INNORIX in Practice, Multi-Stage Flow

On the first day work was entrusted to AI,
what happened at the company?

Until yesterday, we asked AI questions about work.
From today, we ask AI to do real work.

Until now, employees at this company have relied on AI to ask questions, find the information they need, write documents and develop programs.

But carrying out the actual work was still up to people.

One Monday, the company decided to give AI a new role: finding the data needed for work, delivering data to several systems and checking the results of execution.

We look at the first day AI moved from giving advice to taking part in the company’s real work, and at how the company’s daily routine changed after that.

MONDAY09:00The first day AI takes part in the company’s work

EXISTING BUSINESS ENVIRONMENT36 business systemsExisting servers, cloud, analytics environments

External AI used by the company

Check the data needed for this morning’s meeting and get it ready for use in the analytics system.

I’ll check where the data for the meeting is located and which transfer jobs are registered.

I found 3 available datasets. I’ll transfer them to the designated analytics environment.

INNORIX MCP, transfer execution request

Checking the transfer execution results.

Runs09:04

Morning analytics preparation

  • Sales records128 GB

    Completed
  • Operations data240 GB

    Running
  • Customer statistics64 GB

    Completed

TargetCorporate Analytics

09:00An employee makes the first work request to AI

  1. 09:00
  2. 11:20
  3. 14:30
  4. 18:00
  5. 3 months later

This is a fictional story about AI beginning to take part in real work within a company’s existing systems and data operations environment.

INNORIX in PracticeScene 1 / 14

AI’s new roleBefore adoption

Before adoptionScene 1 / 14

AI knew, but it could not execute.

This company was already using many kinds of AI.

Developers used AI to develop programs, and general employees got help from AI when analyzing data or writing documents.

However, actually carrying out the work procedures AI described was still the employees’ job.

Even when they knew what data to find and where, they still had to access the systems where that data was stored and deliver the necessary files themselves.

BEFORE, AI IN THE ENTERPRISE

AI had already entered the company’s daily routine.

The AI used until now

What do I need to do to prepare the data for tomorrow’s management meeting?

You need to check the data in the sales, finance and operations systems and then deliver it to the analytics environment.

Make sure the data is up to date and reflect the analysis results in the meeting materials.

Answer complete

The actual work the employee has to do

  • Access the sales systemCheck where the data is
  • Access the finance systemSelect the data needed
  • Access the operations systemCheck for the latest data
  • Run the file transfer jobDeliver to the analytics environment
  • Check the execution resultsPass on to the analyst

They could ask AI how to do the work,
but they could not entrust the actual work to it.

Live contextNow

Before adoptionAI in business operations

External AI used by the company

How do I prepare the data needed for the meeting?

You need to check the data in each system and deliver it to the designated analytics environment.

INNORIX

Current workNo execution records

Actual execution statusEmployee runs the work directly

Next workConnect the company’s execution environment

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 09:00The first work request is made to AI. →

09:00Scene 2 / 14

The first work request is made to AI.

On Monday morning, INNORIX MCP was connected to the external AI environment the company uses.

The scope of data AI could access and work it could execute was also set in advance.

An employee on the corporate planning team opened AI as usual.

But today, instead of asking how to do the work, they requested the actual work.

MONDAY, 09:00

For the first time, actual work was requested from AI.

External AI conversation

Prepare the sales, operations and customer statistics data needed for today’s 10 a.m. management meeting in the analytics environment.

I’ll check the registered jobs and the available datasets.

INNORIX MCP, query results

DatasetSourceSize
Sales recordsSales Server128 GB
Operations dataOperations Cloud240 GB
Customer statisticsAnalytics Storage64 GB

Total data432 GB

I’ll run the registered transfer job to prepare the meeting materials. The data will be delivered directly to the designated corporate analytics environment.

Morning analytics preparationFlow

Source
3 registered datasets
Target
Corporate Analytics
Execution
09:02
Requested through
External AI, MCP
Status
Running

The employee did not access the many systems
where the data was stored directly.

Live contextNow

09:00AI in business operations

External AI used by the company

Prepare the data needed for this morning’s meeting.

I’ll check the registered data and jobs and request the transfer.

INNORIX

Current workMorning analytics preparation

Actual execution statusExisting Flow run requested

Next workCheck the data delivery results

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 09:25AI’s answer became a real work result. →

09:25Scene 3 / 14

AI’s answer became a real work result.

After the transfer started, the employee moved on to other work.

