USE CASES
Real Environments. Massive Potential. Local Deployment.
Every deployment runs inside existing infrastructure. No migration is required. No new stack is needed. This technology is built for those wondering how AI agents function in a live environment. There are no experts in this new frontier, only new users.
KITSAP CREDIT UNION · LOAN AUDIT AUTOMATION
From 5% sampling toward 100% audited coverage.
Many organizations manually review only a small fraction of their loan portfolio. Coverage is often incomplete and risk remains hidden until a regulator asks a question.
What is possible:
- Move from 5% sampling toward 100% audited coverage.
- Deploy D8:INGEST and D8:CURATE to clean and validate records at the source.
- Use D8:SCOUT to enforce local governance and D8:FLOW to route human review handoffs.
- Audit times could drop from 30 minutes to under one minute per loan.
AGENTS DEPLOYED
D8:INGEST
Connect
Connected to existing loan data systems. No migration. Data never left Kitsap Credit Union's infrastructure.
D8:CURATE
Validate
Cleaned and validated loan records before audit logic ran. Caught incomplete data before it became an error.
D8:SCOUT
Govern
Enforced access controls and masked sensitive fields at every step. Governance built in, not added after.
D8:FLOW
Orchestrate
Coordinated the full pipeline: routing, policy checks, and human review handoffs.
D8:OBSERVE
Monitor
Tracked every audit decision in real time. Generated audit-ready reports with full traceability.
D8:VIEW
Deliver
Delivered findings as a clean dashboard for the compliance team. Exceptions surfaced for human review.
97% faster
Audit time per loan
100% coverage
Every loan audited, every night
99.5% accuracy
Consistent policy interpretation
30% fewer
Human review discrepancies
4–6 weeks
Kickoff to production, not 18 months
Compliance teams focus on decisions rather than data entry. Every loan is audited nightly in under one minute. This occurs without new infrastructure or data migration.
The thread: the same governed data foundation that made KCU's audits trustworthy is now extending to new use cases, without a second platform, a second migration, or a second trust-building exercise. One governed body of knowledge, more than one job done from it.
A LEADING OUT-OF-HOME MEDIA COMPANY · BILLING AUTOMATION
Calculations that took hours can now run in seconds.
Manual billing for lease payments often involves repetitive rules and high compliance risk. Processing billboard rent across a large portfolio can take hours of manual effort.
What is possible:
- Calculations that took hours can now run in seconds.
- D8:FLOW handles the deterministic calculation pipeline while D8:OBSERVE uses an LLM supervisor for narratives.
- D8:VIEW delivers system-ready output without the need for manual reformatting.
- Significant reduction in API calls per record through a hybrid agent pattern.
AGENTS DEPLOYED
D8:INGEST
Connect
Connected to lease and contract data by facility ID. Data stayed in the client's environment.
D8:CURATE
Validate
Validated lease records and calculation inputs before logic ran.
D8:FLOW
Orchestrate
Ran the deterministic calculation pipeline. Managed configurable reporting periods and bundled multiple leases per run.
D8:SCOUT
Govern
Applied access controls and logged every calculation for audit purposes.
D8:OBSERVE
Monitor
One LLM supervisor call per run reviewed output and wrote notification narratives. Down from ~30 API calls per record.
D8:VIEW
Deliver
Delivered system-ready output: validated, splits confirmed, formatted for direct entry. Zero manual reformatting.
Milliseconds
Full billing run, end to end
100% coverage
All error types classified on every record
1 LLM call per run
Down from ~30 API calls per record
Zero reformatting
Output enters the system directly
4–6 weeks
From scoping call to live pipeline
Billing teams no longer touch spreadsheets. The pipeline runs and the system receives validated data every cycle.
PUBLIC EDUCATION · FEDERAL GRANT COMPLIANCE
Grant funding expired unspent. Now every dollar is tracked in real time.
Reconciling federal grant expenditures across PDFs and payroll exports is a manual quarterly burden. When deadlines are missed, funding expires unspent.
What is possible:
- D8:INGEST and D8:CAT extract data from unstructured sources and categorize it by funding restriction.
- D8:CURATE matches expenses to compliance terms and D8:STAGE prepares audit documentation.
- D8:VIEW provides a real-time dashboard of grant utilization for leadership.
- Potential to eliminate grant funding expiration and recover unbilled indirect costs.
AGENTS DEPLOYED
D8:INGEST
Connect
Extracted data from unstructured PDF award letters and sub-recipient spreadsheets. No migration. Connected to existing ERP and payroll exports.
D8:CAT
Categorize
Categorized expenditure types against specific federal award restrictions. Every dollar classified by funding source.
D8:CURATE
Validate
Matched line-item expenses to Grant Award Notification compliance terms. Flagged discrepancies before they became audit failures.
D8:SCOUT
Govern
Kept PII in research payroll data encrypted and localized. Data never left the district's infrastructure.
