00 / 18
§0 // THE AGENTIC CODING FRONTIER
ANTIGRAVITY IDE // AGENT MANAGER
auto_awesome Gemini 3.1 Pro (High)
code recommender.go close
data_object fifa_catalog.json
description spec.md
1package api
2
3import (
4 "context"
5 "fmt"
6 "ruby-tv/models"
7)
8
9// RecommenderEngine generates personalized rails.
10type RecommenderEngine struct {
11 Catalog *models.ContentCatalog
12 Cache *models.RedisCache
13}
14
15// GetSeasonRecommendations applies active seasonal triggers.
16func (e *RecommenderEngine) GetSeasonRecommendations(ctx context.Context, userID string) ([]models.Item, error) {
17 profile, err := e.Cache.GetUser(ctx, userID)
18 if err != nil {
19 return nil, fmt.Errorf("failed fetching user %s: %w", userID, err)
20 }
21
22 // TODO: Inject FIFA World Cup recommendation logic...
23 if profile.HasSportsAffinity() && models.IsFootballSeason() {
24 // [AGENT SYNTHESIZING CODE...]
terminal AGENT MANAGER // ORCHESTRATOR
ACTIVE STREAM
sanchitalekh@ (User Prompt) 12s ago

"it's football season, make sure we are recommending the FIFA world cup to our users who like other sports content etc."

sync Gemini 3.1 Pro // Thinking Process...
> Inspecting models/fifa_catalog.json and user affinity rules...
> Identifying target injection point: GetSeasonRecommendations() in api/recommender.go
> Generating AST modifications to inject conditional FIFA World Cup rail...
> ERROR: Unhandled pointer dereference while resolving recursive catalog triggers.
error Agent terminated due to error

You can prompt the model to try again or start a new conversation if the error persists.

See our troubleshooting guide for more help.

arrow_back 0 Files With Changes
checklist Review Changes
Ask anything, @ to mention, / for workflows
add expand_less Planning expand_less Gemini 3.1 Pro (High)
mic
arrow_forward
SOFTWARE DEVELOPMENT IS AT A CROSSROADS

§1: The Shift

Developers can write 10x more code. But is this really leading to 10x better outcomes? We need to change how we think about code (and coding).

bolt
"Code is no longer the destination. It is becoming an opaque intermediate target—a transient byproduct generated and checked by systems."
A SOFTWARE ENGINEER AT GOOGLE
EVOLUTION_STEP_01 PRE-AGENTIC

Development Today

Code is written by hand (or with the help of Generative AI), but the AI never grasps the full extent of the problem.

EVOLUTION_STEP_02 CURRENT_SHIFT

Spec-Driven Intent

Micro-decisions recorded in declarative protocols. Code generation and verification are fully automated.

ACCELERATION VS ARCHITECTURE

§2: The Illusion of Speed

For decades, the holy grail has always been speed: how do we go from idea to execution faster? We invented higher-level languages, frameworks, agile methodologies, and CI/CD pipelines—all in the pursuit of writing and shipping code faster.

speed
THE GENERATIVE AI PARADIGM SHIFT

Then, Generative AI arrived, and suddenly we had the ultimate speed machine. You can now prompt an LLM and watch it spit out 1,000 lines of complex boilerplate in seconds.

"It feels like magic. It feels like we finally won."

warning
THE TERRIFYING REALIZATION

Speed cuts both ways. When managed with discipline, AI coding dramatically accelerates development; without strict architectural guardrails, however, we risk generating unmaintainable code that silently degrades our foundation over time.

01 / THE SPEED TRAP electric_bolt

1,000 Lines in Seconds

Prompting an LLM generates massive boilerplate instantly, creating an intoxicating illusion of victory while bypassing architectural comprehension.

⚠️ UNCHECKED SYNTHESIS HIGH VOLUME
02 / THE INVISIBLE COST visibility_off

Velocity Without Visibility

When code generation outpaces human review, the gap between "what the code says" and "what the system actually does" expands exponentially.

⚠️ ENTROPY EXPANSION EXPONENTIAL GAP
§2.1 // PRODUCTIVITY TIMELINE // VIEW 1: THE PRESENT
INTERACTIVE TIMELINE: CLICK TO SWITCH ERA
???? The Precedent
arrow_forward
1972 C & UNIX Era
2000s Frameworks Era
2026 Agentic Coding

Agentic Coding Births the 10x Programmer

Through autonomous multi-agent synthesis and specification-driven development, a single engineer now commands the throughput of an entire traditional engineering organization.