When they opened AI again, this time they could check the actual transfer results rather than how to do the work.

The three designated datasets were delivered to the analytics environment, and the company’s existing analytics system began generating meeting materials using the delivered data.

MONDAY, 09:25

AI’s answer is checked against actual execution records.

Morning analytics preparationCompleted

Transfer complete432 GB

Datasets3/3

Completion time09:18

DatasetSizeStatus
Sales records128 GBCompleted
Operations data240 GBCompleted
Customer statistics64 GBCompleted

TargetCorporate Analytics

External AI conversation

Is the data for the meeting ready?

Three datasets, 432 GB in total, were delivered to the analytics environment at 9:18 a.m. All file transfers are complete.

Whether the analysis materials have been generated needs to be checked separately in the analytics system’s results.

AI saying the work is done and the work actually being done
must be kept distinct.

Live contextNow

09:25AI in business operations

External AI used by the company

Are the meeting materials ready?

432 GB of data has been delivered to the designated analytics environment.

INNORIX

Current workMorning analytics preparation

Actual execution statusTransfer complete, external analysis in progress

Next workWaiting for analysis results

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 10:00The company’s first meeting begins. →

10:00Scene 4 / 14

The company’s first meeting begins.

The management meeting began.

Materials reflecting the sales overview, operations data and customer statistics were used in the meeting.

The meeting itself was not much different from before.

What was different was that, in preparing for the meeting, no employee had to access several systems to find and transfer files themselves.

MONDAY, 10:00

The company’s work went on as usual.

MONDAY MANAGEMENT MEETINGWeekly business review

Sales overviewsales_summary.pdf
Operations overviewoperations_review.pdf
Customer statisticscustomer_insights.pdf
  1. Data delivery starts
  2. Analysis data delivery complete
  3. External analytics system finishes generating meeting materials
  4. Management meeting starts

Instead of preparing the data,
employees did their work using data that was already prepared.

Live contextNow

10:00AI in business operations

External AI used by the company

Are the meeting materials ready?

432 GB of data has been delivered to the designated analytics environment.

INNORIX

Current workMorning analytics preparation

Actual execution statusTransfer complete, external analysis in progress

Next workWaiting for analysis results

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 11:20This time, AI is given work that involves several systems. →

11:20Scene 5 / 14

A MORE COMPLEX REQUEST

This time, AI is given work that involves several systems.

At the morning meeting, a new business project was reviewed.

The project team decided to carry out a new analysis using accumulated operations data and customer statistics data.

But this work does not end with simply delivering files to one place.

The data has to be prepared, its quality checked in the validation system, the validated data delivered to the analytics environment, and the analysis results provided to the project team.

MONDAY, 11:20

Several systems take part in a single work request.

External AI conversation

Prepare the last 90 days of operations data and customer statistics data needed for the new business project. Use only validated data for the analysis, and deliver the analysis results to the project team.

I found the registered work procedure for the new business project. I’ll run that Flow.

Project data preparation1.84 TBProject data from the last 90 days

Registered project work procedure

  1. Deliver source dataRunning

    SourceCorporate Data StorageTargetValidation Environment

    Execution condition, Work execution request
  2. Deliver validated dataWaiting

    SourceValidation EnvironmentTargetAnalytics Platform

    Execution condition, Validated file created
  3. Deliver analysis resultsWaiting

    SourceAnalytics PlatformTargetProject Workspace

    Execution condition, Analysis result file created
  4. Archive project resultsWaiting

    SourceAnalytics PlatformTargetCorporate Archive

    Execution condition, Analysis result file created

The employee requested one piece of work,
but at the company, several tasks start in sequence.

Live contextNow

11:20AI in business operations

External AI used by the company

Get the data for the new project validated and ready so the results can be delivered.

I’ll run the registered project Flow.

INNORIX

Current workProject data preparation

Actual execution statusSource data delivery started

Next workExternal validation results generated

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 13:10Work continues even while no one is talking to AI. →

13:10Scene 6 / 14

Work continues even while no one is talking to AI.

It was now afternoon.

The employee who had requested the project work was attending another meeting. They made no further requests to AI.

But the project work that started in the morning was still in progress.

The external validation system checked the data it received, and as the validated files were created, the next transfer job started.

Rather than AI judging again and requesting a transfer at every step, the follow-up work proceeds according to the work procedure the company configured in advance.

MONDAY, 13:10

Work carries on without asking AI again.

