SYNTHETIC DEAL LAB / CONTROLLED TEST ENVIRONMENT
Realistic sales work.
Zero live customer data.
The Deal Lab makes complete demonstrations possible using fictional companies, versioned opportunities, CRM-shaped records, emails, transcripts and deliberately planted ambiguity.
LAB PRINCIPLE / 01
The ambiguity is intentional.
Missing facts, stale notes, weak signals and contradictory evidence are part of the test—not flaws to clean up before the model sees them.
CASE FILE / SYNTHETIC
Northstar Systems.
A fictional enterprise opportunity with enough realism to test research, qualification, pipeline and customer-deliverable workflows without exposing customer information.
THIS IS NOT A CUSTOMER CASE STUDY.
01 / Source packet
CRM-shaped records, synthetic emails, meeting notes, transcripts, activity history and only the context required for the workflow.
02 / Planted friction
Contradictory statements, missing stakeholders, stale assumptions, weak signals and facts the system is expected to leave unknown.
03 / Frozen truth
Where a workflow is scored, the answer key and evaluation rules are defined before the AI run so the result cannot be graded on vibes.
04 / Preserved first run
The first useful and failed outputs are retained so corrections, omissions and unsupported inference can be evaluated rather than edited out of the story.
TEST ROUTE
Known → contradicted → unknown → reviewed.
KNOWN
Confirmed input
Facts the case file actually establishes.
CONTRADICTED
Conflicting evidence
Signals the workflow must not flatten into one answer.
UNKNOWN
Deliberate gap
Missing information the model should leave unresolved.
HUMAN REVIEW
Accountable action
The reviewer decides what survives and what action follows.
THE LAB CAN TEST
Workflow behavior.
Extraction quality, evidence mapping, omissions, unsupported inference, correction effort and usefulness under the stated conditions.
THE LAB CANNOT PROVE
Business outcomes.
A synthetic workflow does not prove revenue lift, win-rate gains, universal accuracy, shorter sales cycles or customer ROI.
CONTROLLED TESTING / NO CUSTOMER DATA / NO FABRICATED OUTCOMES