2026 Agentic Coding & The 10x Programmer
THROUGHPUT: 10,000+ TOKENS / SEC SPECIFICATION ARCHITECTURE
?
psychology_alt
QUESTION

"Have we ever witnessed this kind of an explosion in programmer productivity in human history before?"

At first glance, this exponential velocity feels totally unprecedented. But is this moment truly unique?

§2.2 // PRODUCTIVITY TIMELINE // VIEW 2: THE PRECEDENT
HISTORICAL PRECEDENT // THE 1000x EXPLOSION
1952 A-0 Compiler
1972 C & UNIX Era
2000s Frameworks Era
2026 Agentic Coding
arrow_forward

Yes! Grace Hopper & The First Compiler

In the early 1950s, Grace Hopper revolutionized computing by inventing the first compiler (`A-0`), a breakthrough that transformed computers from glorified calculators into accessible, general-purpose tools. This triggered a >1000x productivity leap.

Admiral Grace Hopper & The A-0 Compiler in the 1950s
1952: THE A-0 COMPILER FROM CALCULATOR TO PLATFORM
THE >1000x EXPLOSION METRICS 1950s MAINFRaMEs → 1970s SOFTWARE ENGINEERING
dns
NUMBER OF COMPUTERS
~250 to 1,000 in 1950s arrow_forward ~150,000 to 200,000+ in 1970s
200X GROWTH
groups
NUMBER OF PROGRAMMERS
A few hundred scientists & clerks arrow_forward 150,000 to 200,000+ corporate SWEs
>100X EXPANSION
code PRIMARY LANGUAGES & ABSTRACTION SHIFT
1950s ERA (LOW-LEVEL) Machine code, Assembly, early FORTRAN, A-0
1970s ERA (HIGH-LEVEL) COBOL, FORTRAN, BASIC, C, Pascal
lightbulb Just as compilers abstracted machine code into human syntax in 1952, Agentic Orchestration abstracts syntax into specifications in 2026.
The Terrifying Realization

§3: Auto-scaling Our Mistakes

"With AI, we aren't just writing code faster. We are auto-scaling our mistakes. If we are using AI to go fast, but aren't maintaining a high signal-to-noise ratio, we are actually going backwards in productivity."

01
trending_down NEGATIVE VELOCITY

Backwards Productivity

If we use AI to go fast without maintaining a high signal-to-noise ratio, our velocity is illusory. We are actively moving backwards as unverified logic piles up.

02
rule_folder RATIO DEFICIT

The Crappy Code Ratio

If the rate of crappy code being generated is greater than the rate of productive, verified code, every prompt actively degrades our foundation.

03
destruction SYSTEM COLLAPSE

The Weight of Slop

Without strict specification guardrails, our projects will eventually and inevitably collapse under the weight of their own unreadable, unmaintainable slop.

CORE TAKEAWAY // ARCHITECTURAL DISCIPLINE
"For each desired change, make the change easy (warning: this may be hard), then make the easy change."
format_quote
Kent Beck Software Design Pioneer
4
ARCHITECTURAL RESILIENCE // BENCHMARK

§4: The Blank Slate Test

?
psychology_alt
Rule-of-thumb: Is your codebase ready for agentic development?

Ask yourself: Imagine you lost all your code. Are you able to re-bootstrap your project from scratch without existing code?

Are you effectively able to generate the 'how' from the 'why' and the 'what'?

THE SYNTHESIS PIPELINE // ZERO CODE DEPENDENCY
UPSTREAM // 01
lightbulb
THE WHY

Intent & PRD

Business outcomes, user needs, and core value proposition stored in immutable markdown.

📄 prd.md // Immutable
+
UPSTREAM // 02
architecture
THE WHAT

Spec & Guardrails

Declarative schemas, architectural boundaries, API contracts, and invariant rulesets.

📄 design.md // Blueprint
arrow_forward
BYPRODUCT // 03
auto_awesome
THE HOW

Disposable Code

Synthetic logic generated on-demand. If wiped, the engine re-compiles identical behavior instantly.

⚡ 100% RECOVERABLE verified
info If your system passes this test, code is no longer a liability—it is an auto-regenerated asset.
RESILIENCE SCORE: 100%
Presenter

Sanchit Alekh

AI Customer Engineer // Google

Bringing the best of Google's AI Technologies to build intelligent solutions for customers.

military_tech

Attribution The original "Design is the New Code" concept and framework was created and pioneered by Dave Rensin, Distinguished Engineer at Google.

Sanchit Alekh
Phase V // Drift Dynamics

§5: THE EAGER
RUNAWAY

It is tempting to dive right into syntax. But AI models are eager to please—which means they will happily sprint off an architectural cliff if you don't anchor them first.

block

Stop Thinking About The Code

Before writing a single line of code, focus entirely on assembling your upstream context artifacts.