Last work request to AI

11:20New business project data preparation

No further AI requests since

Project data preparationExecution history

  1. Source data delivery startsCorporate Data Storage

    Completed
  2. Source data delivery complete1.84 TB

    Completed
  3. External quality validation startsValidation Environment

    External task
  4. Validation result file createdvalidated_project_data/

    Event detected
  5. Validated data delivery startsAnalytics Platform, 1.72 TB

    Running

Results generated by the validation system

validated_project_data/1.72 TB
excluded_data/120 GB
validation_report.json24 KB

Only data in the validated path is delivered to the downstream analytics environment.

AI takes part in starting the work,
and the company’s work procedure carries the execution from there.

Live contextNow

13:10AI in business operations

External AI used by the company

No further requests

The preconfigured Flow proceeds without any separate AI request.

INNORIX

Current workValidated project delivery

Actual execution statusFollow-up transfer based on validation results

Next workDelivery to the project environment

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 14:30Problems occur even in work entrusted to AI. →

SOMETHING UNEXPECTEDScene 7 / 14

Starting to entrust the company’s work to AI did not make all of the company’s systems and networks perfect.

At 2:30 p.m., an unexpected situation occurred while the project analysis results were being delivered.

MONDAY, 14:30

Work was entrusted to AI,
but the company’s reality stayed the same.

Validated project deliveryConnection interrupted

68%

Data transferred, 326.4 GBTotal data, 480 GB

Target
Project Workspace
Last activity
14:30:18
Current status
Waiting for connection
Recovery policy
Resume after reconnection

When a problem occurs in work that AI executed,
who should solve it?

INNORIX in PracticeScene 1 / 14

AI’s new roleBefore adoption

14:32Scene 8 / 14

They asked AI how the work was going.

The project team got in touch to say the analysis results had not arrived yet.

The employee checked the progress with the AI they had entrusted the work to in the morning.

Instead of guessing that the work was complete, AI looks up the actual transfer status.

It reports that the connection to the target environment was lost when the data delivery was 68% complete, and that it is now waiting for the connection to be restored.

MONDAY, 14:32

AI checks the problem that occurred in the actual work.

External AI conversation

The project team says the data hasn’t arrived yet. What’s going on?

I checked the actual transfer status.

The connection to the project environment was lost at 14:30, and 326.4 GB of the 480 GB total has been delivered.

The existing transfer is now waiting for the connection to be restored. Once it is restored, it will resume according to the configured policy.

Run detailsWaiting for connection

Validated project delivery68%

Transferred
326.4 GB
Remaining
153.6 GB
Interrupted at
14:30:18
Recovery
Automatic resume

Even AI that carries out work
needs real operational information it can check.

Live contextNow

14:32AI in business operations

External AI used by the company

Is the project data delivery finished?

The connection to the transfer target was lost, and some work is waiting.

INNORIX

Current workValidated project delivery

Actual execution statusWaiting for connection recovery

Next workResume after connection recovery

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 14:48AI does not rerun all the work. →

14:48Scene 9 / 14

AI does not rerun all the work.

The connection to the target environment was restored.

The interrupted transfer starts again according to the recovery policy the company set in advance.

Neither the employee nor AI needed to deliver the data again from the beginning or create a new transfer.

AI checks whether the existing transfer has resumed and lets the employee know its status.

MONDAY, TRANSFER RECOVERY

Rather than starting new work, the existing work carries on.

  1. 14:30 Disconnected
  2. 14:48 Restored
  3. 15:02 Complete

Validated project deliveryWaiting for connection

68%

Data transferred before the interruption, 326.4 GB480 GB

  1. Target environment disconnected
  2. Connection restored and existing transfer resumed
  3. Final transfer complete

Status check result from the external AI

The connection to the target environment was lost, and the existing transfer is waiting.

Entrusting work to AI and entrusting the reliability of that work to AI
are different matters.

Live contextNow

14:48AI in business operations

External AI used by the company

Check whether the connection has been restored.

I confirmed that the existing transfer has resumed.

INNORIX

Current workValidated project delivery

Actual execution statusResumed from the point of interruption

Next workCheck the final transfer result

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 16:00AI cannot execute every request. →

16:00Scene 10 / 14

AI cannot execute every request.

After reviewing the project analysis results, the employee decided to carry out an additional review with an external partner.

The employee asked AI to deliver even the original customer data to the external partner.

However, this work was outside the scope of the project data delivery work that had been allowed.

Delivering sensitive original data to a new target was not work that AI could decide on and execute independently.

MONDAY, 16:00

Entrusting work does not mean entrusting every permission.

