Agentic chat with thumbnails
Talk to the AI like any modern assistant, but every conversation is saved as a thumbnail, so context never gets lost.
More AI · Smarter spend · Enterprise-grade
More AI for less budget.
Priced per token. Tighter walls.
Why pay $20, $30, $200 a month for a chat that just talks? Xuda gives you five text models billed per token, from a nano default that handles the everyday to frontier reasoning when you ask for it, so you pay for what you actually use.
Five text models to pick from
Pay per token, not per seat
No-code agents
On-prem ready
Xuda · agentic chat
Give me a sales forecast for next month using our internal data.
B
AI
On it. Routing to your Sales Forecast agent and pulling from
orders and pipeline. tool · query_table(orders, pipeline)
Forecast · May
Projected revenue: $418,200 · +12.4% MoM
Top driver: Enterprise tier · 38 active deals
$0.05
Per million input tokens, default model
5
Text models, priced per token
0
Lines of code to build an agent
100%
On-prem option for enterprises
Watch
See the AI workspace in action.
Watch how one prompt puts agents to work, and your whole workspace stays in sync, in one place.
Priced per token
Stop overpaying. The premium model is overkill for "summarize this".
Most AI subscriptions hand you one model and charge premium for every prompt, easy or hard. Xuda prices per token across five text models you pick from in the composer, from GPT-5 Nano standing as the default at five cents a million input tokens, up to GPT-5.6 Sol when the work earns it. Pick per request, pay for the tokens you actually burn.
Your prompts
"Reformat this list as a table."
"Summarize today's emails from Acme."
"What is in invoice #4421?"
"Plan a 3-month migration to AWS."
"Generate a sales forecast model."
Xuda
Catalogue
Catalogue
Five text models
you pick from
you pick from
You pick the lane
Fast lane · the default
$0.05/1M
GPT-5 Nano answers unless you pick another, and the project assistant steps up one model on an action or a long prompt. Pennies per million tokens. Reformatting, summaries, lookups, classification, basic Q&A.
Premium lane · you pick it
$2 to $30/1M
Frontier reasoning + tool-using models. One pick in the composer. Multi-step planning, code gen, deep analysis.
Per-token pricing across the whole catalogue, not a flat premium seat.
You pick per request. The project assistant steps up one model on its own, never into the premium lane.
The Workspace
Everything an AI workspace needs, arranged as one system.
Stop hopping between an inbox, a CRM, a chatbot, an app launcher, and a project tool. Xuda gives you one command center, chat, apps, contacts, agents, and profiles, in five tabs that map to the five things you actually do all day.
No-code agents in minutes
Describe what the agent does, pick its tools, and ship. A virtual workforce without writing a line of code.
Contacts as one timeline
Email, chat, and app notifications collapse into one conversation view per person. Stop hunting for context.
Profiles + pooled credits
Split your credit pool across work, client, and personal contexts. Switch with one click; everything follows.
A unified command center
Five tabs. Every conversation, app, and contact. One scroll away.
Open a chat, jump to an app, message a contact, hand work to an agent, and switch profiles when you need to wear a different hat. The workspace stays out of the way. Your work stays in flow.
- Conversations saved as thumbnails, share with a link.
- Apps you build, install, or receive, one tab.
- Contacts unify email, chat, and notifications per person.
- Agents you create can be shared with your team or published to the Marketplace.
AI Chats12
Apps7
Contacts214
AI Agents9
Profiles3
Q2 forecast review
SalesRouted to Sales Forecast agent · pulled from orders + pipeline · $418k projected
AI Today · 14:32
Onboarding template
HRGenerated a checklist + email draft, scheduled for new hires starting Monday.
AI Today · 11:08
Invoice #4421
FinanceParsed PDF, extracted line items, ready for AP review.
AI Yesterday
Migration plan
Eng3-month AWS migration plan with phases, dependencies, and risk matrix.
AI Yesterday
Agentic AI Chatbot
A chat that doesn't just talk. It ships.
One prompt. The chat pulls your private data, fires the workflow, drafts the email, generates the code, and packages an answer you can act on. Every conversation is saved as a thumbnail you can revisit or share with one click.
