Explore Keystone with pre-seeded data.
This is a read-only CFO workflow demo with synthetic AI spend, live dashboard logic, and no signup wall. It shows the command center, the AI CFO Brief, and the onboarding path in one pass.
Scope
45 days of synthetic OpenAI, Anthropic, and Bedrock usage mapped into Keystone Units.
Mode
Read-only. No credentials, no email verification, and no backend writes required.
Rates
Estimated Rates power the cost calculations shown in the charts and brief.
What To Look For
Spend trend, budget risk, anomaly queue, department attribution, and the board-ready brief.
Scenario
This demo uses a fixed seeded scenario ending on 2026-05-31 so every visitor sees the same spend patterns, anomalies, and CFO brief.
Total AI Spend
$0.00
0 usage records
Keystone Units
0.0 KU
Normalized across providers
Projected EOM
$0.00
on track vs $2.4k budget
Blind Spots
0
No anomalies detected
Command Center
Daily spend trend across the latest 30 days of seeded usage.
No data yet — upload a CSV to populate
Spend by Model
Real dashboard chart components running against demo data.
No data yet — upload a CSV to populate
Budget Outlook
on track
Current month spend is $0.00 with a 7-day run rate of $0.00/day.
Days to budget: n/a • Variance: $2.40k
Largest Anomaly
No anomaly spikes in the current demo sample.
Benchmark Status
105
Ahead of the reference operating baseline
Reference baseline reflects efficient saas teams operating with 2+ providers, sub-35% top-model concentration, and cost per 1k tokens near $0.0035.
AI CFO Brief
Board-style summary generated from the same seeded dataset.
AI CFO Brief
Executive summary
Keystone analyzed 0 usage records totaling $0.00 (0.00 KU). Projected month-end spend is $0.00 against a $2400.00 budget.
Spend vs. prior period
The current 7-day run rate is $0.00/day versus a 30-day run rate of $0.00/day.
Forecast and budget risk
Budget status is on track. Days to budget: not available.
Biggest cost drivers
Top model: none. Top provider: none.
Blind spots and anomalies
Detected 0 spend anomalies requiring review.
Recommendations
- Keep the current budget posture, but review model mix weekly as usage grows.
- Continue monitoring for shadow AI by requiring team/use-case tags on new usage.
- Upload provider usage to unlock model-level optimization recommendations.
Onboarding Path
What a real customer does after the demo convinces them.
1. Connect
Link providers or drop a CSV export.
OpenAI, Anthropic, Google billing exports, and Bedrock feed the same normalized spend model.
2. Review
Check trend, budget, and anomaly queues.
Keystone surfaces run-rate risk, model concentration, and spike alerts without waiting for month-end close.
3. Report
Generate the AI CFO Brief for leadership.
Export the board-facing narrative after the metrics and source data look right.
Illustrative Connection Snapshot
A clearly labeled mock modeled after the live connections view.
OpenAI
Usage sync healthy. GPT-4o and GPT-4o-mini mapped into KU.
Anthropic
Claude spend normalized and available for benchmark comparisons.
AWS Bedrock
Read-only example shows a provider awaiting refreshed credentials.
Next Step
Move from seeded demo to your own data.
The live product keeps the same views, but replaces this synthetic dataset with org-scoped provider syncs, CSV ingestion, and saved CFO reports.