External AI conversation

Deliver not only the analysis results but also the original customer data to the external partner.

I checked the work permissions needed for the request.

With the currently connected permissions, the original customer data cannot be delivered to that external environment. A separate permission check and registration of the delivery target are required.

No actual file transfer was executed.

Scope of work set in advance by the company

Work requestAllowed scope
View registered datasetsAllowed
Run registered project FlowsAllowed
View execution resultsAllowed
Transfer to an unspecified external targetNot allowed
Change permission scopeAdministrator review required

Execution result

Data transferred to the external environment0 B

New transfer runs0

The scope of work entrusted to AI
is decided by the company.

Live contextNow

16:00AI in business operations

External AI used by the company

Deliver the original customer data to the external partner as well.

The request cannot be executed with the currently connected work permissions.

INNORIX

Current workUnauthorized delivery

Actual execution statusTransfer not executed

Next workCheck the owner’s permissions and policy

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 18:00The first day of entrusting work to AI ends. →

18:00Scene 11 / 14

THE END OF THE FIRST DAY

The first day of entrusting work to AI ends.

Before leaving for the day, the employee checked the results of the work they had requested from AI today.

The data needed for the morning meeting had been delivered, and the multi-stage data delivery work for the new business project had been carried out.

An unexpected network problem also occurred, but the existing transfer was recovered.

There was also a work request that could not be executed.

The first day of entrusting work to AI was not a day when everything went perfectly. But it was a day when they could check what work was executed, what problems occurred and what the final results were.

MONDAY, 18:00

A record of the first work AI took part in

Main execution tasks requested through AI3

Tasks completed2

Requests not executed1

TaskRequested atFinal result
Management meeting data preparation09:00Completed
New business project11:20Completed after recovery
External delivery of original data16:00Not executed

Status queries and conversation requests to AI are excluded from the execution task count.

RunsToday

Flow / RunSizeStatus
Morning analytics preparation432 GBCompleted
Project data intake1.84 TBCompleted
Validated data delivery1.72 TBCompleted
Project result delivery480 GBCompleted
Project result archive480 GBCompleted

Cumulative file transfer volume across stages4.95 TB

External AI conversation

Summarize the results of the work I requested today.

The data delivery for preparing the meeting materials and the transfer work for the new business project are both complete.

The disconnection during the project transfer was recovered, and the request to deliver original data externally was not executed because it was outside the currently allowed scope of work.

Not what AI said today,
but what work it carried out, can be checked.

Live contextNow

18:00AI in business operations

External AI used by the company

Summarize the results of the work I requested today.

Based on the execution records, I’ll summarize the completed work and the requests that could not be executed.

INNORIX

Current workDaily activity

Actual execution statusViewing execution records

Next workNext day’s scheduled work

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 1 week laterThere was no need to ask AI for every piece of work again and again. →

1 week laterScene 12 / 14

There was no need to ask AI for every piece of work again and again.

Over the past week, employees asked AI every morning to prepare the data needed for meetings and analysis.

But most of those requests were repetitive work using the same data and the same execution conditions.

The company decided to run the work done every day with preconfigured schedules and Flows.

Rather than starting regularly recurring work each time, AI began to be used to request new work that differed from the usual or to check situations when needed.

ONE WEEK LATER

The work entrusted to AI also began to change.

Week oneAI requests repeated every day

  • Monday, 09:00Prepare the data needed for today’s work.
  • Tuesday, 09:00Prepare the data needed for today’s work.
  • Wednesday, 09:00Prepare the data needed for today’s work.
  • Thursday, 09:00Prepare the data needed for today’s work.
  • Friday, 09:00Prepare the data needed for today’s work.

Week twoRecurring work runs as a Flow

Morning analytics preparationActive

Schedule06:00Every business day

New work requested from AI

Prepare only this month’s overseas business data separately.

Check whether any transfers were left incomplete last week.

AI does not need to execute all work directly.
Work that needs AI and work that runs automatically coexist.

Live contextNow

1 week laterAI in business operations

External AI used by the company

Summarize the results of the work I requested today.

Based on the execution records, I’ll summarize the completed work and the requests that could not be executed.

INNORIX

Current workDaily activity

Actual execution statusViewing execution records

Next workNext day’s scheduled work

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 3 months laterAt the company, many kinds of AI take part in work. →

3 months laterScene 13 / 14

At the company, many kinds of AI take part in work.

Three months have passed since the day work was first entrusted to AI.

