Platform data flows in
Location, telemetry, sensor, alarm, event and production records land in CreaAI's data layer the moment they are created on the platform. There is no separate transfer job or overnight batch load to wait for.
Automation · Artificial Intelligence Layer
CreaAI in-house artificial intelligence layer
The in-house artificial intelligence that talks to your live platform data. You ask the question in your own words; the answer arrives within seconds and the same result turns into an Excel report in a single step.
BLE location data, drum status, geofence violations, event records and production data collected from factory machines all accumulate on the same platform, and CreaAI sits on top of that. The model is hosted on the organization's own infrastructure; questions and records are never sent to a third-party AI service. This page covers the artificial intelligence layer of our industrial automation solutions in detail.
- Vector Index
- Embeddings
- In-House LLM
- Natural Language Query
- Follow-Up Questions
- Excel Report
- CSV / PDF
- Scheduled Reports
- Permission Filter
- Source Listing
- On-Premise Model
- Private Cloud
01 HOW IT WORKS
In-house AI
In-house AI
that talks to your platform data
CreaAI is an in-house artificial intelligence layer that runs inside the platform. Live data from the field is indexed as vectors and interpreted together with a language model; the question asked is answered within seconds, and the same result turns into an Excel report.
BLE location data, drum status, geofence breaches, event records and production data collected from factory machines all accumulate on the same platform. CreaAI sits on top of that: instead of raising a report request, building a filter or hunting for the right screen, you ask the question in your own words.
The model is hosted on the infrastructure the platform runs on. Questions and records are never sent to a third-party AI service; the answer is built from the organization's own data, and the records it came from are listed.
Data becomes vectors
Records and text are converted into embedding vectors and stored in a meaning-based index. Similar records sit next to one another in that space, so search works on meaning rather than keywords.
The model reasons with context
The vectors closest to the question are selected and given to the in-house language model as context. The answer is built from the organization's real records, not the model's general knowledge.
An answer or a file
The result appears in the chat as text and a table. The same result can be turned into an Excel, CSV or PDF file and shared in one step.
THE DECISIONS BEHIND THE FLOW
- Meaning-based search: "The drum emptied", "discharge took place" and "the pour finished" are worded differently but land in the same region of the vector space.
- Current context: The index refreshes as data is created, so a record entered today can be used in a question asked the same day.
- Numerical accuracy: Operations such as totals, averages and sorting are not left to the language model; they are computed in the platform's own query layer.
- Source citation: The records each answer was produced from are listed, so the result can be traced back to its source.
- Permission filter: The vector search is limited by the asking user's platform permissions; a record outside those permissions never enters the context.
- Modular connection: When a new module is added to the platform its data flows into the same index; no separate development is needed for CreaAI.
02 ANSWERS IN PLAIN LANGUAGE
Write the question,
Write the question,
read the answer
Using CreaAI needs no query language, report design or knowledge of the screens. You write the question in everyday language, and the answer comes back as text, a table and a short comment together.
Which records were used appears beneath the answer. To narrow the result, move it to a different period or change the breakdown, you simply carry on in the same conversation.
- Follow-up questions: When you say "now produce the same list for the previous shift", the context is kept and the question is not rebuilt from scratch.
- Mixed questions: Data from different modules — BLE location, drum status, event records — can be combined in a single answer.
- Short comment: Alongside the figure, the notable deviation and any recurring pattern are summarized too.
CREAAI · PLATFORM ASSISTANT
LIVE DATA · CONNECTED
User
Rank the lines with the most downtime last shift, and the causes, by duration.
CreaAI
I scanned 412 event records from the night shift. Total downtime 96 minutes; the three longest causes are below.
| Line | Downtime cause | Duration |
|---|---|---|
| Line 3 | Feed wait | 38 min |
| Line 1 | Sensor alarm | 24 min |
| Line 3 | Operator intervention | 17 min |
| Line 2 | Maintenance | 11 min |
| Line 4 | Material change | 6 min |
The feed waits on Line 3 recur in all of the last three shifts; you may want to look at them alongside the forklift position records.
