Features Use Cases Cloud Developers Security Open Source Mission Pricing Get Early Access ↗
Sansten · Cognitive Layer
NINAI
Cognitive Operating System Beta

Your organisation already knows the answer. Ninai finds it.

Persistent memory, causal reasoning, and autonomous action — across every silo, every system, every timezone.

Get Early Access ↗ See how it works
Early access · No credit card required · Set up in minutes
The Problem

Every organisation leaks knowledge. Every day.

A decision gets made in a Slack thread. A customer insight lives in one person's head. A past incident and its fix disappear when someone leaves.

Your teams restart from zero on problems they've already solved. Your AI tools answer questions but forget everything the moment the session ends.

The knowledge is there. The intelligence isn't.

What Ninai Is

The memory layer your organisation has always needed.

Not a chatbot. Not a dashboard. A cognitive layer that runs alongside your organisation — remembering, reasoning, and acting on your behalf.

01

Remembers

Persistent memory across all teams and tools. Nothing gets lost when the tab closes or someone leaves. Your organisation's knowledge compounds over time.

02

Reasons

Causal, not just correlational. Ninai understands why things happen — not just what happened. Root cause in minutes, not days.

03

Acts

Autonomous action engine with full audit trail. Ninai can trigger workflows, run playbooks, and escalate — not just surface information.

Capabilities

What Ninai actually does

80+ cognitive capabilities across memory, reasoning, and action — built for teams where forgetting is expensive.

Memory · 01 / 03

Persistent Cross-Silo Memory

Stores, connects, and surfaces institutional knowledge across all departments in a shared memory graph. When engineering fixes an issue, support knows the same day — not six months later when it becomes tribal knowledge.

  • New hires get full institutional context from day one
  • Knowledge survives turnover, reorgs, and tab closes
  • Cross-department signals connected automatically
NINAI ENG OPS SALES SUPPORT PRODUCT
Memory · 02 / 03

Episodic Memory & Timeline Reasoning

Groups events into episodes — not just individual data points. Ninai understands that the three Slack messages, the spike in error logs, and the customer complaint last Tuesday were all part of the same incident.

  • "What changed before the Q3 churn spike?" — answered from memory
  • Incidents automatically linked to their predecessors
  • Trends visible across months without building a report
Episode 1 Episode 2 Now Nov 14 Dec 02 Today same pattern
Memory · 03 / 03

Memory Decay & Consolidation

Not all knowledge stays relevant forever. Ninai tracks the age and usage of every memory, automatically deprioritising stale information and surfacing fresh signals. Overlapping memories are consolidated — the clearest version always wins.

  • Stale procedures don't compete with current guidance
  • Duplicate knowledge is merged, sources preserved
  • Working memory kept separate from long-term knowledge
2yr ago 1yr ago 6mo ago 1mo ago Current Relevance
Reasoning · 01 / 03

Causal Reasoning Engine

Traces root causes backward and projects impact forward. Ninai builds a causal graph of your organisation — which systems affect which outcomes, which decisions led to which results. Root cause in minutes, not days.

  • Do-calculus: interventional reasoning for policy decisions
  • Counterfactual: "if we hadn't shipped that, what happens?"
  • Causal chain visible before you act
Root Cause Service A Service B Latency ↑ Errors ↑ Churn ↑
Reasoning · 02 / 03

Predictive World State Monitor

Continuously models the current state of your organisation and projects it forward. When the projected state diverges from what's expected, Ninai flags the discrepancy — before it becomes a crisis. No dashboards to watch.

  • Early warning for delivery risks and capacity problems
  • Projected impact of decisions before they're made
  • Continuous background monitoring, zero effort
! Expected Now → Baseline
Reasoning · 03 / 03

Anomaly Detection

Learns the normal operating envelope of your systems and teams, then detects meaningful deviations. Anomalies are ranked by severity, cross-referenced against causal context, and routed to the right person — not just dumped into an alert channel.

  • Severity ranking with causal cross-reference
  • Routed to the right person, not a noisy channel
  • Reduces alert fatigue — only what matters surfaces
Anomaly P1 Time
Action · 01 / 03

Autonomous Action Engine

When a situation matches a known playbook and confidence is above threshold, Ninai acts — triggers webhooks, creates tickets, sends alerts. Every action is audited, explained, and reversible. When confidence is low, it escalates to a human.

  • Configurable confidence thresholds per action type
  • All actions logged with full reasoning trace
  • Rollback capability for reversible actions
NINAI Webhook Create Ticket Slack Alert Human Review confidence: 94% → act
Action · 02 / 03

Proactive Intelligence Push

You don't have to ask. When Ninai detects a signal that's relevant to someone who probably doesn't know about it yet, it pushes the information proactively. The right person gets the right context — not whoever happened to be watching the right dashboard.

  • Org-wide attention model weights signals by who needs them
  • No dashboards to watch — Ninai watches for you
  • Push by Slack, email, or API webhook
Signal NINAI CTO PM CSM
Action · 03 / 03

Human Review Queue

Not everything should be automated. When Ninai's confidence is below threshold, it queues the decision for human review — with its full reasoning, evidence, and proposed action already written up. Every review decision feeds back into Ninai's learning.

