Claude
GPT · routed
Gemini
✓Guardrails
⌕Web search
{ }Python
EnterpriseEnterprise AI conversational agent →

Secure multi-model
AI conversational
platform.

ChatLite is an enterprise-first AI workspace: self-hostable, customisable, and connected to your systems, data feeds, and agents. It also gives individuals and power users a clean way to use OpenAI, Claude, Gemini, Mistral, and more.

ChatLite secure workspaceGuardrails on
New chat ⌘K
Models
Anthropic Claude
OpenAI
Google Gemini
Mistral
DeepSeek
xAI Grok
SSO · RBAC · Audit logs
Never used for training
Claude · cited answers
What changed in the EU AI Act timeline this year?
Searched the web · 14 sources · 3.2s
Obligations for general-purpose AI models now apply, with high-risk system rules phasing in later. Providers must publish training-data summaries and keep technical documentation up to date. [1] [2]
1Official Journal of the EUeur-lex.europa.eu
2Commission AI Office updatedigital-strategy.ec.europa.eu
Summarise what that means for our product team.
Document model sources, add a transparency notice to AI features, and assign an owner for incident reporting before Q2.
Where are we on the Northwind onboarding?
Searched workspace · Drive, Slack, Jira · 9 matches
Kickoff is done and SSO is configured. Two blockers remain: the data import ticket is waiting on their IT team, and the security review is due Friday. [1] [2] [3]
Onboarding plan.docx[1] Drive
#northwind-rollout[2] Slack
NW-142 Data import[3] Jira
Draft a status update for the account channel.
Northwind update: SSO live, kickoff complete. Waiting on IT for data import (NW-142). Security review due Friday.
Search the web…Search your workspace…Web searchWorkspace↑
Routes to every leading model
OpenAI
Anthropic Claude
Google Gemini
Mistral
DeepSeek
xAI Grok
[ 01 ]Platform

Minimal interface. Serious capability.

The chat experience stays light. The platform underneath gives teams the model choice, security, and control they need.

01

Every model, one workspace

Switch between OpenAI, Anthropic Claude, Google Gemini, Mistral, DeepSeek, and xAI Grok, including reasoning and multimodal models, without changing tools or losing context.

02

Governed by default

Centralized access, guardrails, and admin controls help teams adopt AI without losing oversight.

03

Web search, built in

Answers grounded in live sources, with citations your team can verify before sharing.

04

Files and data

Summarize documents, analyze spreadsheets, and read ~30 formats (PDF, DOCX, XLSX, CSV, and code) or pull files straight from Google Drive and OneDrive.

05

Code and execution

Run sandboxed Python, generate charts, and call Wolfram Alpha for exact computation, orchestrated automatically without leaving the chat.

06

Clean by design

A minimal interface that gets out of the way so the work stays in focus.

07

Image generation

Produce on-brand visuals, diagrams, and concepts from a single prompt, no separate tool needed.

08

Team workspaces

Share prompts, threads, and presets so teams build on each other's best work safely.

09

Self-host and customize

Run the entire platform in your own cloud or data center, customised to your workflows, branding, and policies.

10

Custom agents and feeds

Connect internal systems, live data feeds, and your own agents into one governed plane alongside chat and research.

[ 02 ]Capabilities

Built to remember, optimise, create, and act.

Six capabilities that turn a chat window into a working platform, each governed by the same access, guardrails, and audit trail.

Memory · Personalisation

Remembers what matters. Forgets on request.

ChatLite learns your preferences, tone, projects, and recurring context, so every conversation starts where the last one left off. Memory is visible, editable, and scoped to you or your team.

+Preferences, writing style, and working context carried across chats
+View, edit, pause, or delete any memory at any time
+Personal and team-scoped memory, governed by admins
+Never used to train third-party models
Memory · Personal4 items
◆Prefers concise answers with a bullet summary
◆Writes in British English
◆Current project: Q3 vendor review
◆Works in GBP, fiscal year starts April
Stored for your account onlyManage memory →
Token economics · Neuro-symbolic optimisation

Spend tokens only where intelligence is needed.

A neuro-symbolic optimiser pairs deterministic rules with neural routing. It decides whether a task needs a model at all, which model fits, and how much context to send, so cost drops without losing quality.