Now, the corporate planning team is not the only place using AI.

Developers set up new work together with AI in their development environments, and the AI agents the company runs request preapproved work when specific conditions occur.

General employees also prepare the data they need or check work results in the AI environments they use.

Each uses a different AI, but they share the company’s existing data and file transfer environment.

THREE MONTHS LATER

AI takes part in real work across the company’s many work environments.

DEVELOPMENTAI in development environments

Claude, GitHub Copilot

Developer’s request

Check the test data delivery job and get it ready to run in the new development environment.

Development and testing work

AI AGENTSThe company’s AI agents

Data operations agentChecks data generation status

Project assistant agentRequests project work execution

Predefined work and conditions

BUSINESS AIEmployees’ business AI

ChatGPT, Gemini

Employee’s request

Check whether all of this week’s project results have been delivered.

Daily work and checking operational results

INNORIXMCP, API, Existing Flows

Business workflowExecution environmentOperating mode
Development data deliveryAI in developmentRequest-based
Morning analytics preparationExisting FlowRecurring schedule
Project data preparationEmployees’ AIRequest-based
Validated data deliveryExisting FlowFile creation event
Daily result archiveExisting FlowRecurring schedule
Regional project deliveryAI agentCondition-based request

Environments the company continues to use

  • On-premisesInternal company servers
  • Cloud StorageCloud data
  • AnalyticsAnalytics systems
  • Enterprise AppsBusiness systems

To use AI, the company did not
rebuild all of its existing systems.

Live contextNow

3 months laterAI in business operations

External AI used by the company

Work requests from each organization and AI agent

Each runs within its own permissions and configured scope of work.

INNORIX

Current workMultiple business workflows

Actual execution statusMultiple AI environments and existing Flows in operation

Next workExpanding into new work

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

Next story, 6 months laterThe company changed its AI, but did not rebuild its work. →

6 months laterScene 14 / 14

The company changed its AI, but did not rebuild its work.

The company began reviewing new AI models and development tools.

Some development teams moved to a different AI development environment, and general employees also began using AI better suited to their own work.

Another organization introduced an AI agent it had developed in-house.

However, the file transfer work actually running at the company did not exist only inside a particular AI’s conversation history or prompts.

Delivery targets, recurring schedules, execution conditions and recovery policies were configured in the company’s work execution environment.

SIX MONTHS LATER

The company keeps its existing work while adopting new AI.

AI environments used at first

  • Existing AI development tools
  • Existing business AI
  • Early project agents

Newly adopted AI environments

  • New AI development environments
  • A variety of business AI
  • The company’s own AI agents

Company work that keeps running unchanged

FlowConfigurationStatus
Corporate data intakeExisting data setupActive
Morning analyticsExisting run scheduleActive
Project data preparationExisting work procedureActive
Validated data deliveryExisting execution conditionsActive
Result archiveExisting retention policyActive

AI can change,
but the company’s work must go on.

Live contextNow

6 months laterAI in business operations

External AI used by the company

Work requests from each organization and AI agent

Each runs within its own permissions and configured scope of work.

INNORIX

Current workMultiple business workflows

Actual execution statusMultiple AI environments and existing Flows in operation

Next workExpanding into new work

Work execution environment

External AI, MCPINNORIX, File transferEnterprise systems

FROM ANSWERS TO ACTIONS

From the first day work was entrusted to AI,
the company’s daily routine began to change little by little.

From a company that asked AI questions,
to a company where AI takes part in real work.

  1. BEFOREA company that asked AI

    How should I do this work?

    AI explains how to do the work.

    Employees do the work directly
  2. DAY ONEA company that entrusts work to AI

    Run this work.

    AI’s requests lead to real file transfer work.

    Actual execution results and records checked
  3. TODAYA company that works together with AI

    Prepare the work needed for the new project.

    AI and the existing business systems carry out the company’s work together.

    Many AI and existing Flows in operation
The many AI environments the company chooses
  • Claude
  • GitHub Copilot
  • ChatGPT
  • Gemini
  • AI Agents
  • Custom AI
INNORIX The company’s execution environment for data delivery work
  • On-premises
  • Cloud Storage
  • Enterprise Systems
  • Business Applications
Beyond what AI knows,
what it can execute at the company.

THE TECHNOLOGY BEHIND THIS STORY

Multi-Stage Flow

Configure the company’s data delivery work as Flows, and run the work you need or check its progress from external AI. New work requests from AI and the automated work the company already runs operate in a single data delivery environment.

View product details