Initialising KEN

Ken AI

Intelligence without the noise.

Think, create, code, research, and explore with Ken AI — a private AI assistant at ken-ai.tech, built for the way you work.

Introducing Ken AI

One intelligence.
Many possibilities.

KEN sits between the question you have and the work you need to finish. Ask it to reason through a decision, draft the thing you have been avoiding, read the file you never opened, or ship the feature you sketched on a napkin.

ThinkWriteCodeAnalyzeResearchCreateLearn

The model

Built for intelligence.
Optimized for speed.

Reasoning, context, vision, code, knowledge, and tools work as one system — so KEN keeps the whole problem in view instead of answering one piece at a time.

Model family

Choose the intelligence
that fits the task.

Switch models mid-conversation. KEN keeps the context, the files, and the thread.

KEN Fast

Everyday questions, quick drafts, and back-and-forth that should feel instant.

Speed
Depth
Context
Standard
Best for
Daily chat

Available

KEN Reason

Multi-step problems, tradeoffs, and work that benefits from thinking before answering.

Speed
Depth
Context
Extended
Best for
Hard problems

Available

KEN Vision

Screenshots, diagrams, scans, and photos — read, described, and reasoned about.

Speed
Depth
Context
Extended
Best for
Visual input

Available

KEN Code

Reads repositories, writes changes, explains failures, and stays in your stack.

Speed
Depth
Context
Long
Best for
Engineering

Available

The interface

A workspace, not a
text box.

ken.ai / workspace
KEN Code SELECTED
FilesWebRun
YOU
📎 checkout-service.ts · 14 KB
Our checkout retries are hammering the payment API. Can you add backoff and explain the fix?
KEN
The retry loop fires every 200 ms with no ceiling, so one slow response turns into a burst. Exponential backoff with jitter spreads the retries out and keeps a failing dependency from cascading.
checkout-service.tsdiff
async function retry(fn, attempts = 5) {
  for (let i = 0; i < attempts; i++) {
    try { return await fn(); }
    catch (err) {
      if (i === attempts - 1) throw err;
      // 200ms → 3.2s, ±30% jitter
      const wait = 200 * 2 ** i * (0.7 + Math.random() * 0.6);
      await sleep(wait);
    }
  }
}
YOU
Add a circuit breaker too.
KEN
Opening the breaker after five consecutive failures and probing every 30 seconds. Want me to write the tests as well?
Ask KEN anything
⌘K tools

Capabilities

More than a chatbot.

01

Think

Reason through difficult problems and hold the whole thread while doing it.

02

Create

Generate content and ideas in your voice, then refine them line by line.

03

Code

Build, debug, and understand software across the languages you actually use.

04

Research

Explore information across sources and synthesize what actually matters.

05

Analyze

Read documents, spreadsheets, and images, and turn them into an answer.

06

Create images

Turn a description into a visual, then iterate on it in the same conversation.

07

Remember

Keep the context that matters to you, and forget what you ask it to forget.

08

Use tools

Connect intelligence to action — search, run code, and call your own systems.

01 — BUILD

Build

Turn an idea into a working application — scaffolding, logic, tests, and the parts you would have looked up.

02 — RESEARCH

Research

Explore complex topics, compare what sources disagree on, and extract what matters to your decision.

03 — CREATE

Create

Transform rough thoughts into polished content that still sounds like you wrote it.

04 — ANALYZE

Analyze

Turn documents and data into insight — what changed, what is unusual, and what to do next.

05 — LEARN

Learn

Understand difficult concepts faster, at the depth you ask for, without the lecture.

Developers

Build with KEN.

One endpoint, streaming by default, with tool calling and structured outputs that behave the same across every model in the family.

REST APIPredictable schemas, versioned
SDKsTypeScript, Python, Go
StreamingToken-level server-sent events
Tool callingLet KEN call your functions
Structured outputSchema-validated JSON
Model accessFast, Reason, Vision, Code
stream.tsTSPYGO
import { Ken } from "@ken-ai/sdk";const ken = new Ken({ apiKey: process.env.KEN_API_KEY });const stream = await ken.messages.stream({  model: "ken-reason",  messages: [{ role: "user", content: "Summarise this contract." }],  tools: [fileReader, calculator],});for await (const event of stream) {  if (event.type === "text") process.stdout.write(event.text);}

Performance

Fast enough to feel instant.

Responses begin streaming as they are generated, so you read the first sentence while KEN is still writing the last one.

RequestYour prompt, files, and context leave the client.
ProcessingKEN plans, retrieves, and reasons over the task.
ResponseTokens stream back the moment they exist.

Privacy & security

Your intelligence stays yours.

Privacy by default

Your conversations are not used to train models unless you choose to share them.

Secure infrastructure

Encrypted in transit and at rest, with isolated workloads and audited access.

Data controls

Decide what KEN retains, per workspace, with retention settings you can change.

Conversation management

Export a thread, archive it, or delete it permanently — including its attachments.

Account security

SSO, two-factor authentication, device sessions, and role-based access for teams.

Clear terms

Plain-language policies describing what is stored, for how long, and why.

Pricing

Start with KEN.

Upgrade when you hit a limit, not before. Every plan includes the full model family.

Free

$0/ month

  • KEN Fast for everyday chat
  • File and image uploads
  • Conversation history
  • Web and mobile apps

Daily message limit

Start free

Pro

Recommended

$20/ month

  • Every model, including Reason and Code
  • Higher limits and priority capacity
  • Memory, tools, and image generation
  • Early access to new releases

Extended usage

Get Pro

Team

$30/ user / month

  • Everything in Pro
  • Shared projects and prompt library
  • Admin console, SSO, and roles
  • Central billing and usage reporting

Pooled workspace usage

Contact sales

Questions

Before you start.

Ken AI is an AI assistant and platform, available at ken-ai.tech. You can chat with Ken in the app, upload files and images for it to read, ask it to write or debug code, and call the same models from your own software through the API.

Ken is built around one workspace rather than a series of disconnected chats. Models, files, memory, and tools stay in the same thread, so switching from a quick answer to deep work does not mean starting over. The honest answer on quality is to try it on your own work.

Four in the KEN family: Fast for everyday conversation, Reason for complex problems, Vision for images and visual input, and Code for engineering work. You can switch between them mid-conversation from the model selector.

Yes. Attach documents, spreadsheets, code, and images, and Ken reads them in context. It can summarise, compare, extract structured data, or answer specific questions against what you uploaded.

KEN Code writes, reviews, and explains code across common languages and frameworks. It can work from an error message, a file you paste in, or a description of what you want to build.

On paid plans Ken can remember context you choose to keep — how you like your code formatted, what you are working on, who your team is. You can review what it remembers and delete any of it at any time.

Yes. The API exposes the full model family with streaming, tool calling, and structured outputs, with SDKs for TypeScript, Python, and Go.

There is. The Free plan includes KEN Fast, file uploads, and history, with a daily message limit. No card required to start.

The future of thinking starts here.

Meet KEN.