Download as Excel
Turn into a chart
Source records
EVERY PLATFORM MODULE IN ONE CONVERSATION
-
AUTOMATION
List the vehicles that breached a geofence this week and how long each breach lasted.
Boundary breach records are grouped by vehicle and zone, xlsx can be downloaded. -
BLE PLATFORM
Which zones had the most forklift-to-pedestrian near misses yesterday?
RTLS trace records zone and hour breakdown, with risk density ranked. -
DRUM AUTOMATION
Show the mixers that discharged outside the plant yesterday.
Geofence breaches by drum status assessed on the same time axis. -
FACTORY INTEGRATION
Report the machines with the most faults last month and their downtime.
PLC and SCADA event records are collected per machine and downtime cause grouped accordingly. -
ENVIRONMENTAL MONITORING
Report the sensors that breached a threshold this week and for how long.
Sensor, threshold, duration and recurrence count in one table, with an optional PDF summary. -
FLEET
List the 5 vehicles with the highest fuel consumption per km last month.
Trip and fuel records are combined per vehicle and summarized by deviation is calculated.
03 REPORT GENERATION
The same question,
The same question,
this time as an Excel file
The answer on your screen turns into a shareable file in one step. Columns, breakdown and sort order are built exactly as the question described them, with no template to prepare.
| Line | Downtime cause | Events | Total duration | Avg. duration | Share |
|---|---|---|---|---|---|
| Line 3 | Feed wait | 9 | 38 min | 4.2 min | 40% |
| Line 1 | Sensor alarm | 6 | 24 min | 4.0 min | 25% |
| Line 3 | Operator intervention | 4 | 17 min | 4.3 min | 18% |
| Line 2 | Maintenance | 2 | 11 min | 5.5 min | 11% |
| Line 4 | Material change | 3 | 6 min | 2.0 min | 6% |
Excel Workbook
Headers, formatting and number alignment come ready. Summary and detail can be produced on separate sheets, and the file is ready to share as it is.
Raw Data Export
A raw table for moving into your own analysis tools, with column order and encoding kept fixed.
Management Summary
A one-page summary you can take into a meeting, with a short comment and a chart beside the table.
Recurring Report
A question asked once can be made recurring: the report is produced automatically on the chosen day and time and sent to the people concerned.
04 IN-HOUSE MODEL
The model inside the organization,
The model inside the organization,
the data inside it
CreaAI is an in-house model: it is hosted on the infrastructure the platform runs on. Questions, records and the reports produced are never sent to a third-party AI service.
Data Stays Inside
The model and the vector index run on the platform's own infrastructure. Location, production, staff and customer data never leaves the organization, and no query is sent to a service open to the internet.
Permission-Aware Answers
CreaAI sees only the data the asking user's platform permissions allow. A record the user cannot see on screen cannot enter the answer either.
Traceable Result
Every answer lists the records it was produced from. A line in the report can be traced back to its source record, so the result stands up to audit.
INSTALLATION AND OPERATION
- On-premises or private cloud: The model can run on your existing platform servers or in a private cloud environment of your own.
- Record keeping: Questions, the answers produced and the reports downloaded are all logged per user.
- Scope limitation: Which modules' data enters the index is decided by the organization.
- Verification: For critical reports, an audit view is provided that lets the answer be compared against the source records.
05 WHY CREATECH
Why choose Createch
Why choose Createch
solutions?
Home-grown engineering strength, sector experience and a way of working focused on measurable results.
Home-Grown R&D
Through our in-house R&D center we develop innovative, sustainable solutions built for the needs of industry.
Sector Expertise
Many years of experience in the sector mean we know the operational needs of sector leaders at first hand.
Continuous Support
Our expert technical team provides fast, effective service to keep your operations running.
Measurable Results
The solutions are designed to deliver concrete efficiency gains and cost savings in operational processes.
06 CONTACT
Let your data speak
Ask your question without opening a report request, building a filter or hunting for the right screen. CreaAI brings together the location, telemetry, sensor, alarm, event and production data that accumulates on your platform in a single conversation, and turns the result into a file you can share.
Get in touch for detailed information, project consultancy and demo requests. Our technical team is ready to plan a CreaAI deployment around your organization's data scope.