Industrial AI.Your data. Your cloud. Your model.
Sightwell connects the industrial systems you already own and makes their operational knowledge safely usable by AI.
- AVEVA PI
- Asset Framework
- AVEVA Historian
- Ignition
- OPC UA
- SharePoint
- File shares
- SQL
- Excel
- Power BI
- Discharge temperature increased 8.2% between 13:40 and 14:35.
- It follows a suction pressure drop at 13:22 and an anti-surge valve opening at 13:38.
- A similar pattern in March ended with a seal replacement.
- No limit currently exceeded. Temperature is below the inspection threshold in the operating procedure.
- 14 MARSimilar excursion · WO-48213
- 13:22Suction pressure low
- 13:38FV-1050 opens to 34%
- 13:40–14:35Temperature rise
Your AI should come to your industrial data. Not the other way round.
Connect what you already own.
Historians, asset models, SharePoint, SQL and spreadsheets stay where they are.
Reason across data and documents.
Ask why production changed and see the trends, events and procedures behind it.
Every claim traced to evidence.
Tags, time ranges, assets, document pages and formulas. No evidence, no confident answer.
Identity decides the answer.
Operators, engineers, vendors and executives each get their own permitted context.
Start where you are
Find the calculation before it becomes the problem.
Critical numbers are often computed in spreadsheets, PI calculations and dashboards with hard-coded assumptions nobody has checked in years. Sightwell finds them and compares them with the plant data underneath.
- Operations Daily Report.xlsx
- Gas Balance Dashboard · Power BI
- Monthly Production Workbook
Operations Daily Report.xlsx uses a fixed conversion factor. Measured energy from the incoming gas meter varies over the selected period, and exceeds the contract limit on 3 days where the spreadsheet stays below it.
The meter should be treated as the authoritative operational measurement, subject to confirmation by the responsible OT owner.
A hidden spreadsheet assumption can become an operational decision.
Nominations, contract limits, production reports and investor figures often trace back to a single cell.
Trace calculations back to the plant data underneath.
Excel, PI calculations, Power BI, SQL and scripts: where each input comes from, and which copies exist.
Discovery and governance, not automatic approval.
Sightwell surfaces constants, thresholds and duplicated formulas. Your engineers decide what changes.
Your P&IDs, connected to the historian.
Drawings sit in SharePoint. Tags sit in PI. Sightwell links every instrument on the drawing to its PI Asset Framework element, so an investigation can follow the process, not just the numbers.
Interactive drawing requires JavaScript.
Drawings become navigable.
Select an instrument on a P&ID and see its live and historical data, its asset and its documents.
Investigations follow the process.
Sightwell uses drawing connectivity to look upstream and downstream, not just at the tag that alarmed.
Find where documents and data disagree.
Instruments missing from PI, tags missing from drawings, and drawings older than the last change.
Trace every dashboard number back to the plant.
Board packs come from Power BI. Power BI often comes from Excel. Excel is typed in by people. Sightwell traces each KPI back to its source and checks it against what PI actually recorded.
| Date | Stop | PI event frame | Downtime Log.xlsx | Hours |
|---|---|---|---|---|
| Unplanned hours counted | Report 9.4 h | PI 25.9 h | ||
Finding RL-0009 routed to the report owner. Sightwell shows the difference and its evidence; it does not change the report.
Know which number your board is seeing.
Every KPI traced through Power BI, Power Query and Excel to the people and systems that produced it.
Reconcile reports with the historian.
Manual logs compared event by event with PI, so gaps and reclassifications show up before the board pack does.
Keep your reporting tools.
Power BI and Excel stay as they are. Sightwell reads their lineage and adds the evidence underneath.
Service the pump before it becomes an outage.
Two readings drifting together, a drawing, a manual, a maintenance plan and an outside expert. Sightwell puts them in one case file, so the team can decide early and with evidence.
Patterns, not single alarms.
Speed and bearing temperature each stayed inside their limits. Together they described a pump that was wearing.
The backup is part of the answer.
The drawing showed the standby; the maintenance plan showed it hadn't been proven. Both changed the decision.
Experts in, historian closed.
The OEM engineer saw two pumps and one manual for 30 days. Nothing else, and no network connection.
The $400k multiplier.
A number that's slightly off can cost real money once it leaves the building. Sightwell checks reports against the method your documents require and the data your historian holds, before they are sent.
The combustion emission factor is calculated monthly from the carbon content measured by online gas chromatograph AT-0215, weighted by fuel flow FQ-0201. Default factors are not used.
Check the method, not just the number.
Sightwell compares how a report is calculated with how your approved documents say it must be calculated.
Recalculate from the historian.
Measured composition and flow are already in PI. The correct number can be rebuilt and compared month by month.
Before it leaves the building.
Emissions returns, customer quotes and contract statements are cheaper to fix before they are sent.
Every answer traceable to operational evidence.
Industrial decisions need more than plausible text. Each claim links to the tag, time range, document page or formula it came from.
No confident answer.
When sources don't support a conclusion, Sightwell says so and shows what is missing.
Bring Your Own AI. Keep it out of the control loop.
ChatGPT, Claude, Sightwell AI or your internal agents consume the same governed industrial context. Sightwell sits between AI and your infrastructure, read-only.
Add AI without putting AI in control of your plant.
Sightwell starts as a read-only intelligence layer. PLC, DCS, SIS, historians and operational control keep running independently, so you can start small and expand on your own terms.
Already standardized on ChatGPT or Claude? Keep it.
MCP is an open standard; Sightwell implements it. What you are buying is everything underneath: the connectors, context, permissions and evidence.
Your infrastructure. Your identity. Your data.
Sightwell runs where your industrial data belongs. No mandatory SaaS environment, and no upload of plant data to us.
Customer Azure
AVAILABLESightwell operates inside your Azure environment, under your subscription and policies.
On-premises
AVAILABLERuns on Kubernetes within infrastructure you control, next to the systems it reads.
Hybrid
AVAILABLEPlace collectors and processing between plant, edge and cloud where your policy allows.
- Microsoft Entra ID
- Private networking
- Encryption
- RBAC / ABAC
- Audit trails
- Data sovereignty
Industrial AI without an 18-month transformation programme.
No multi-year data project is required to start getting value. Timelines depend on site scope and data readiness.
- Deploy→Sightwell in your Azure or Kubernetes.
- Connect→Historian, SharePoint and identity, read-only.
- Discover→Import assets. Map documents and calculations.
- AskStart investigating, with evidence.
Delivered by the partners who already know your plant.
Qualified industrial technology partners deploy, integrate and support Sightwell. Sightwell builds and backs the platform.
- Deployment
- Historian & OT integration
- Configuration & dashboards
- Training
- L1 / L2 support
- Managed services
- Software
- Product updates
- Platform engineering
- Advanced support
- Roadmap
Connect what you already own.
- 01You don't have to replace what you already use.
- 02Your industrial data stays under your control.
- 03Your AI can reason across plant data and engineering knowledge.
- 04You can find the assumptions you didn't know existed.