Login AI
AI data readiness audit
We survey your sources before an AI project: we build the register, profile the data, and review PII and access. You get a quality scorecard and a preparation roadmap.
We collect the sources into a register, grade quality, map PII and access, and hand over a scorecard and roadmap.
Pricing
from ₽300,000
per project
This is the lower bound of a project. We scope the exact quote after a brief: it depends on volume and the number of integrations.
Capabilities
Source register
We collect every source in one list: systems, tables, files, queues, and stores. For each source we record the owner, volume, refresh rate, and export path.
Quality scorecard
We score completeness, duplicates, freshness, and consistency for each source. The scores show which data is ready for a model and which needs cleanup.
PII and access review
We find personal data, check the processing basis, and map role permissions. We also see who can export sensitive fields and how.
Roadmap
We split the preparation into steps with time and effort estimates. The plan tracks dependencies between sources and the order of defect fixes.
From request to result
1
Access and inventory
We get read access and build the source register with the data owner. We record systems, volumes, and refresh rates.
2
Profiling
We count completeness, duplicates, distributions, and freshness on samples. Each source gets a report with numbers and sample records.
3
Scorecard
We fold the scores into a scorecard by source and attribute. It shows where data is ready and where you need cleanup or a new extract.
4
Roadmap and handover
We assemble the work plan by priority and estimate the timeline. We hand over the register, scorecard, and map in a team meeting.
Check if it fits you
Good fit
Three sources or more
An AI project draws on CRM, ERP, stock, and analytics. The register and scorecard show how to join them into one loop.
A data owner is named
Someone on your side confirms access and priorities. With that person you agree on samples and settle disputed fields faster.
Not a fit
No access to the sources
Without table and file reads, profiling runs on descriptions. That report shows structure and misses defects inside the data.
An audit as a cleanup substitute
The audit records defects and priority, while the team or a separate project removes duplicates and fixes reference data.
What you get
−90%
fewer discrepancies in reports
Our team removed 1,200 duplicate counterparties and linked stock to orders. Stock reconciliation takes 5 minutes, and report discrepancies fell by 90%.
3
Sources in one register
The three sources in the base package get an owner, volume, and refresh rate. The register shows how to join CRM, stock, and reporting into one loop.
Scorecard
Scorecard as the base artifact
The scorecard grades completeness, duplicates, and freshness for each source. It gives you a numeric basis for the data preparation talk.
5 min
Reconciliation in ERP
In the ERP case, stock reconciliation takes 5 minutes after our team removed 1,200 duplicates. We use that result as a benchmark for cleanup scope.
What the delivery includes
Source register
The register lists systems, tables, files, and queues with owner, volume, and refresh rate. It shows where each attribute comes from.
Quality scorecard
The scorecard grades completeness, duplicates, freshness, and consistency for each source. Scores and sample records set the cleanup priority.
Roadmap
The roadmap splits preparation into steps with dates and owners. It ties source defects to team tasks and the order of work.
How the system is built
Source inventory
We walk the systems with the data owner and write each source, its tables, files, and queues into the register. Each entry carries volume, refresh rate, and export path, so the picture stays verifiable.
Data profiling
On a sample we count completeness, duplicates, distributions, and freshness for each data attribute. Python scripts repeat the calculation on new extracts, and you see quality move between iterations.
Scorecard and roadmap
Scores roll into a scorecard by source and attribute, and defects get a priority. The roadmap orders the steps by dependency, and the team starts with the source that covers 3–5 priority scenarios from the register.
Where the solution pushes back
- 01
Read access
The audit relies on read access to tables and files. Full access speeds up profiling: the first source numbers arrive by the end of week one.
- 02
Source quality
The depth of the findings follows the completeness of the sources. With 40% empty fields, the scorecard shows that gap and the roadmap puts cleanup first.
- 03
Cleanup as a separate project
The audit records defects, priority, and scope. The client team or our separate project removes duplicates and fixes reference data to the plan.
Technologies
The technologies we use in our work and their role in a project.
Survey stack
We pull extracts through Python, check transactional data in PostgreSQL, aggregates in ClickHouse, the event stream in Kafka, and files in MinIO.
Cost calculator
This is an indicative range, not a quote. We scope the exact estimate after a brief: it depends on volume and the number of integrations.
Sources
Data volume
PII audit
Scorecard
Add-ons
Support
Year one: updates and monitoring, a reply within 2 business days.
Service
AI data readiness audit
The estimate covers our team's work. Hardware, third-party service licences and travel are added on request.
Estimated cost
₽450,000
Timeline
3 wk
Frequently asked questions
How long does the audit take?
Three to four weeks for three sources. The timeline depends on table count, sample volume, and how fast access is approved.
What is in the result?
A source register, a quality scorecard, a PII and access map, and a roadmap. We also hand over the samples and profiling scripts.
How do you handle PII?
We find personal data through the catalog and samples, check the processing basis, and map roles. We mask sensitive fields in the reports themselves.
What comes next?
You pick the first step from the roadmap and start cleanup or a pilot. We can run that stage as a separate project.
What teams add to this service
Related Services
Part of Solutions
Implementation Examples
−90%
Discrepancies in reports
Ready to discuss your task?
Leave a request – we'll get back with a proposal and a preliminary estimate.
Contact us