/workspace
ContextBase, put to work.
The workspace is where your team uses ContextBase. Ask questions, run agents, build apps and dashboards, check value cases, and automate integrations.
01Conversational analytics
Ask a question. Get the answer and its evidence.
Type it in plain English. With the answer, you see the tables it read, each query it ran, and its checks.
What are the top denial drivers across the estate?
1. Rate raises not applied. 407,098 claims, $8.38M underpaid to providers. This creates prompt-pay and dispute exposure.
The net is only $1.6M, but $18.3M gross is mispriced. A net view would hide it.
2. COB: the leak is to dual members, not the plan
- Medicaid never paid ahead of Medicare. That direction is clean.
- On 42,039 paid claims (about 2%) secondary coverage did not post, and the full cost share went to the member.
3. Copay drift
- Claims adjudicate at the configured copay 30 to 50 times more often than at the filed one.
Evidence · 3 sourcesWorth a look · 7 checks
Answer 17 queries
worth a look7 checks
Answer 234 queries
worth a look7 checks
Claims line detail
Claims source claim_line
Evidence used here
9 queries · 91 rows returned · 2.7M rows staged
select c.claim_id, c.allowed_amt, k.contract_rate from claims.claim_line c join network.provider_contract k on k.provider_id = c.provider_id where c.service_date between k.eff_date and k.term_date
About this source
- FQN
claims_dw/claims/claim_line- Fields
- 48 columns
- Context
- No context yet
Demo data
02Agents
Agents run your checks on a schedule.
Tell an agent what to check and when to run. It is read-only by default, and it posts each report where your team works.
RADV Audit
Demo dataChecked the 1,240 risk-adjusted diagnoses submitted this year against the encounter records on connection 1. 18 have no encounter that supports them.
| Why it was flagged | Diagnoses |
|---|---|
| No encounter on the date of service | 11 |
| Encounter from a provider type CMS does not accept | 5 |
| Diagnosis missing from the encounter it cites | 2 |
The run report lists all 18 by provider. The risk adjustment team was notified.
03Apps and dashboards
Build an app or a dashboard from one sentence.
Describe what you need, and ContextBase builds it on your data. It opens with fresh numbers, and any panel shows its source and its SQL.
Live rev #1 · data from 32d · · built from demo_ind_ma
$16.0MTotal lab plan paid
$3.3MCarve-out eligible
783,758Lab lines
5.6%Denial rate
$655.7K2026-06 · −25.1% vs 2026-05
04Value Lab
See which value cases your data supports.
Value Lab checks every case in its library against your data. Each comes back Ready, Gaps, or Blocked, with the data it still needs.
Value Lab
Where the value is
125 high-value cases across 10 domains, led by Claims and payment with 31. 1 of 41 quick wins are on the workbench.
- Library
- 375
- cases, checked against your data
- Quick wins
- 41
- High value, low effort
- Workbench
- 10
- 3 Now · 3 QA · 2 Next · 2 Later
- Data ready
- Not checked
05Data integration
Automate integrations safely.
Data Pipelines sends eligibility, claims, and accumulator files to each vendor on its schedule, in production. A new integration starts as a copy of one you already run, and runs in test first.
New integration
- Copies
- Demo Logistics
- Relay
- Named from the employer
- File
- Named from the employer
- Runs
- Monthly on the 10th, 13:00 UTC · 8 GB
06Impact
What teams built.
Each is a published impact study, on the team’s own data.
- Large national payerCaught the copay defect review missed.→
Large national payerCaught the copay defect review missed.Read the study→ - Third-party administratorEmployer reports now run on schedule.→
Third-party administratorEmployer reports now run on schedule.Read the study→ - Health plan, enterprise performanceBusiness users own their dashboards.→
Health plan, enterprise performanceBusiness users own their dashboards.Read the study→ - Large third-party administratorMore data work. No new hires.→
Large third-party administratorMore data work. No new hires.Read the study→
Put the first one to work.
Pick a question or a recurring check. We connect what it needs, and your team sees it run.