ABSORB

Appearance

ABSORBCan your organization absorb AI speed?

AI made code cheap to write. It did not make software cheap to deliver. ABSORB finds where your delivery waits, decides what can run with less review, and checks whether you were right.

Take the assessment
How it works
GitHub
A delivery path drawn as blocks, with one long amber stage

Find where work waits

Time each stage of a typical change and split it into hands-on work and waiting. Faster coding only shrinks the work. The constraint is usually where changes wait.

The method

Decide what needs less review

Four autonomy tiers, from no human to a change board. Readiness for each comes from your weakest evidence, not an average score.

Levels and tiers

Disagree on purpose

Two people score the same team and swap codes. The statements where they differ most are where the useful conversation starts.

Compare with a colleague

Be wrong in public

Each claim has a written test and a threshold that would count against it. Decision records turn your own predictions into evidence.

Evidence and validation

Try the idea in ten seconds ​

Coding is part of the path from request to customer, not all of it. Drag the sliders: making coding much faster barely moves the whole path, while cutting the waiting does.

CodingOther hands-on workWaiting
1.25×faster end to end

With coding 3× faster, a change that took 100 days takes 80.

Amdahl's law: only the part that gets faster gets faster. The assessment does the same calculation from your own stage timings.

The idea in three lines ​

  1. Development speed and delivery speed are different quantities. AI mostly raised the first.
  2. The slowest stage sets the pace, and the constraint moves when you relieve it.
  3. Verification and human attention become the scarce resources. Readiness depends on how well the system around the code absorbs the speed.

The first line is not new: DORA's 2025 report calls AI an amplifier of an organization's existing strengths and weaknesses. ABSORB is a way to act on it. See prior art.

What you get ​

An offline assessment that runs in your browser and sends nothing anywhere:

  • a readiness level and your weakest of eight dimensions
  • a pipeline view of where a typical change waits, with what-if sliders
  • which autonomy tiers your evidence supports
  • first moves for your two weakest dimensions
  • a comparison with a colleague's scores, and a decision record you can check at the next assessment

Take the assessment

Status

ABSORB is a proposed framework, not a validated standard. The levels, tiers and thresholds are proposals, and the validation plan says in advance what would count against them. Case studies and critique are the most useful contributions.