Where AI meets numbers people already trust
AI on top of a governed data platform — because a natural-language answer is only as good as the model underneath it.
A natural-language layer over untrusted data produces confident nonsense
"Ask your data a question" demos beautifully and fails quietly. If two teams disagree on what revenue means, an AI assistant will pick one definition without telling anyone, and the answer will be presented with more confidence than a human analyst would use.
Fix the semantic layer first, then add the assistant
We model the data and agree the definitions before any AI touches it. The assistant then queries the modelled layer rather than raw tables, and shows the query it ran — so an analyst can check the work rather than take it on faith.
Capabilities
AI-powered dashboards
Generated narrative summaries alongside the visuals, grounded in the same governed model.
Natural-language analytics
Questions answered against a semantic layer, with the generated query always visible.
Automated reporting
Recurring packs assembled from source systems instead of rebuilt by hand.
Data quality automation
Tests on transformations and freshness monitoring that alert before a stale number is presented.
Data classification
Tagging sensitive fields so downstream tools can enforce handling rules.
Anomaly detection
Alerting on values outside expected bounds, with the threshold explained.
Forecasting
Baseline forecasts with published error ranges, not a single confident line.
Data assistants
Self-service question answering for teams who currently queue for an analyst.
Power BI AI integration
AI features wired into existing Power BI estates rather than replacing them.
Automated KPI explanations
Drafted "why did this move" narratives for a human to verify before circulation.
How it fits together
From source system to business user
The assistant sits at the end, behind the semantic layer. That ordering is the whole design — reverse it and you get fast answers to the wrong question.
- 01Business database
- 02Data pipelineTested transformations
- 03Data warehouseModelled, documented, versioned
- 04Power BIGoverned semantic model
- 05AI analytics assistantQueries the model, shows the query
- 06Business userCan verify the work
Systems we build against
Platforms and services TOOKLI actively works with on client engagements.
Microsoft data platform
- Power BI
- Microsoft Fabric
- Dataverse
- Azure Data Factory
- Azure Synapse
Databases
- SQL Server
- PostgreSQL
- Azure SQL
Transformation
- dbt
- Python
- SQL
Models
- Azure OpenAI
- OpenAI
This list reflects systems we integrate with, not partnerships, resale agreements or certifications. If something you rely on is missing, ask — we will tell you honestly whether we have used it.
What this looks like in practice
Examples of what we can build for each sector. Work we have actually delivered appears in our case studies, with named outcomes.
- Operations
The monthly pack, assembled automatically
Two days of a capable person's month returned, with the figures traceable.
- Finance
Variance explanations, drafted
A first-pass narrative on what moved and why, for a human to check and sign off.
- E-commerce
Anomaly alerting on trading metrics
Told when conversion drops, rather than discovering it in next week's review.
- Real estate
Portfolio reporting across systems
One governed view where there are currently several partial spreadsheets.
Delivery process
- 01
Audit the current reporting
What exists, who uses it, and where two reports disagree. There are always some.
- 02
Agree the definitions
Written, versioned, reviewed. This is the step that makes everything after it possible.
- 03
Model the semantic layer
Tested transformations, documented metrics, monitored freshness.
- 04
Rebuild reporting on the model
Existing dashboards repointed, so nobody loses what they rely on.
- 05
Add AI where it earns its place
Summaries, anomaly explanations, natural-language querying — each measured.
- 06
Monitor quality
Freshness and test failures alert an engineer before an executive notices.
How we keep it safe and accountable
Engineering practices, not certifications. TOOKLI holds no security certifications and does not claim any.
The generated query is always visible
A natural-language answer shows the SQL or DAX it ran. An analyst can check it, which is the difference between a tool and an oracle.
We do not claim vendor features that do not exist
AI capability in the Microsoft data stack changes quickly and is licensed unevenly. We confirm what is actually available on your tenant and licence before designing around it.
Definitions are governed
Metric definitions live in version control and change through review, so a number cannot quietly shift meaning between two reports.
Forecasts carry error ranges
A forecast without a stated confidence interval invites false precision. We publish the range and the backtest.
Common questions
Related services
Business & Workflow Automation
Business process, document, data and system automation. We combine ordinary software workflows with AI only where judgement is actually needed.
Learn moreAI Integration
Bring AI capabilities into your existing applications — search, extraction, classification, summarisation and assistants — without building or training models yourself.
Learn moreData Analytics & Business Intelligence
One trustworthy version of the numbers, and dashboards people actually open.
Learn more
Do people trust your numbers?
If two reports disagree, that is the problem to fix first. We will audit the reporting and tell you what is actually wrong before proposing any AI.