Standard chat, agentic muscle
Talk to the AI like any modern assistant. Behind the scenes, agents handle the work: queries, automations, code generation, file operations.
Conversation thumbnails
Every chat is saved as a thumbnail so revisiting yesterday's analysis is one click away. Share a thumbnail and your teammate gets the full context.
Tap private data, safely
Ask the chat to use your internal tables, files, and integrations. Permissions and scope live with the agent, not with the conversation.
Generate code, ship apps
Ask for a form, a landing page, a small tool, or a full report. The chatbot spins up apps directly into your workspace and studio.
Trigger workflows
Send group emails, dispatch messages, fire off complex functions in connected apps, all by chatting. The chat is the universal remote.
Group conversations
Pull a colleague into a chat, hand off a thread, or co-pilot with your team. Conversations behave like a modern messaging app, with AI inside.
Use case · sales forecast
From a one-line question to a number you can act on.
Ask the Xuda AI chatbot for a sales forecast using internal data. Behind the scenes, the system automatically tasks an AI agent to securely retrieve that information from your database: no SQL, no glue code.
- 01
You ask in plain language
"Forecast next month's revenue from our pipeline."
- 02
Chatbot routes to the right agent
A Sales Forecast agent is selected based on the request and your private data scope.
- 03
Agent queries your data
It calls fetch_table_data on orders and pipeline, applies your filters, and aggregates.
- 04
You get a usable answer
A clean number, the breakdown, and a link to dig in. The whole conversation is saved as a thumbnail.
Xuda AI Agents
Build a virtual workforce. In minutes. With no code.
Agents are the multitasking workers behind your prompt. Define one in minutes, no code, no complex config. Pull data from a database, draft the weekly report, ship an application, answer an inbox. Share with your team or list in the Xuda Marketplace.
🗂️
Add Data to Table
Drop in a CSV, a paragraph, or a screenshot. The agent maps fields, validates types, and writes rows to the right table.
save_docvalidatefind_query
📊
Fetch Table Data
Ask in plain language. The agent translates the request into a Mango query and returns the records, scoped to your permissions.
find_queryget_table_doc
📨
Auto Response Triage
Watches incoming messages, classifies intent, and either drafts a reply for review or sends an answer when it is confident.
classifydraftsend
📈
Weekly Report
Pulls metrics from the apps you pick and writes the narrative, so the weekly update is one ask instead of a lost afternoon.
aggregaterenderdeliver
🛠️
App Builder
Describe what you need. The agent scaffolds tables, programs, and UI directly in your studio, ready for you to refine.
create_docbuild_program
🧭
Triage Agent
A meta-agent that reads the request and routes it to the right specialist agent, so users never have to pick the right tool.
routehandoff
Contacts & Conversations
Stop hunting for context. Find the person, find everything.
Xuda re-organizes communication around the people you talk to, not the apps you use. Connect your email in one click and every message, chat, and app notification collapses into one conversation view per contact.
One-click email sync
Connect your inbox in one click. Xuda groups every message by contact, so you stop hunting for context across folders, labels, and threads.
One timeline per person
Email, chat, and app notifications collapse into a single conversation view per contact. The interaction history of a relationship, in one scroll.
Act from the chatbot
Take action on contacts directly from the AI chatbot. Send group emails, dispatch messages, and trigger specific functions in connected apps.
Auto-replies that respect tone
Let an agent watch a thread and reply when you are away. You set the boundaries, the tone, and the moments that should always reach a human.
Credit management
One pool of credits.
Every dollar accounted for.
Per-seat AI subscriptions punish you twice: full price for a heavy user who barely chats, full price again for the intern who only summarizes documents. Xuda flips the model. Pool credits at the workspace level, then split them across account profiles, separate buckets for work, a client engagement, marketing, support, your weekend side project.
Switch profiles in one click and the chat, agents, and contacts come along. Admins see clean, real-time usage broken down by profile and by user. No surprise bills, no end-of-month spreadsheet archeology, no rewarding waste. Credits renew every 30 days.