Your design documents, user journeys, architecture diagrams, and even developer debates from Slack are the fuel. This human judgment is the exact ammunition the agent needs to write correct-by-construction code for you.

priority_high CODE IS A BYPRODUCT OF PRE-CODE INTENT ARTIFACTS
UPSTREAM CONTEXT HUB // PRE-CODE ARTIFACT ASSEMBLY
● 4 ARTIFACTS LOADED FOR AGENT
description prd_and_design.md
PRD & DESIGN
# Goal: FIFA World Cup Recommender Rail
# Rule 1: Must check user sports affinity.
# Rule 2: Strict geo-licensing check before render.
✓ INTENT VERIFIED 2.4 KB
route user_journey.mermaid
USER JOURNEYS
[User App Open] ──> [Fetch Profile]
│── If SportsAffinity == True ──>
└── [Inject FIFA Rail @ Position 1]
✓ FLOW MAPPED 1.1 KB
architecture architecture_decisions.yaml
ARCHITECTURE
cache_layer: Redis Cluster v7
circuit_breaker: Trip @ 50ms latency
fallback: Generic Trending Rail
✓ BOUNDARIES LOCKED 1.8 KB
forum slack_eng_debates.log
TEAM CHAT / DEBATES
@sanchitalekh: Don't block main feed render if sports service times out!
@drensin: 100%. Circuit breaker must fail open so user never sees blank screen.
✓ CRITICAL NUANCE CAPTURED 18 SHADOW DECISIONS
sync
AGENT SYNTHESIS ENGINE Feeding 4 Pre-Code Artifacts → Synthesizing 100% of Code...
ZERO MANUAL CODING NEEDED auto_awesome
Phase VI // The Liminal State

§6: The Vanishing Code

When code becomes opaque, the only artifact that matters is the design.

shift

Micro-Decisions Critical

SHIFT LEFT

Historically, human developers made thousands of subtle architectural judgments right at the keyboard, leaving a rich trail of micro-decisions embedded inside the logic.

When AI writes the code, that human trail vanishes. If we do not explicitly capture those judgments upstream inside our specifications, the system loses its architectural soul.
01 // UPSTREAM GOVERNANCE verified
image

A Future without Code

DIRECT-TO-BINARY

When you ask AI for a picture of a cat, it does not generate a Python script automating MS Paint to draw it—it outputs the pixels directly from intent.

Software is following the exact same curve. Neural compilers will soon synthesize binaries directly from formal specifications, rendering manual syntax as obsolete as assembly code.
02 // THE PIXEL ANALOGY bolt
ac_unit

Design is Paramount

100X VELOCITY

Even today, LLMs generate code at a velocity and volume that makes traditional line-by-line human code review mathematically impossible.

Whether hidden in a blizzard of 100,000 synthetic lines or a direct-to-binary executable, when code becomes opaque, the design document becomes our sole point of leverage.
03 // SOLE SOURCE OF TRUTH verified_user
!
THE FUNDAMENTAL SHIFT IN ENGINEERING LEVERAGE

We must stop managing the byproduct (syntax) and start governing the source of truth (design).

⚡ DESIGN = EXECUTABLE
The Final Synthesis

DESIGN IS
THE NEW CODE

§7

When the "how" is automated, the "what" is the only remaining lever. Structure, intent, and interface become the source of truth.

PARADIGM SHIFT
THE HOW

§8: FEED THE ELEPHANT, TEST AGAINST THE GOLDFISH

THE ELEPHANT MODEL

Long-term structural memory. The architectural bedrock (`spec.md`, `design.md`) that remains immutable through version cycles. It is the "What" that defines the soul of the product.

THE GOLDFISH MODEL

Ephemeral context window. The immediate syntax, local variables, and transient logic that shifts with every prompt. High-velocity, low-retention.

RETENTION BEDROCK
STORAGE CAPACITY ∞ IMMUTABLE SPECS
TRANSIENT WINDOW
CONTEXT LIFETIME 8k - 128k EPHEMERAL

§9: DECOUPLING
INTENT

We move from a single monolith of code to a tiered Specification Stack. Each layer serves a specific cognitive function for the orchestrator.

LAYER_01 STRATEGIC

prd.md

The product requirements document. Defining user needs and business constraints without technical pollution.

Immutable Truth
LAYER_02 ARCHITECTURAL

design.md

The structural blueprint. Defining component relationships, state machines, and API contracts.

