Managed AI operations
CodeAustral installs and manages a role-based AI employee that works 24/7 on incoming work through the business WhatsApp or Telegram you choose—without a second number. It can research, use a controlled browser, complete documents, draft content, coordinate calendars, operate approved tools, and hand sensitive or external actions to a person for approval.
Plans at USD 49, USD 99, or USD 199/month, plus annual billing options and no setup fee. Every plan includes all employee capabilities; USD 99 is the recommended private-VPS plan with higher AI usage.
Functional sandboxes
Choose a business context, enter a fictional workspace, and ask the operator to investigate, prepare, route, and explain work across a real multi-step flow. These public sandboxes are intentionally narrower than a configured commercial employee and do not enable every controlled-browser, research, document, content, calendar, or tool capability.
Every installation can customize the role, menu, sources, workflows, tools, approval boundaries, branding, and artifact templates around the business.
Patient continuity
Move from patient intake to schedule coordination and a reviewable clinical handoff without asking the assistant to diagnose or approve care.
Clinical decisions stay with the practitioner
Field-to-market coordination
Review field status, work orders, input inventory, weather windows, and commercial follow-up without asking the assistant to make agronomic or financial commitments.
Agronomic decisions, purchases, and commitments stay with people
Exception-led operations
Organize client documents, investigate reconciliation exceptions, prepare deadline queues, and route the work that needs an accountant.
Filing and sign-off stay with the accountant
Matter coordination
Qualify an intake, prepare a conflict-check packet, organize matter deadlines, and draft a client update for attorney review.
No legal advice; attorney review is explicit
Lead-to-viewing desk
Qualify a buyer or tenant, match requirements to available properties, coordinate a viewing, and prepare the next follow-up.
Offers and listing changes stay with the agent
Dispatch and job control
Turn a service request into a triaged job, a workable schedule, a technician brief, and a clear customer update.
Estimates and material schedule changes require review
One secure entry point, six professional contexts.
The demo uses fictional records and simulated writes. It cannot contact a real patient, client, court, grower, property lead, supplier, or field-service customer. A production role is configured around the customer’s approved operating boundaries.
Configurable 24/7 role
The installed employee works continuously on incoming work. Each capability is enabled only for the agreed role; sensitive or external actions remain permission-bound and human-approved. Public demos are intentionally narrower and may not expose every capability.
Navigate approved sites and complete defined browser steps inside the role’s policy.
Find, compare, and summarize public sources with links and traceability.
Complete approved forms, working documents, and recurring records.
Prepare messages, briefs, articles, proposals, and campaign drafts for review.
Coordinate availability, bookings, meetings, reminders, and follow-up.
Run reusable tools and multi-step workflows with visible permission gates.
From demo to a defined job
The best first role has repeatable work, reliable sources, clear permissions, and a person who owns approvals and exceptions. The role can span up to three agreed starting workflows.
Core role
Owns a defined queue across the systems your team already uses, prepares decisions, executes permitted actions, and routes exceptions with context.
Customer operations
Responds from approved knowledge, updates supported records, and hands conversations to the right person when policy or judgment is required.
Revenue operations
Researches, drafts, enriches, and follows up inside a bounded sales or marketing workflow without inventing claims or sending outside policy.
Control model
Every role has a written action policy, named owners, visible failure states, and a way to stop or revise the workflow.
The role receives only the accounts, data, and actions required for its agreed job. Credentials remain revocable.
Financial, contractual, publishing, account, and other sensitive actions can pause for a named human approver.
Ambiguous requests, conflicting data, and policy exceptions move to a person with the relevant context attached.
Runs, tool calls, errors, approvals, and outcomes are made visible enough to inspect and improve the workflow.
The runtime pattern is chosen around the company’s data, security, network, and infrastructure constraints.
Health, failures, costs, and workflow drift are monitored under an agreed operating and change process.
Systems boundary
The role reaches only the systems, records, and actions named in its operating policy.