  • Reviewers see evidence, reasoning, and proposed action
  • Approve, modify, or reject — all choices are learned from
  • Trust is built incrementally, autonomy expands over time
Review Required · Confidence 61% Proposed: Escalate ticket #4421 Approve Modify Reject feeds back into Ninai's learning
How It Works

Up and running in minutes. Valuable in days.

01

Connect

Point Ninai at your existing systems. No rip-and-replace. Ninai listens and learns from the data you already have.

Slack · Teams · Jira · Linear · GitHub · Confluence · Notion · PostgreSQL · REST APIs · Email
02

Ninai Builds the Picture

Over the first few days, Ninai builds your organisation's knowledge graph, identifies recurring patterns, and maps the causal relationships between teams, systems, and outcomes.

03

Your Team Gets Smarter

Ninai pushes relevant signals proactively, answers questions with full context, executes low-risk actions autonomously, and gets better with every decision your team makes.

Who It's For

Built for teams where forgetting is expensive.

Engineering

Incidents That Don't Repeat

Episodic memory links current incidents to past resolutions. Before the on-call engineer opens a terminal, Ninai has already surfaced what happened last time and what fixed it.

"This pattern matches the incident on 2025-11-14. The fix was [X]. The root cause was [Y] in service [Z]."

Product

Customer Insights That Reach the Roadmap

When a customer pattern emerges in support, it surfaces in product — not through another Jira ticket that gets lost, but through Ninai's proactive intelligence push.

14 support tickets in two weeks, same underlying cause, filed in different language. Ninai connected them before sprint planning.

Operations

Playbooks That Run Themselves

Ninai crystallises resolution patterns from past cases into reusable playbooks — without asking anyone to write documentation. When a similar issue arises, the playbook surfaces immediately.

Executive

A Briefing That Actually Means Something

Not a dashboard full of numbers. A narrative synthesis of what's happening — what's on track, what's at risk, what needs your attention, and why.

Compliance

Audit Prep in an Afternoon

Every decision, inference, and action is logged immutably with full reasoning attached. GDPR data exports are one-click. Audit prep becomes an afternoon, not a sprint.

Small Teams

Knowledge Infrastructure Without the Overhead

Ninai gives a team of 8 the knowledge infrastructure of a 200-person organisation — built automatically from the Slack messages, commits, and decisions you're already making.

Enterprise

Security is not a feature. It's the foundation.

Built for organisations where data is sensitive by definition. Every architectural decision was made with security as a primary constraint.

Isolation

Row-Level Security

PostgreSQL RLS enforced at the database engine — not the application layer. One tenant's data cannot reach another, even with an application bug.

Identity

SSO / SAML / OIDC

Integrates with Okta, Auth0, Azure AD, and Google Workspace. Ninai never sees your corporate credentials. MFA required for admin roles.

Network

Cloud Armor WAF

OWASP Top 10 protection, rate limiting on all endpoints, TLS 1.3 for all external traffic. All internal services are VPC-internal only.

Audit

Full Audit Trail

Every decision, inference, and autonomous action is logged immutably with reasoning attached. Required for compliance. Essential for trust.

Compliance

GDPR · SOC 2 · HIPAA

GDPR data export and erasure built in. SOC 2 Type II available on Professional and Enterprise plans. HIPAA-eligible configurations available.

Residency

Data Residency

Your data stays in the region you choose — US, EU (Belgium, Zurich), Asia (Singapore, Tokyo), Canada. Enforced at the infrastructure level.

Open Source

Transparent by design.

Ninai's deployment tooling, connector SDK, and API specification are open source. The cognitive intelligence engine is proprietary — it's what your organisation pays for, and what we invest in continuously.

Open · Apache 2.0

  • Connector SDK — build connectors to your tools
  • Helm chart — self-hosted deployment config
  • Terraform modules — GCP, AWS, Azure stubs
  • OpenAPI specification — full API reference
  • Python & TypeScript SDKs — client libraries
  • Demo notebooks — showcase and examples

Proprietary · Commercial Licence

  • Cognitive agents — 80+ capabilities, the core product
  • Causal Reasoning Engine — the primary differentiator
  • Memory architecture — sleep cycle, consolidation, decay
  • Theory of Mind & Org Attention Model
  • Continuous Learning Pipeline
  • Strategy Evolution Service

Enterprise self-managed customers receive full source under a commercial licence for auditability and deployment on their own infrastructure.

Contribute

Help build the cognitive layer.

Ninai's open components are built in the open. If you're a developer who thinks about memory systems, causal reasoning, or enterprise tooling — there's meaningful work to do here.

