KONST
AI Training CalculatorContact us
EN
繁體中文English
SERVICE | Enterprise AI adoption

Connect AI to business processes,
with usage governed throughout

Enterprise AI adoption connects models to business processes: pre-adoption assessment and consulting, custom application development and system integration, and usage governance after launch. Adoption runs on Horizon AI's ATP platform, where one project key connects to more than 70 open-source models, every usage record is attributed to its project, and billing is prepaid, deducted as used, with no monthly fee.

Horizon AI adoption service ATP Token
Single entry point via project key1keyEvery usage record attributed to a project
HORIZON AI | Platform governance and application rollout

Adopt applications first, or govern model usage first

AI ADOPTION

Horizon AI adoption service

Connect AI to daily workflows so it runs every day

From blueprint to real rollout

Go to Horizon AI adoption service
TOKEN AS A SERVICE

ATP Token

One project key connects to more than 70 open-source models

One key governs all model usage

Go to ATP Token
PROBLEM | Problems customers face

Three fragmentation risks of connecting AI models yourself

Several sets of connection logic must be maintained, AI costs cannot enter the cost allocation process, keys have no boundaries, and routing and failover have to be built in-house. None of this causes trouble in a demo, but once the system goes live and external users appear, it becomes a hidden risk.

SOLUTION | KONST's approach

One SDK, one bill, one key

One project key connects to each provider's models, every usage record is attributed to its project, and permissions are bounded by the key. Adoption proceeds in three stages: assessment, custom development, and governance after launch.

  1. Problem

    Scattered models: switching models means changing code

    Each provider has its own SDK, keys, and model list, so switching models requires code changes.

    Solution

    One SDK: one key reaches every provider's models

    The interface is compatible with major providers' SDKs, and in most cases you only need to change base_url and the key.

  2. Problem

    Scattered billing: cost never reaches the allocation process

    Token usage and costs for each project are spread across each provider's console and cannot be combined into one report.

    Solution

    One bill: every usage record attributed to a project

    Token usage and costs for each project are collected in one report that can go straight into the allocation process.

  3. Problem

    Scattered keys: no boundary on permissions or usage

    Keys are scattered, permission boundaries are unclear, and usage has no cap, which becomes a risk once external users appear.

    Solution

    One key: permissions bounded by the key

    Each key can call only the models it is authorized for, and the usage cap and permission scope are set key by key.

PROCESS | Workflow

One key covers everything from model access to Agent deployment

An event triggers the flow, then a model is selected, tools are called, results are checked and retried, and finally the flow is deployed as an external-facing Agent. Every step runs under the same metering and governance, so no Agent can use up credits in the background without anyone knowing.

  1. 01Event trigger

    Review which steps in the process suit a model, confirm that data access is secure, then decide how to implement it.

  2. 02Select a model and provider

    The routing layer picks a source among multiple upstream suppliers based on availability and cost.

  3. 03Tool calls and execution

    Call external tools and internal systems as the workflow defines, obtaining results step by step.

  4. 04Result check and retry

    Check whether the output meets the validation criteria, and retry or hand off to a person according to the rules if it does not.

  5. 05Deploy as an Agent

    After the workflow passes validation, it is deployed as an external-facing Agent, and usage and costs stay under the same governance.

EVIDENCE | Track record and data

Model access, metering, and governance,
in a single entry point

Each provider has its own SDK and model list, and billing is spread across separate consoles. ATP connects to more than 70 open-source models with one key, bringing access, metering, and governance into one entry point.

70+Models available on the platform and still expanding
1keySingle entry point via project key
0monthly feeTop up first, deduct as you use
base_urlCompatible with major SDKs, connect by changing settings
PRICE | Pricing and fees

Credit-based billing: top up first, deduct as you use

Credits are deducted based on input and output token counts and each model's rate. You top up first and usage is deducted as you go, with no monthly fee. Usage is visible in real time and attributed to projects, and high-volume users can ask about an annual contract.

Billing

DescriptionInput + output token count multiplied by each model's rate, then deducted from credits

Payment

DescriptionTop up first and usage is deducted as you go, with no monthly fee, and credits do not expire

Visibility

DescriptionUsage and spend are visible in real time and can be attributed to projects

High volume

DescriptionEnterprises with larger usage can negotiate capacity and annual contracts separately

FAQ | Common questions

Common questions on adoption assessment, access, billing, and governance

Where should an enterprise start when adopting AI applications?
Take stock of your processes, not models. First find steps with clear inputs and outputs, secure access to data, and results that can be verified. A step is ready to start only when all three are in place. With a unified entry point, models can be swapped at any time, while redesigning a process costs far more.
How much code has to change to move an existing application to the platform?
The interface is compatible with the SDKs of major providers, and in most cases you only need to change base_url and the key. The platform sits on top of existing call methods and does not require the application to adopt a new protocol.
How is usage cost allocated to each department?
The token count and the credits deducted for every call can be traced back to a project, workspace, and organization. Credits are allocated from the top down and usage is attributed from the bottom up, so each department's cost comes directly from the platform without separate estimates.
How does the platform charge?
Billing is credit-based. The input plus output token count is multiplied by each model's rate and deducted from credits. You top up first and usage is deducted as you go, with no monthly fee, and credits do not expire. Enterprises with larger usage can ask about enterprise discount programs.
If a single provider has a problem, are calls interrupted?
The routing layer switches between multiple upstream suppliers. When one source has a problem, other sources take over automatically, with no reliance on manual switching.
Does adoption always require a full custom build?
Not necessarily. After the assessment you can choose an existing application, custom development, or a mix of both. The scope of system integration depends on your existing information architecture.
What ready-made enterprise AI applications does KONST offer?
Through Horizon AI you can already adopt four types of applications: an enterprise knowledge base, real-time multilingual AI meeting translation, document AI, and an AI factory digital twin. Custom development is also available to fit industry needs.
RELATED | Related services

Services often evaluated together

GPU Cloud ServicesActivated on sign-up, billed per minuteGlobal compute sites16 deliverable data centersKonstra AIData center infrastructure and bare-metal compute delivery

Ready to connect AI to your business processes?

Start with a process inventory and plan your company's adoption path.

Contact our team