Decision boundary
Sensitive or ambiguous work stops at the gate and moves intact to a named owner.
Runtime boundary
Custom connectors, health signals, and reviewable event records live inside the agreed deployment pattern.
Delivery
Define the queue, sources, decisions, outputs, owners, exceptions, and measurable acceptance tests.
Validate integrations, set least-privilege access, write action limits, and choose approval points.
Run real examples in a narrow scope, compare outcomes, and correct the workflow before permissions expand.
Deploy and monitor the role, then teach the method and reusable templates your team uses to create, duplicate, test, and adjust its own workflows.
Simple plans
All plans include the complete AI employee, approved workflows, browser and research capability, documents, content, calendars, human approvals, and ongoing management. Plans differ only by cloud isolation and AI usage capacity.
Shared
$49/month
Shared managed cloud
Standard AI usage
Private
$99/month
Private managed VPS
Higher AI usage
Private Plus
$199/month
Private managed VPS
Highest AI usage
No setup fee. Cancel future renewals at any time. Your current billing period remains available through its end.
A commercial role can be configured to work continuously on incoming tasks, use a controlled browser, research public sources, complete approved templates and documents, draft content, coordinate calendars or appointments, and operate approved tools and workflows. Capabilities are enabled only when needed; sensitive or external actions stay permission-bound and human-approved. Each public demo is intentionally narrower and does not imply that every capability is enabled there.
No. A conversational interface can be part of the system, but the deliverable is a managed operational role: a defined queue, approved knowledge, selected tools, action limits, human handoff, and monitoring.
Potentially. CodeAustral first validates supported access, permissions, data quality, rate limits, and recovery paths. An unsupported or unstable integration is narrowed, redesigned, or excluded before launch.
No. The employee works through the business channel you choose—your existing WhatsApp or Telegram. CodeAustral does not provide or require a second number. Channel access and policy fit are validated during onboarding.
Yes. CodeAustral configures up to three agreed initial workflows and teaches the operating method with reusable templates so your team can create, duplicate, test, and adjust its own workflows. Professional design, implementation, or integrations beyond that starting scope are quoted separately according to complexity.
Only inside the permissions approved for that role. Sensitive actions can require confirmation, and unknown or conflicting situations are escalated to a named owner.
Choose USD 49, USD 99, or USD 199 per month, with monthly or annual billing and no setup fee. Every plan includes the complete AI employee, approved workflows and tools, controlled browser and research capability, documents, content, calendars, human approvals, monitoring, and updates. USD 49 uses managed shared cloud; USD 99 and USD 199 use a private managed VPS and include progressively higher AI usage capacity.
The launch plan is set after CodeAustral validates the workflow, examples, access, approval owners, and integrations. A role using reliable existing access is faster to validate than one that needs a custom connector or data cleanup.
The USD 49 plan uses a managed shared cloud with logical customer separation. The USD 99 and USD 199 plans use a private managed VPS. In every plan, access and data boundaries are configured around the approved role and sources.
The operating policy defines whether the role stops, retries, requests approval, or hands the work to a named person. Failures and handoffs are logged so the workflow can be reviewed and corrected.
Before launch, CodeAustral and the workflow owner define acceptance measures for the chosen queue, such as completion quality, processing time, handoff rate, error rate, or backlog reduction. The role is reviewed against those measures rather than judged as a generic AI demo.
Each demo opens a fictional vertical workspace with role-specific navigation, synthetic records, and multi-step work. Ask the operator to investigate, prepare a draft, route an exception, or explain the evidence behind its answer.
No. Public demos use synthetic records and simulated writes. Sensitive actions stop at an approval step, and nothing is sent to a real patient, client, court, grower, property lead, supplier, or field-service customer.
Start with up to three workflows
Send your chosen business channel, up to three recurring workflows, current systems, approval points, and exception owner. The first reply will focus on feasibility and scope.
Reviewed by the CodeAustral team · Last reviewed .
FeedbackReviewed by the CodeAustral team · Last reviewed .
Feedback