Build a connector

Connect Ninai to a tool that isn't supported yet. The Connector SDK is Apache 2.0 — pick a source (a SaaS tool, a database, an API) and add it to the ecosystem.

ninai-connect on GitHub ↗

Improve the SDKs

The Python and TypeScript client SDKs are open. Add missing endpoints, improve error handling, write tests, improve docs — all contributions welcome.

ninai-sdk on GitHub ↗

Write demo notebooks

Show what Ninai can do with real-world scenarios. Notebooks that demonstrate memory, reasoning, or action capabilities become part of the official examples library.

ninai-demos on GitHub ↗

Report issues

Found a bug, a gap in the API, or behaviour that doesn't match the docs? Open an issue. Every report makes the product better — for everyone running it.

Open an issue ↗
Get started in 3 steps
1
Fork the repo

Find the component you want to work on under github.com/sansten and fork it.

2
Read CONTRIBUTING.md

Each repo has a contributing guide covering setup, code style, and the PR process. Takes 5 minutes to read.

3
Open a pull request

All PRs get reviewed. First-time contributors welcome — no contribution is too small.

View on GitHub Questions? Talk to us
Cloud

Ninai Cloud — managed for you.

Sign up and start in minutes. No infrastructure to provision, no Kubernetes to babysit. Ninai runs on Google Cloud in the region you choose — we handle the ops, you get the intelligence.

Fully managed

Ninai handles all infrastructure: database, vector store, Redis, backups, and zero-downtime upgrades. You connect your tools and use the product.

Up in minutes

Create your workspace, connect your first integration, and Ninai starts building your knowledge graph. No engineers required to get started.

Always on

99.9% uptime SLA. Hourly backups with 30-day retention. Ninai watches your organisation around the clock — including 3 AM on a Sunday.

Your region, your data

Choose from US, EU (Belgium or Zurich), Asia (Singapore or Tokyo), or Canada. Your data — database, vector store, and backups — stays in that region.

Enterprise-grade security

Row-level tenant isolation, SSO/SAML, Cloud Armor WAF, TLS everywhere, and a full immutable audit trail — on every plan, not just Enterprise.

GDPR compliant

Data export and right-to-erasure built in. DPA available on request. Ninai never uses your data to train models or improve the service without your consent.

Connect in minutes
SlackMicrosoft TeamsJiraLinear GitHubConfluenceNotionGoogle Docs PostgreSQLREST APIsEmail (IMAP)+ more via SDK
Ready to start?
Early access is free. Your data carries over when paid plans launch.
Create your workspace ↗
Developers

Start building in under 10 minutes.

Ninai has a full REST API, Python and TypeScript SDKs, and a Connector SDK for building custom integrations. Everything you need to embed Ninai's cognitive capabilities into your own stack.

1

Create a workspace

Sign up at admin.ninai.sansten.com and create your organisation. Takes 2 minutes.

2

Get your API key

From the admin console → Settings → API Keys. Keys are scoped to your organisation and optionally to a specific role.

# Settings → API Keys → Create
NINAI_API_KEY=nai_live_xxxxxxxxxxxx
3

Install the SDK

# Python
pip install ninai-sdk

# TypeScript / Node
npm install @ninai/sdk
4

Write your first memory

from ninai import NinaiClient

client = NinaiClient(api_key="nai_live_xxx")

client.memory.store(
  content="API redesign decision: moved to REST from GraphQL for client compatibility",
  tags=["engineering", "api", "decisions"]
)
5

Query with context

result = client.memory.query(
  question="Why did we move away from GraphQL?"
)

# Returns: answer + source memories + confidence score
print(result.answer)
Pricing

Pricing is coming soon.

Beta · Early Access

We're in beta. Access is free while we are.

We're working with early customers to shape the product before publishing pricing. Sign up now — early access accounts keep their data when paid plans launch.

Get Early Access ↗
Starter
Coming soon
Professional
Coming soon
Enterprise
CustomContact us

Enterprise enquiries: support@sansten.com

Get Started

Your organisation is already generating the knowledge.

Ninai makes it work.

Get Early Access ↗ Explore on GitHub
Where we're heading

Most AI tools are tools.
Ninai is meant to be something else.

The capabilities in Ninai — episodic memory, causal reasoning, theory of mind, world state modelling, goal decomposition, continuous learning — are not features assembled for enterprise convenience. They are the building blocks of a system that understands the way a person does.

A system that remembers what your organisation has lived through. That reasons about why things happen, not just what happened. That holds goals across time. That knows what it doesn't know, and says so.

We are early. The version you're using today is a fraction of what this becomes. But every organisation that runs Ninai now is contributing to that picture — their decisions, their patterns, their knowledge compounding into something that gets meaningfully smarter over time.

This is the direction. We thought you should know.

01
Memory
Episodic, semantic, and procedural memory — persistent across time, organisations, and systems.
02
Reasoning
Causal, counterfactual, and predictive reasoning — understanding consequence, not just correlation.
03
Agency
Goal decomposition, autonomous action, and self-correction — with human oversight as a first principle.
04
Understanding
Theory of mind, attention modelling, and semantic awareness — knowing who knows what, and what it means.
— Sansten, 2026