Compendium · Compositions · Comparison Strips

Comparison Strips

Items

of 1002 total

Full-featured composed scenes — one per row so each design has room to breathe. Click for the live render, the spec, and the source.

Featured 14 Atoms 762 Compositions 240

№ 111

Agile vs DoCoDeGo

Scene STA

The translation

The values inherit. The bottleneck moves.

DoCoDeGo does not contradict Agile — it restates each value for a world where AI carries the build.

Agile (2001)

→

DoCoDeGo (2026)

Why the shift

Working software over documentation

DO Intent over implementation

Specs are now the command interface for AI agents.

Individuals and interactions over processes

CO Direction over production

AI agents are participants — humans direct, machines produce.

Responding to change over planning

DE Flow over releases

AI collapses build time; the constraint is validation.

Customer collaboration over contract negotiation

GO Governance over process

Active alignment replaces passive compliance.

№ 112

Spec Quality Comparison

Scene STA

ICS · same intent

One score. Two specs. Same feature.

Both teams set out to ship password reset. Only one wrote down what "done" meant.

SPEC-074 · draft

Vague intent

ICS 42

Intent: Let users reset password.; ✕
Acceptance: —; ✕
Constraints: —; ✕
Failure modes: —; ✕
Owner: —; ✕

SPEC-074 · approved

Complete intent

ICS 92

Intent: Recover account access without contacting support, while resisting credential-stuffing.; ✓
Acceptance: 6 criteria — token TTL, rate limit, audit log, error states, mobile flow, negative path.; ✓
Constraints: SES quota · token ≤ 15 min · no PII in email body.; ✓
Failure modes: 4 modes — quota exhaustion, replay, race on dual reset, mailbox bounce.; ✓
Owner: @auth-platform; ✓

№ 113

AI With Spec vs Without

Scene STA

Same model · same task claude-4.5-sonnet

AI never writes wrong code. It writes code that's right about the wrong thing.

prompt: "build password reset" Drift


    01
    def reset_password(email):
    
  

    02
        user = User.get(email=email)
    
  

    03
        new = secrets.token_urlsafe(8)
    no TTL
  

    04
        user.password = hash(new)
    silent overwrite
  

    05
        user.save()
    
  

    06
        send_email(email, f"new pw: {new}")
    PII in body
  

    07
        return {"ok": True}

prompt: SPEC-074 @ v1.2 (ICS 92) Aligned


    01
    def request_password_reset(email):
    
  

    02
        if not rate_limiter.allow(email):
    SPEC §AC-3
  

    03
            return RESET_THROTTLED
    
  

    04
        token = mint_token(ttl_seconds=900)
    SPEC §C-1 (≤15 min)
  

    05
        audit.log("reset_requested", email)
    SPEC §AC-5
  

    06
        mailer.send_template("reset", to=email,
    no PII in body
  

    07
                             vars={"url": reset_url(token)})
    
  

    08
        return RESET_SENT

№ 114

Sprint vs Flow

Scene STA

DE pillar · cadence 8-week window

Sprints wait for the ceremony. Flow releases on validation.

Agile · sprint cadence 6 releases · scheduled

S1 release

S2 release

S3 release

S4 release

S5 release

S6 release

DoCoDeGo · flow cadence 18 releases · gate-driven

stream release on green

Sprint result

Value waits in queue between ceremonies.

Flow result

Value moves at the speed of verification.

№ 115

Governance On vs Off

Scene STA

GO pillar · drift same project · 8 checkpoints

Governance does not slow you down. It clips drift.

governance · OFF

Drift accumulates

incident

drift threshold

governance · ON

Drift clipped at gates

shipped

drift threshold

№ 1141

Agile vs DoCoDeGo — Dimension Table

Scene STA do

DO pillar · dimension table

Eight dimensions. Two operating models.

Same engineering goals, different leverage. Each shift belongs to one pillar.

Dimension

Agile

DoCoDeGo

Owner

Primary artifact

Working software

Approved specification

Cadence

Two-week sprints

Continuous validation flow

Build labour

Humans, paired

AI agents, directed

Planning unit

Story points

Intent + acceptance

Quality gate

Definition of done

ICS ≥ 80 + GTR run

Change response

Reprioritise the backlog

Edit the spec, recompile

Governance posture

Trust the team

Bounded autonomy

Failure handling

Retrospective

AAR → spec patch

№ 1142

Waterfall vs DoCoDeGo

Scene STA de

DE pillar · cadence shift

Waterfall lines up. DoCoDeGo loops.

Same six concerns. One model serialises them; the other compiles them.

Waterfall · serial gates ~14 mo · rework: high

P1 Reqs

P2 Design ↶ rework

P3 Build ↶ rework

P4 Test ↶ rework

P5 Deploy ↶ rework

P6 Operate ↶ rework

DoCoDeGo · spec loop ~6 wk · rework: bounded

Intent

Spec

Compile

Verify

Release

Reconcile

Waterfall outcome

Rework discovered after the gate has closed behind you.

DoCoDeGo outcome

Rework happens at the spec, before code is compiled.

№ 1143

Velocity vs Alignment

Scene STA go

GO pillar · twin metrics

Velocity climbs. Alignment falls.

Four quarters at the same team. Throughput trended up; the shipped product drifted from the approved spec.

