Knowledge was scattered.

• Employees searched across Foundation Software, CompanyCam, SiteDocs, SharePoint, email, and a proprietary bidding platform to find answers.
• There was no single source of truth – critical answers were buried in 17 years of spreadsheets, PDFs, and project close-out folders.

Decisions were made blind.

• Supervisors, PMs, and executives had no unified operational view across active jobs and historical records.
• Site disruptions and bid reviews forced calls based on incomplete or conflicting information.

Institutional expertise didn’t scale.

• Hard-won knowledge lived in the heads of long-tenured staff – not in documented workflows.
• When senior employees were unavailable or new hires joined, the whole organization slowed down.

Sphere Solution

SphereIQ is a custom AI assistant that sits on top of a company’s existing systems. It doesn’t replace any software the client already uses – it reads from them, reasons over them, and answers questions in natural language. The client continues working in Foundation Software, CompanyCam, and SiteDocs as before. SphereIQ is the layer that makes all of those systems queryable from one place.

The integration work

Sphere’s team spent the first phase of the engagement mapping every system the client depended on and how data flowed between them. Eight primary sources were connected through a combination of native APIs, authenticated service accounts, and custom connectors for the client’s proprietary bidding and job-tracking software. For the 17 years of legacy content – spreadsheets, scanned PDFs, Word documents, project close-out folders – Sphere ran a full ingestion pass: documents were parsed, OCR’d where needed, chunked, and indexed into a vector store with metadata preserved so that every answer could be traced back to the original file.

Data freshness was a core requirement. Financial data from Foundation Software, live bid pipeline data, and field photos from CompanyCam sync on short intervals so that a project manager asking about job cost variance on a Tuesday morning gets numbers that reflect Monday’s transactions. Historical and document-based sources are re-indexed on a slower cadence.

The assistant layer

SphereIQ runs on a large language model orchestration layer that routes each question to the right combination of data sources. A supervisor asking “What’s our current incident rate on the Austin project?” pulls from SiteDocs. A PM asking “What were the change orders on last year’s hospital build?” pulls from Foundation Software plus the archived project folder. An executive asking “Which open bids are we most likely to win based on past patterns?” pulls from the proprietary bidding system combined with historical bid outcomes.

Every response cites its sources. Every answer links back to the originating document, photo, or record so that users can verify and drill in. Users can ask follow-up questions, upload a document mid-conversation, or request a summary that gets exported to Word or PDF.

Role-based access architecture

Access control was built into the foundation of the system – not layered on afterward. Sphere worked with the client’s leadership to map every role in the org chart to an explicit data-access policy. When a field crew member logs in and asks about their assignment, SphereIQ can only see the data that role is authorized to see. When the CEO asks the same question, the system accesses a different set of sources.

This was implemented at the retrieval layer, meaning the model itself never has access to data the user isn’t authorized for. Authorization is enforced before any context reaches the assistant.

Deployment and rollout

Full production deployment took under two weeks from contract signature. Sphere ran a short internal pilot with 10 users across three roles – a PM, a site supervisor, and a member of the back-office team – to validate the answers, tighten prompts, and confirm access boundaries. After pilot sign-off, the system was rolled out company-wide in a single week, with role-based onboarding sessions for each user tier.

Sphere continues to maintain the deployment: monitoring query accuracy, adding new data sources as the client’s tech stack evolves, and tuning the system as usage patterns reveal new high-value workflows.