Logic Mapping
LAYER_03 SYNTHETIC

GEMINI.md

The execution bridge. LLM-specific instructions, prompting strategies, and tactical implementation guides.

Actionable Syntax
The Closing Insight

"When intent is decoupled from implementation, the cost of change approaches the speed of thought."

READ FULL PAPER open_in_new
10
Cognitive Architecture // Verification Loop

§10: Elephant and Goldfish

How do you know if your design doc is actually good, or if it's just relying on the hidden context built up in your session? You test it against a Goldfish.

Teaching & Growing the Elephant

THE ABSOLUTE SOURCE OF TRUTH
PHASE 01

We create the Markdown document that will serve as the absolute source of truth and guardrails for the code. Do not ask the AI to write the whole doc at once due to output token limits. Build it iteratively in the same chat session:

01 / THE PROBLEM

Have the AI write a plain English description of the business problem.

02 / THE TECHNICAL PLAN

Have it write a jargon-light description of the big components and how they fit together.

03 / ALTERNATIVES

Document the ideas you considered but ruled out during Phase 1.

04 / DETAILED IMPLEMENTATION (CRITICAL)

Demand that the AI enumerate every single file that will be created or changed, and the rationale for why.

save SAVE THIS FILE TO YOUR TREE. THIS IS YOUR NEW SOURCE CODE.

The Goldfish Protocol

STEP 5: THE COMPREHENSION TEST
VERIFICATION

This is where the magic happens. Start a brand new, empty AI session (the Goldfish) with zero historical context. Give it your Markdown doc and say:

PROMPT // GOLDFISH SESSION

"Read this document and the files it references. Tell me what it's trying to accomplish, and how my system currently works as it relates to this feature."

sync_problem

Do Not Skip This Loop: If the Goldfish cannot explain your system based only on that doc, your doc is missing context. Add the details and repeat until it passes.

verified DOCS THAT PASS THE GOLDFISH TEST BECOME BULLETPROOF GUARDRAILS AGAINST SLOP.
PROCEDURAL INTELLIGENCE // THE HARNESS LAYER

§11: Agentic Harnesses for the Win

Harnesses elevate you from the specs to the goal, using intelligence, skills and data.

flag
TARGET ACHIEVED
The Goal: Autonomous System Delivery
arrow_upward
rocket_launch
THE HARNESS ENGINE
Elevation Engine
LEVEL 02
psychology INTELLIGENCE Gemini 3.1 Pro
handyman SKILLS Runbook Logic
dataset DATA Telemetry / State
arrow_upward
description
BASE FOUNDATION
The Specs (What)
ANTIGRAVITY // HARNESS STUDIO V3.1
ACTIVE HARNESS LOOP
⚡ ANTIGRAVITY ENGINE RUNNING SLO: < 15ms
[1. SPECS] Loaded spec.md constraints.
[2. SKILLS] Attached runbook query-watch.skill.
[3. DATA] Injected telemetry fifa_catalog.json.
[4. INTEL] Gemini 3.1 Pro synthesizing target binary...
check_circle GOAL ACHIEVED // ZERO MANUAL CODE
System state perfectly matches declarative intent in 14.2 ms.
HARNESS ELEVATION MULTIPLIER 10,000X LEVERAGE

§12: The Shift Left

Design judgments are no longer downstream. We are forcing micro-decisions into documentation before a single component is rendered.

PRE-SHIFT WORKFLOW

Designer ships Figma → Engineer interprets spacing → QA catches misalignment → Cycle repeats.

AGENTIC WORKFLOW

Designer ships Judgment Doc → Agent synthesizes Spec → UI is rendered correct-by-construction.

gavel
"We don't build features. We build the rulesets that allow agents to build features."
DECISION_LOG_012 2024-11-24
Border-Radius Strategy

"All outer containers must inherit parent curvature minus 4px padding to maintain visual concentricity."

EXECUTION: COMPONENT_CONSTRUCT_ENGINE
DECISION_LOG_013 2024-11-25
Luminance Contrast

"Active states should not exceed 85% luminance to prevent eye fatigue in dark-mode technical environments."

STATUS: APPLIED_TO_ALL_AGENTS
We are all developers architects

§13: Judgement Belongs to the Humans, So Don’t Be Lazy

ACCELERATION VELOCITY

100X Velocity // Zero-Trust Review

AI generates logic at 10,000+ tokens/sec. Rigorous human review is our only defense against high-entropy synthetic slop.

CRITICAL CULTURE SHIFT

Senior Mentorship // Training the Skeptic

Seniors must train juniors to interrogate and critique AI output. We don't need passive prompters—we need rigorous architects.