Velocity (story points / wk)

▲ +31%

Alignment (spec match %)

▼ −34%

// the governance question

Throughput is not progress when the destination drifts.

№ 1144

Throughput vs Quality

Scene STA de

DE pillar · two-axis flow

Throughput is half the story. Escape rate is the other half.

Eight weeks of ships. The gated team got faster without getting worse.

With validation gates · DE-managed escape rate · bounded

Without validation gates · velocity-only escape rate · climbing

low escape

high escape

low throughput

high throughput

№ 1145

Spec vs Production Diff

Scene STA do

DO pillar · reconciliation diff

What the spec required. What production shipped.

SPEC-074 · ICS 92 · six requirements · two drifted, one absent.

Spec requires

Prod ships

Status

Token TTL ≤ 15 min

Token TTL = 15 min

MATCH

Rate limit 5/IP/hr

MATCH

Audit log: every issue + every redeem

Audit log: redeem only

PARTIAL

Negative path: expired token error page

Generic 500 on expired token

PARTIAL

Mobile flow: deep-link to app

—

MISSING

No PII in email body

MATCH

№ 1146

Autonomy Tier Comparison

Scene STA go

GO pillar · autonomy ladder

As autonomy rises, four levers move at once.

Bounded autonomy is not "less review" — it is review concentrated where it counts.

Assisted

Supervised

Bounded

Strategic

Approval cadence

every change

Approval cadence

every change

Approval cadence

sampled batch

Approval cadence

exception only

Sampling rate

100%

Sampling rate

100%

Sampling rate

25%

Sampling rate

Kill-switch

any reviewer

Kill-switch

any reviewer

Kill-switch

on-call

Kill-switch

platform owner

Review time

~12 min

Review time

~4 min

Review time

~30 sec

Review time

~0 sec

№ 1147

Hat Distribution by Team Size

Scene STA co

CO pillar · hat distribution

Every team carries all four hats — in different proportions.

Solo operators spec and build. Orgs concentrate GO at the platform.

Solo

1 person

DO 40% CO 35% DE 15% GO 10%

Squad

4–8 people

DO 30% CO 30% DE 25% GO 15%

Tribe

25–50 ppl

DO 25% CO 25% DE 25% GO 25%

Org

500+ ppl

DO 20% CO 15% DE 25% GO 40%

DO · documentation

CO · construction

DE · delivery

GO · governance

№ 1148

Tool Stack — Before / After

Scene STA co

CO pillar · stack migration

The tools stay. Their roles change.

Five categories of toolchain, each rewired around specs and gates rather than stories and sprints.

Governance Cost Comparison

Scene STA go

GO pillar · cost of oversight

Governance is not the tax. Ungoverned remediation is.

Same five activities, monthly hours. Bounded autonomy front-loads the small cost and avoids the large one.

Model A · ad-hoc oversight

Reactive · after the fact

Approval reviews: 60 h/mo
Audit prep: 35 h/mo
Incident remediation: 90 h/mo
Compliance evidence: 40 h/mo
Spec & ICS upkeep: 0 h/mo

Model B · bounded autonomy

Front-loaded · at the spec

Approval reviews: 18 h/mo
Audit prep: 6 h/mo
Incident remediation: 22 h/mo
Compliance evidence: 10 h/mo
Spec & ICS upkeep: 14 h/mo

№ 1150

Drift by Quarter

Scene STA do

DO pillar · drift over time

Drift bounded versus drift compounded.

Two teams, identical specs. One reconciles every quarter; the other waits "until it matters."

Reconciled · spec patched per quarter peak drift: 14%

Q1 6%

Q2 12%

Q3 4%

Q4 9%

Q5 14%

Q6 5%

Q7 11%

Q8 7%

Unchecked · spec never reconciled peak drift: 64%

Q1 6%

Q2 14%

Q3 22%

Q4 31%

Q5 39%

Q6 48%

Q7 56%

Q8 64%

№ 1151

Team Shape — Startup / Scale / Enterprise

Scene STA co

CO pillar · team-shape map

Three stages. Three shapes. Same framework.

DoCoDeGo absorbs scale by re-distributing the four hats — not by changing them.

1–10 eng

Startup

Humans per AI: 1 : 3
Hat overlap: all four · per person
Spec ownership: founder + tech lead
Governance: lightweight · trust-first

40–150 eng

Scale-up

Humans per AI: 1 : 8
Hat overlap: paired · DO/CO and DE/GO
Spec ownership: product + platform
Governance: bounded · sampled review

500+ eng

Enterprise

Humans per AI: 1 : 20
Hat overlap: specialised · one hat per team
Spec ownership: spec council + domain leads
Governance: tiered · platform-enforced

— Full detail →

—

// source

Alpha · Honest about it

The framework is real.
The community is forming now.

DoCoDeGo is in Alpha. The framework is documented, the practices are battle-tested at small scale, and the next release is being shaped in public.

If it produces anything, it should produce engineers and teams who think more clearly about what they are building and why.

Two doors in

Join the Discord

Discord is where specs are debated, the framework gets sharper, and decisions land in writing. The conversation is the artefact.

Comparison Strips

The framework is real.The community is forming now.

The framework is real.
The community is forming now.