AI productsIn development

TurbotaAI

AI SaaS product · Customer app + operations console

A product a customer can open, try, subscribe to and keep using on their own — and that someone inside the business can actually run: see activation and subscriptions, check system health, look up one account for support, and have every privileged action written to an audit log.

TurbotaAI — Landing — the offer, the trial counter and the entry point into a conversation — Landing — the offer, the trial counter and the entry point into a conversation

Context

An AI companion product built in public. People talk to an AI by text, voice or video in their own language, behind an account with a trial, a subscription and their own consent and data controls — and the business runs it from an operations console.

The business problem

An AI demo is not a business. Most AI projects stop at the chat window: the account, the trial limit, the payment, the consent record, the error states and the people who actually operate the thing are all left as "later". That is exactly where a product either becomes sellable or stalls.

The solution

How it works

1Visitor
2Try the AI (limited trial questions)
3Register and onboard
4Choose permissions and start a conversation
5Hit the trial limit → paywall
6Subscribe or redeem a promo code
7Manage access, consent and data in the profile

What was delivered

A multi-modal AI conversation — text, voice and video
Accounts with sign-in, registration and a guided onboarding
Trial access with a request allowance, then unlimited access on subscription
A monthly subscription with a paywall, promo-code redemption and self-serve access management
Per-capability permission toggles, so a customer chooses what the product may use
Consent history and a data & account area covering AI processing, memory, notifications and deletion
An operations console: activation, engagement, capabilities, subscriptions, AI usage and cost, today/push/system health, single-account support lookup, access grants and an audit log
Role-based access control on the admin area, enforced server-side
Three locales — Ukrainian, Russian and English — with locale-aware pricing
Designed at 1440 desktop and 390 mobile across every route
Deliberate loading, empty, partial, locked, offline and error states

Key capabilities

User accounts & onboardingSubscription, paywall & promo codesUsage limits & entitlementsRole-based access controlAdmin & operations consoleAudit loggingConsent & data controlsMulti-modal AI (text / voice / video)Multilingual (uk / ru / en)Responsive 1440 / 390

Visual walkthrough

Screens from the real product.

TurbotaAI — Pricing — monthly plan, trial allowance, promo redemption and access management in one place — Pricing — monthly plan, trial allowance, promo redemption and access management in one place

TurbotaAI — Three locales at 390px — the same product in Ukrainian, Russian and English, with locale-aware pricing — Three locales at 390px — the same product in Ukrainian, Russian and English, with locale-aware pricing

TurbotaAI — Operations console — activation, engagement, subscriptions, AI usage and cost, system health, support lookup and an audit log — Operations console — activation, engagement, subscriptions, AI usage and cost, system health, support lookup and an audit log

TurbotaAI — Role-based access — the admin area refuses accounts the server does not recognise as administrators — Role-based access — the admin area refuses accounts the server does not recognise as administrators

TurbotaAI — Error state — a failed load explains itself and offers a retry, instead of showing an empty screen — Error state — a failed load explains itself and offers a retry, instead of showing an empty screen

TurbotaAI — Mobile pricing at 390px — Mobile pricing at 390px

Technical implementation

Next.jsAI assistant integrationAuth & user profilesSubscription billingAdmin consoleScreenshot/QA harness

What exists today

The screens below are the real product at a specific build (V1, commit fc6ed09) — captured by the project's own screenshot harness, not mocked up
Route coverage was captured systematically: every route at 1440 desktop and 390 mobile, in three locales
The admin area, audit log, consent history and error states are implemented, not planned
In active development: the subscription screen currently reports that payments are unavailable, and the store badges on the landing page are links, not evidence of a published app

Outcome

A complete AI SaaS is standing: the customer journey from first trial question to paid subscription and self-managed data, and the operator side that runs it — activation, subscriptions, AI cost, system health, support lookup and an append-only audit trail. Development continues against a working product, not a specification.