D8:FLOW
Orchestrate
Synced validated grant expenses directly into the existing ERP. No manual entry.
D8:OBSERVE
Monitor
Monitored for non-compliant spending patterns that risk federal clawbacks. Flagged issues before the quarter closed.
D8:STAGE
Deliver
Prepared audit-ready documentation for annual federal audit requirements.
D8:VIEW
Report
Gave the CFO a real-time dashboard of grant utilization and indirect cost recovery, without waiting for month-end.
85% reduction
Monthly grant reconciliation time
$400K annually
Previously unbilled indirect costs recovered
35 hrs/week
Returned to strategic planning
100%
Grant funding expiration eliminated
4–6 weeks
Deployed in a single grant cycle
Finance teams no longer chase PDFs. Federal dollars are tracked from award to expenditure in real time. Audit season shifts from a months-long scramble to simple report generation.
A CASE-MANAGEMENT SERVICES PROVIDER · CASE DATA INGESTION & KNOWLEDGE GRAPH
From scattered case files toward a governed knowledge graph.
Case-management history handling sensitive, HIPAA-governed data had no relationship-level view of how cases connect. And no AI system could safely touch it without a compliance framework built first.
What is possible:
- Build a HIPAA-aligned compliance framework (discoverable vs. non-discoverable classification) before a single record moves.
- Deploy D8:SEC and D8:INGEST to structure a 100-case representative set from unstructured historical documentation.
- Layer a knowledge graph on top through D8:Platform, surfacing relationships across cases for the first time.
- Validate the full approach at a clear go/no-go decision point before any production commitment.
AGENTS DEPLOYED
D8:SEC
Govern
Defined a HIPAA-aligned compliance framework (discoverable vs. non-discoverable classification) before a single case record moved.
D8:INGEST
Connect
Structured ingestion of a 100-case representative set from unstructured historical case documentation.
D8:PLATFORM
Deliver
Layered a knowledge graph over the ingested case set, surfacing relationships across cases for the first time.
$54,500
Fixed-price SOW, signed 6/02/2026
100 cases
Representative ingestion set, not the full dataset
60 pages/min
Minimum ingestion rate: SOW acceptance target
5 fields
Facility, date, CPT code, amount, quantity: SOW parsing accuracy target
6–8 weeks
Kickoff to go/no-go decision
HIPAA-aligned
Compliance framework built first
This is an active, early-stage engagement: a go/no-go feasibility build, not a committed path to production. Reportable ROI is expected six to nine months out. What's proven today: a working ingestion pipeline and knowledge graph running against real, compliance-governed case data.
The thread: the compliance framework had to exist before a single case record moved: governance designed in, not inspected for afterward. The knowledge graph isn't just storage for 100 cases; it's the client's own case history becoming something they can trust and query, without HIPAA risk following it around unmanaged.
A REGIONAL CREDIT UNION · ENTERPRISE DATA & CLOUD INFRASTRUCTURE READINESS
Undocumented infrastructure, tribal knowledge. Now a validated roadmap.
A credit union preparing for a transition to Azure had no current, trusted view of its own architecture: infrastructure knowledge fragmented across outdated documentation and tribal knowledge.
What is possible:
- Inventory infrastructure automatically through D8:Fabric, a read-only discovery platform. Nothing moved or replaced.
- Build a business glossary, KPI/metrics inventory, and end-to-end data lineage across three priority domains.
- Run a PII exposure assessment and define a security/governance model before any migration begins.
- Deliver a target-state architecture, a Tableau-to-Power-BI migration plan, and an executive infrastructure readiness roadmap.
DISCOVERY DELIVERABLES
Glossary
Define
A business glossary and KPI/metrics inventory: shared definitions for the organization's core data terms.
Lineage
Map
End-to-end data lineage for three priority domains, replacing tribal knowledge with a documented chain of custody.
PII Exposure
Assess
A consolidated view of where personal data lives across systems. No such view existed before.
Governance Model
Define
A target-state security and governance model ahead of the coming Azure transition.
BI Migration Plan
Plan
A defined path from Tableau to Power BI: a roadmap, not a completed migration.
$552,000
Not-to-exceed, Design Partner pricing
6 months
Phase 0 discovery timeline
3 domains
End-to-end data lineage mapped
Read-only
Discovery mode: nothing moved or replaced
Tableau → Power BI
BI migration plan (roadmap only)
This is Phase 0 of a phased engagement: a validated roadmap and governance foundation, not a working production system. No implementation and no ROI numbers exist yet, and by design this phase isn't meant to produce one. A future Phase 1 would be scoped and delivered separately.
The thread: a credit union with real infrastructure and real data, but no trusted, current view of what it actually has. Phase 0 doesn't move a single byte; it turns undocumented, tribal-knowledge infrastructure into something the organization can finally see and govern, before a single production system is touched.