+Symbolic rules handle structured, repeatable steps at zero token cost
+Routes each task to the smallest capable model
+Context compression and caching across turns
+Exact maths and data handed to Python and Wolfram Alpha
Optimiser traceillustrative
Task · Convert 14 invoices to a CSV
Symbolic · rule: structured extraction0 tok
Neural · small model + Python toolcached context
Naive
41.2k tok
Optimised
9.8k tok
Saved
−76%
Artifacts

Turn answers into work you can ship.

Responses become living artifacts: documents, code, charts, diagrams, and tables that open beside the chat. Edit them, iterate with the model, and keep every version.

+Live preview alongside the conversation
+Version history for every artifact
+Share with teammates or export to the format you need
+Code and charts generated from sandboxed Python
PreviewCodeq3-summary.md · v3
Q3 vendor review
EditVersionsShareExport
Prompt library · Builder

Your best prompts, built once and shared everywhere.

Build structured prompts with roles, context, constraints, and output formats, then save them to personal or team libraries with variables anyone can fill in.

+Guided builder with variables and output formats
+Personal and team libraries with permissions
+Versioning and usage insights per prompt
+Run any saved prompt from a slash command
Prompt builder · /vendor-briefTeam library
RoleProcurement analyst
Context{vendor_name}, contract renewal in {month}
ConstraintsCite sources · max 300 words
OutputTable + three recommendations
v4 · used 212 timesRun prompt
AI agents · Co-work

Agents that work alongside you, not around you.

Hand off multi-step work: gather files, research, draft, and prepare outputs. Agents use your connected tools and pause for approval before anything leaves the workspace.

+Plan, execute, and report on multi-step tasks
+Use connected files, search, and code tools
+Human approval before external actions
+Bring your own custom agents onto the same plane
Agent · Prepare board packrunning
✓Pull Q3 figures from Google Drivedone
✓Research three competitor updatesdone
◐Draft summary and chartsworking
!Share with finance teamApprove
Every external action waits for your approval.
Integrations

Bring your tools and files into the chat.

Connect the apps your team already works in and query them in plain language, with access you can revoke at any time.

+Google Drive and Microsoft OneDrive, available today
+GitHub, Slack, Notion, Jira, Salesforce, and more coming soon
+Query files, code, tickets, and records from the chat
+Custom integrations and live enterprise data feeds
Connected apps2 live · more soon
Google Drive
OneDrive
GitHub
soon
Slack
soon
Notion
soon
+Custom connector
[ 03 ]Enterprise search

Search everything your company knows. From one prompt.

Ask once and ChatLite searches across your connected apps, files, chats, tickets, code, and records, then answers with citations back to the source. Results respect each person's existing permissions.

+One query across Drive, OneDrive, Slack, GitHub, Jira, Notion, and more
+Permission-aware: people only see what they can already access
+Cited answers that link back to the original file or thread
+Blend company knowledge with live web results when you need both
Explore integrations →
⌕What did we agree with Northwind on the renewal?⌘K
All sources · 6FilesChatsTicketsCRM
A two-year renewal at the current rate, with a 5% uplift in year two and SSO added at no cost. Legal signed off on the 14th. [1] [2] [3]
Northwind – Renewal terms v4.docx[1] Google Drive · Legal / Contracts
#northwind-account: "Confirmed 5% in year two"[2] Slack · 3 days ago
Northwind Logistics – Renewal 2027 opportunity[3] Salesforce · Closed won
Permission-aware · only sources you can access
[ 04 ]For teams

A practical AI assistant for every department.

All use cases →
[ 05 ]Deep Research

Turn open questions into source-backed reports.

ChatLite investigates a topic end to end, searches the web, weighs sources, compares claims, and returns a structured brief your team can review, share, or build on.

+Plans the research path before searching
+Pulls from live web and your internal context
+Citations and source quality next to every claim
+Findings, risks, and recommended next steps
Researching: AI market signals · Q2
✓ Drafting search plandone
✓ Scanning 24 sourcesdone
◐ Comparing citationsrunning
○ Writing summaryqueued
Tech press
8 reports
Filings
4 docs
Analyst notes
3 briefs
Watercolor of a quiet office at sunriseMinimalist ceramic vase product shotAbstract geometric pastel compositionFuturistic glass architecture at dusk
PromptA quiet office at sunrise, watercolor
[ 06 ]Image Generation

Generate images right inside the chat.