May · workspace credits
12,400 / 20,000
Work7,820 · 4 users
Personal2,940 · 1 users
Side project1,640 · 2 users
Renews in 11 days3 profiles · 7 users
Enterprise-grade security
The perfect secure environment for serious enterprises.
Most AI tools ask you to send your data to their cloud and trust the marketing copy. Xuda assumes the data is sensitive: scope every agent, guard every input and output, and give the customer a switch to run the whole platform inside their own walls. CISOs love it because it's not a checkbox, it's the architecture.
Per-agent permissions Input & output guardrails Encrypted in transit & at rest Profile isolation Per-agent activity trail On-prem option
Scoped tools, scoped data
Every agent runs with an explicit toolset and explicit data scope. It cannot read tables, hit endpoints, or call functions you have not approved. Least privilege, baked in.
Input & output guardrails
Tripwire-grade guardrails inspect what enters and leaves an agent. Prompts that try to exfiltrate data are caught early. Outputs that violate policy are blocked.
Encryption end to end
Data is encrypted in transit with modern TLS and at rest. Secrets and credentials live in a secured vault: never in prompts, transcripts, or logs.
Profile isolation
Switch from your work profile to your personal profile, and the data, agents, and conversations switch with you. No accidental crossover, no leaked decks.
Audit-ready logs
Every agent run and every credit charge writes a usage row: the model that answered, the profile, the member, the token counts. Plus a per-agent trail of what changed.
Run it on your own metal
Need maximum control? Deploy Xuda standalone on your own premises. Same workspace, same agents, same studio. Your data never leaves your perimeter.
The Xuda Platform
Five modules. One platform. Zero glue code.
Build, deploy, manage, and monetize without leaving the platform. The Agentic AI Workspace, an AI-powered no-code studio, a deployment system that ships from one form, robust hosting, and a marketplace for what you make.
01 · Agentic AI Workspace
Your daily command center. Chat, apps, contacts, and agents in one place.
02 · No-Code Studio
An AI-powered visual editor that turns prompts into apps, tables, and programs.
03 · Deployment System
Ship from prompt to production on one form, with predictable cost.
04 · Hosting System
A robust, distributed hosting tier ready to absorb growth without surprises.
05 · Marketplace
Distribute or monetize the apps and plugins you build. A second income for makers.
The honest comparison
Same money. Massively more capability.
ChatGPT, Claude, Gemini, and Grok are great chat products. They are not unified workspaces. They bill a flat seat instead of the tokens you actually burn, they don't unify your contacts, they can't be deployed inside your perimeter, and you can't sell what you build on top of them.
Xuda You | ChatGPT | Claude | Gemini | Grok | |
|---|---|---|---|---|---|
| Entry paid plan (individual / mo) | Free tier · paid from less | $20 (Plus) | $20 (Pro) | $20 (Advanced) | $30 (SuperGrok) |
| Premium / power tier | Pro tier < competitors | $200 (Pro) | $100-$200 (Max) | ~$30 (Workspace) | $300 (Heavy) |
| Pay per token, not per seat | |||||
| No-code AI agents | ~ | ~ | |||
| Agents that act on your private data | ~ | ~ | ~ | ||
| Build & ship full apps from chat | ~ | ~ | |||
| Unified contact & inbox view | |||||
| Pooled credits with per-profile budgeting | |||||
| Marketplace to sell what you build | ~ | ||||
| On-prem deployment | ~ | ||||
| Custom enterprise SLA | ~ |
For enterprises
Run Xuda on your own premises. Keep your control.
For enterprises requiring maximum control, Xuda can be deployed as a standalone installation on your own premises. To further support business-critical operations, Xuda offers multiple custom support plans and SLAs tailored to your needs.
- On-prem deployment with the same workspace, agents, and studio.
- Custom support plans matched to your business hours and priorities.
- SLAs tailored to business-critical operations and uptime targets.
- Direct access to engineering, with named support contacts.
Common questions
Answers before you have to ask.
Stop wrestling tools. Start shipping outcomes.
A workspace where chat acts, agents work, contacts are unified, credits are clear, and your data stays yours. Try it free, see your first hour saved by tomorrow.
The era of fragmented tools is over.