds your credentials and API keys directly
External process with prompt injection attack surface
Spends half its compute figuring out what's happening
Orbits: native, inside the data
Built into the data model, always there
Knows context because it generated it with you
Sandboxed. Credentials never leave the OAuth layer
Human-in-the-loop before any outbound action
Spends zero compute on context. Already home

Note: Our Orbits are not a response to OpenClaw. We were building this before Clawdbot existed. The viral moment in early 2026 simply confirmed what we already knew: the world wants persistent, proactive agents. The question is whether you build them in a way that is safe.

Built as One Since Day One

Every tool native. Not integrated. Structural.

Most AI products are built first, then AI is layered on. A chat window in the corner. A "generate with AI" button. We built Planless in reverse: the intelligence is structural, the tools share one data model, and Orbis was designed in from the first line of code.

Docs
Write investor updates, briefs, and SOPs while Orbis pulls live metrics from your sheets and full client history from your emails. You never start with a blank page. It already knows what you're working on.
Native
Sheets
A full spreadsheet engine with 500+ functions, multi-sheet support, real modeling depth. Ask Orbis to build a forecast or break down burn rate — it builds it right there. No exporting, no formulas to debug yourself.
Native
Email
Orbis writes every email knowing who you're talking to, what you've discussed, and what you're working on together. Sends from your actual Gmail with attachments, CC, BCC. Sounds like you wrote it. Because it knows you.
Native
Calendar & Scheduling
Set up a booking page with your branding and rules in one sentence. Someone books and it is not just a calendar event. It shows up in your workspace with full context on who they are and why you're meeting.
Native
Forms
Tell Orbis what you want to collect: feedback, lead qualification, client onboarding. It builds a form with your branding, one question at a time. Every response lands in your workspace, not in a separate dashboard.
Native
Voice
The first workspace where voice is native, not an add-on. Sub-200ms. Call Orbis like a co-founder: direct, honest, willing to push back. Brainstorm at midnight, debrief after a meeting, plan your week on a walk.
In development

All six tools share one data model. Change something in a doc and it updates in the task. Reference a sheet in an email and Orbis pulls live data. There are no sync issues because there is no syncing. It was always one thing.

Trust Model

You control how much Orbis acts autonomously. Not us.

The hardest problem with persistent AI agents is not capability: it is trust. OpenClaw sent emails nobody authorized. We built a trust model that puts every dial in your hands. The default is conservative: Orbis always asks. But you can open it up exactly as much as you want, scoped to a specific person, a category of action, or your entire contact list. You set the rules. Orbis follows them.

Default: always confirm
Out of the box, Orbis drafts and waits. Every email, every scheduling link, every form submission gets shown to you before it goes anywhere. Zero surprises. Change this at any time.
Trust a specific person
Tell Orbis it can act without asking when the action involves Sarah, or your co-founder, or your lawyer. It won't touch anyone else without confirmation. Scoped, not blanket.
Trust all your contacts
Open it up to everyone in your address book. Orbis can send follow-ups, reply to intros, and schedule meetings with anyone you already know. Still logging everything. Still fully reversible.
Trust a category of action
Allow Orbis to auto-send follow-ups but always confirm meeting invites. Auto-publish internal docs but gate external ones. The trust rules apply consistently across email, calendar, forms, and Orbits.
Trust settings · Orbis actions
Email follow-ups
All contacts
Auto
Email follow-ups
New contacts
Ask me
Meeting invites
Everyone
Ask me
Form submissions
Internal only
Auto
Scheduling links
sarah@acmecorp.com
Auto
Publish external docs
All
Ask me
Rules apply across email, calendar, forms, and all Orbits
Security & Trust

Verified, sandboxed, and yours alone.

OpenClaw's security critics were right about one thing: an AI agent with broad permissions needs a completely different trust architecture. We built ours from scratch, designed for the level of access Orbis requires.

Google Verified
We went through Google's OAuth security review process. Our Gmail, Calendar, and Drive integrations are verified. Not scraped, not proxy-routed. Google's infrastructure, Google's trust chain.
Verified
AES-256 + TLS 1.3
Data at rest is encrypted with AES-256. All transit uses TLS 1.3. Authentication is OAuth 2.0. We never store your passwords or raw credentials. Your data is opaque to us.
Encrypted
Sandboxed Architecture
Orbis operates inside a sandboxed execution environment. It cannot access file systems, shell commands, or external services outside of explicitly scoped integrations. The attack surface is minimal by design.
Sandboxed
Human-in-the-Loop
Every outbound action (email, calendar invite, published form, scheduling link) requires explicit user confirmation. Nothing leaves your workspace without your approval. Hard requirement, not a default setting.
Gated
No Credential Exposure
Unlike external agents that hold your API keys and OAuth tokens directly, Planless tokens are stored in a scoped credential vault. Orbis acts through the vault. It never sees raw credentials and cannot export them.
Isolated
Transparent Automation
Every Orbit, every background automation, is visible in your workspace. You can see what it's watching, what it has done, and what it will do next. Autonomous does not mean invisible.
Auditable
OpenClaw proved the demand. We built the version you can actually trust.
Persistent, proactive, native. Sandboxed from the ground up. The capability without the liability.
Google Verified AES-256 Sandboxed

The boundary
is yours to move.

Planless is live. Orbis is running. The research is being written. Come build with us, or get notified when we publish.

Try Planless Free Get Research Updates