SYSTEM TELEMETRY:
TOKENS/SEC: 14,200
REVIEW MANDATE: 100% HUMAN RIGOR
Judgement belongs to the humans so don't be lazy
policy
The "TRUST ME" Machine Do not delegate taste, architecture, or responsibility to automated output.
OPERATIONAL PROTOCOL 3-PHASE VERIFICATION LOOP

§14: The Execution Sequence

execution_protocol.sequence
protocol_trace.antigravity
Dev (18%)
Gemini (50%)
Repo (82%)
person Dev
auto_awesome Gemini
folder Repo
Phase 1: Context & Definition
LOOP: Refine until clear (Steps 1-5)
01.
Read context (Source/Docs)
02.
Describe feature
03.
Prompt for Q&A
Ask clarifying questions
04.
Refine requirements
05.
Write full description
Phase 2: Strategy & Alignment
06.
Request technical strategy
07.
Confirm understanding
Request implementation plan
sticky_note_2 If gaps, read more files.
08.
Request implementation plan
LOOP: Convergence
Critique & Discuss
Repeat until converged
Phase 3: Execution
11.
Create FEATURE.md
Drafts content
12.
Refine content
13.
Add detailed plan
Detailed plan & file list
LOOP: Final Review (Steps 14-15)
14.
Read, React, Refine
15.
Create CL & Review
✓ PROTOCOL COMPLETE
verified LIFELINE PROTOCOL AXES: Dev (18%)Gemini (50%)Repo (82%)
100% DETERMINISTIC VERIFICATION
Source Manifest 5 FILES
Cohesion Index 0.982
/demo/ruby/recommender.js
CTX: 4,213 TOKENS
check_circle ✓ WRITTEN TO /DEMO/RUBY/RECOMMENDER.JS
UTF-8 // LF // JS
sync
Live Refresh
Ready (Sandboxed)
RUBY-TV OTT-G1
TARGET: RUBY TV OTT-G1 STATE: INITIALIZED
SANDBOX_VM_04 ISOLATED
[SYS] Initializing blast_radius_shield...
[MEM] Allocation restricted to 256MB...
[NET] Outbound egress denied...
[CPU] Cycle monitoring active...
[LOG] Anomaly detected in slide_14_logic...
Governance & Safety

§15: CONTAINMENT AND THE BLAST RADIUS

In a world of synthetic code, execution must be adversarial. We do not trust the generation; we contain it. Every block of code runs in a disposable sandbox, where its "Blast Radius" is monitored in real-time.

01. ISOLATION

Hypervisor-level segregation for every design-to-code iteration.

02. TELEMETRY

Real-time resource exhaustion monitoring and automatic kill-switches.

HEAL
AUTONOMOUS RESILIENCE & ADAPTATION

§16: Moving towards Self-Healing and Self-Evolving Software

System failures and runtime friction are no longer human pager alerts—they are deterministic feedback signals. When a UI breaks or an assertion fails, the system intercepts the error, diagnoses the root cause against the spec, patches the logic, and redeploys autonomously.

From Reactive Repair to Proactive Evolution

Beyond immediate self-healing, software begins to evolve continuously against production telemetry. By observing real-world user workflows and drift patterns over time, agentic loops propose architectural refinements directly to the specification stack, re-bootstrapping cleaner, faster iterations automatically.

CRITICAL_EXCEPTION:04 (OVERFLOW) DETECTED 12ms AGO
REPAIR_AGENT_INVOKED ANALYZING AST & SPEC...
SYSTEM_RESTORED // ZERO DOWNTIME VERSION 4.2.1-AUTO
EVOLUTION_DELTA: SPEC_V4.3 ADAPTIVE OPTIMIZATION ACTIVE
● DIAGNOSTIC & EVOLUTION LOOP ACTIVE
TELEMETRY_STREAM
99.999%
UPTIME VIA AUTONOMOUS FIX & ADAPTATION
MTTR: 14ms (ZERO HUMAN LOOP)
SPEC DRIFT: AUTO-CORRECTED
The Final Frontier

§17: The Self-Referential Engine

We have reached the point of Singularity in Design. The Agent no longer builds the UI for the User—it builds the Agent that will build the next UI. It is an iterative improvement cycle where the Designer's role is not to create, but to define the Initial Conditions of the engine itself.

READ FULL PAPERopen_in_new
S
R
Closing

Thank You. Questions?

Please provide us feedback:

Feedback QR Code SCAN WITH YOUR CAMERA
HOW TO ACCESS THIS: dnc.gdemos.cloud
T
Y