Describe what you need and ChatLite produces it, illustrations, concepts, product mockups, social assets, using the image models you already trust, in the same workspace as your text work.

+Seven providers: DALL·E, Imagen, Flux, Stable Diffusion, Grok, Ideogram, Recraft
+Iterate on the same prompt without losing history
+Reference uploaded images for style and composition
+Export at the size and format you need
[ 07 ]Guardrails

Safe AI, on by default.

ChatLite enforces your policies on every prompt and response, blocking sensitive data leaks, prompt injection, and content that violates your rules before it ever reaches an end user.

+Detect and redact PII, secrets, and credentials
+Block prompt injection and jailbreak attempts
+Content policies tuned per team or department
+Full audit trail of every decision
Policy check · Marketing prompt
Draft a launch email referencing the customer list in the attached file.
✓No prompt injection
!PII detected · 24 emails redacted
✓Within content policy
✓Model output allowed
Allowed with redactions
[ 08 ]Plans

Start small. Scale into your organization.

Use ChatLite personally, or bring it to your organization with admin controls and guided rollout.

Compare plans in detail →
Base
Freeforever

For individuals getting started with AI.

Start free
✓Access to standard models
✓Standard daily limits
✓Deep research included
✓Image generation included
✓Web search with citations
✓File analysis and uploads
✓Single-user workspace
AscentMost popular
$9per month

For power users who want more of every capability.

Get Ascent
Everything in Base, plus:
✓All leading models: OpenAI, Claude, Gemini, Mistral, DeepSeek, and Grok
✓Higher daily limits
✓More deep research
✓More image generation
✓Google Drive and OneDrive connectors
✓Sandboxed code execution and charts
✓Presentation generation
Zenith
$49per month

For professionals running heavy research workflows.

Get Zenith
Everything in Ascent, plus:
✓Generous daily limits
✓Maximum deep research
✓Maximum image generation
✓Multi-model deep research roles
✓Priority access to new models and features
✓Built for demanding, high-volume workflows
Organization
$49per seat / month

For teams that need a shared, governed workspace.

Get Organization
Everything in Zenith, plus:
✓Shared team workspace, billed per seat
✓Generous per-member limits
✓Invite and manage members
✓Roles and admin controls
✓Organization-wide model allowlist
✓Guardrails on every prompt and response
✓Organization usage analytics
[ 09 ]Enterprise

Self-host it. Customise it. Connect your agents.

ChatLite is enterprise-first. Deploy the whole platform in your own environment, integrate your internal systems and live data feeds, and bring custom agents into the same governed workspace as chat, research, and image generation.

+Fully self-hostable in your cloud or data center
+Custom integrations and enterprise data feeds
+Bring your own agents onto one governed plane
+SSO, RBAC, guardrails, and audit logs throughout
Models
Data feeds
Custom agents
ChatLite · one governed plane
ChatResearchImagesTools
SSO · RBAC · Guardrails · Audit logs
[ 10 ]FAQ

Frequently asked questions

Still curious? Talk to the team.

View all FAQs →

Which AI models does ChatLite support?

ChatLite routes to leading models from OpenAI, Anthropic (Claude), Google (Gemini), Mistral, DeepSeek, and xAI (Grok), including reasoning and multimodal models, from one workspace, so you can switch models without losing context.

Is ChatLite secure for business and enterprise use?

Yes. ChatLite provides SSO, multi-factor authentication, device and session control, encryption at rest, role-based access control, and guardrails that screen every prompt and response for PII, secrets, and prompt injection.

Does ChatLite support SSO and MFA?

Yes. Organizations can sign in through OAuth-based single sign-on, and every account can enable TOTP-based multi-factor authentication with backup codes.

Does ChatLite use my data to train AI models?

No. Your prompts, files, and outputs are processed only to provide the service and are never used to train third-party foundation models unless you explicitly opt in.

How much does ChatLite cost?

ChatLite has five plans: a free Base plan for individuals, Ascent at $9 per month and Zenith at $49 per month for power users and heavy research workflows, an Organization plan at $49 per seat per month for teams of up to 100 users, and Enterprise from $100K per year for 100+ users, with SSO, audit logs, and private deployment. See the pricing page for current rates.

Bring secure AI into everyday work.

Contact ChatLite →Start free