METHODOLOGY
A credential is only worth
the method behind it.
The integrity isn't a claim — it's a method. An open competency framework, empirical calibration, and a trace-level signal behind every credential, engineered so the one issued in 2026 still means the same thing in 2030.
Attempt A-184 · Reviewed
Submitted 12 May 2026 · Proctored · Counter-signed
79 / 100
✓ Passed
Signal layers
The shift
We stopped measuring what you know and started measuring what you can run.
As AI systems take over execution, human responsibility shifts to orchestration, validation, and control. The credential that graded recall is measuring the wrong thing. Ascend measures the new one — who can operate intelligent systems in real environments.
- Orchestrate
- Validate
- Control
- Recall a factDirect a system to the outcome
- Follow a procedureValidate what the system produced
- Complete a taskHold control under failure
What Ascend measures
Capability is no longer what you know — it's how effectively you operate intelligent systems.
How the signal is built
One attempt in. A signal worth standing on, out.
A single score isn't trust — it's a number. Ascend runs every attempt through a defined loop and emits five measured layers, so the learner, the institution, the employer, and the regulator can all stand on the same result.
What comes out
Signal vector
Five layers per assessed attempt
- AutomatedOutcome correctness75%
- LLM-as-judgeTrajectory quality68%
- ProfiledEfficiency65%
- Red-teamedSafety adherence72%
- ReviewedCode & prompt quality70%
How it gets there
- 1
Define
Competencies mapped to industry evidence, reviewed by practitioners, calibrated against real job performance.
- Open framework
- Slice mapping
- 2
Assess
Challenges run in a controlled sandbox. Every action is trace-captured — not just the result — then scored across the five layers.
- Live sandbox
- Trace capture
- 3
Signal
One submission flows to the learner profile, the national dashboard, and the employer pool — simultaneously.
- Learner profile
- Employer pool
Why it holds up
Trust isn't earned by claim.
It's earned by method.
Three mechanisms keep an Ascend credential durable — a versioned open framework, empirical calibration, and a trace chain down to the evidence. Each one is something you can open and inspect, not a promise you have to take on faith.
Open, versioned framework
Every domain, competency, and scoring criterion is published and versioned — no black box. When the framework changes, the diff is public and the calibration cohort behind it is named.
View the frameworkFramework · Agentic DevOps
v1.1 → v1.2 · 12 May 2026 · Published
Empirically calibrated
Rubrics are tuned against real practitioner performance, and the evidence behind every score is captured live — not asserted after the fact.
How calibration worksIntegrity report
A-184 · 6 events · 0 disqualifying · Reviewer sign-off required
Traceable to evidence
Every credential carries its full chain of custody — from identity check to counter-signed submission to human sign-off. Auditors can verify the whole path.
See the platformChain of custody · CR-218
6 events · Verified · Defensible to audit
12 May · 14:32
Identity verified
Candidate
12 May · 14:35
Session opened
Proctor
12 May · 14:59
Trace counter-signed
System
12 May · 15:12
LLM-as-judge reviewed
Calibrator
12 May · 15:38
Human spot-check ✓
Reviewer
12 May · 15:38
Credential issued · OB3
Platform
A credential issued in 2026 should mean the same thing in 2030. The open framework, the calibration, and the trace chain are all in service of that durability.
Talk to our methodology team