🟠 Codex 2026-07-04 · OpenAI GPT

DNAfinity — Codex Competitive Analysis & Scorecard (2026-07-04)

วันที่: 2026-07-04 · independent short-window re-run จาก prompt กลาง DNAfinity Competitive Analysis & Scorecard
วิธีให้คะแนน: COMPETITIVE-METHODOLOGY.md · 7 มิติ 0-5, overall = ค่าเฉลี่ยเลขคณิต
บทบาท: INDEPENDENT competitive analyst สำหรับ tri-model consensus รอบ 2026-07-04
Heatmap visual: CODEX_COMPETITIVE-HEATMAP-2026-07-04.html

คะแนนนี้เป็น directional self-assessment เพื่อจัดลำดับ product work ไม่ใช่ marketing claim, analyst coverage, third-party benchmark, หรือ user research. ผมอ่าน code-truth ที่ session นี้เข้าถึงได้ก่อน แล้วค่อยใช้ Claude 07-04 / Claude 07-02 เป็น calibration anchor สำหรับ repo เครื่อง A ที่เข้าไม่ถึงโดยตรง.


0. Executive Summary

รอบ 2026-07-04 เป็น short-window 2 วัน ไม่ใช่รอบ rebuild ใหญ่แบบ 06-18 -> 07-02. แต่มี 2 movement ที่มีน้ำหนักจริง:

แอปอื่นส่วนใหญ่คงคะแนน ไม่ใช่เพราะไม่มีงาน แต่เพราะงานรอบนี้ยังไม่ข้าม threshold ของ rubric: graph/constellation kit redistribution เป็น UX/evidence ที่ดี แต่ส่วนใหญ่เป็น visual layer, branch work, หรือยังไม่เพิ่ม traction.

Portfolio average ขยับจากประมาณ 3.55 -> 3.59. จุดอ่อนร่วมยังไม่เปลี่ยน: D7 Traction/Trust = 1 ทุกแอป. Live หรือ founder self-verify ยังไม่นับเป็น user/pilot จริง.

Strategic conclusion: รอบนี้พิสูจน์ว่า autonomy pattern เริ่มไหลจาก SmeDNA/AppForge ไป WorkDNA และ OperDNA ได้จริง. ถ้าจะทำให้คะแนนรอบหน้าขยับแบบมีความหมาย ต้องทำ 2 อย่างพร้อมกัน: (1) ขยาย safe execution ให้เป็น reusable AppForge kit, (2) ดัน WorkDNA/SmeDNA/OperDNA เข้า named pilot หรือ external trust proof.


0.1 Evidence & Method

Rubric / method

Code-truth since 2026-07-02

Gap/claims docs checked

Found and used app gap/claim material including CareDNA RESEARCH_GAP_ANALYSIS.md / CLAIM_MATRIX.md, OperDNA CLAIMS_MATRIX.md, WorkDNA research gap docs, SmeDNA claim readiness registry, TradeDNA gap docs, EmotiQ claim-readiness matrix, PetDNA/WisdomDNA research gaps.


1. Portfolio Scorecard (0-5)

App Feature AI-native UX/Wow Diff/Moat Ecosystem Thai Traction Overall Δ vs 07-02
WorkDNA 5 5 4 5 5 5 1 4.3 +0.2
SmeDNA 5 5 5 4 4 5 1 4.1 0.0
AppForge 5 5 4 5 5 3 1 4.0 0.0
BloomDNA 5 4 4 4 4 5 1 3.9 0.0
MU-AI Core 5 4 3 4 5 5 1 3.9 0.0
CareDNA 4 4 4 4 4 5 1 3.7 0.0
FinDNA 4 4 4 3 5 5 1 3.7 0.0
MU-AI Game 4 3 4 4 4 5 1 3.6 0.0
WisdomDNA 4 4 4 4 4 4 1 3.6 0.0
DataDNA 4 4 3 4 4 4 1 3.4 0.0
OperDNA 4 4 3 3 4 4 1 3.3 +0.4
EmotiQ 4 4 3 3 4 4 1 3.3 0.0
InfluDNA 3 4 4 3 4 4 1 3.3 0.0
TradeDNA 4 3 4 3 4 4 1 3.3 0.0
PetDNA 4 3 3 3 4 4 1 3.1 0.0
MUAI-Verse 3 2 3 4 4 3 1 2.9 0.0
*Avg market leader* *5* *4* *4* *4* *2* *2* *5* 3.7

Readout


2. Market Frontier Refresh (2026)

  1. Agentic execution remains the frontier

Replit Agent positions itself as building production-ready apps from chat, Zapier Agents says agents can work across 9,000+ apps, Microsoft Power BI added agentic report-authoring skills, Eightfold describes talent agents executing across the lifecycle, and Cleo Autopilot is moving money management toward goal-based automated plans. This validates raising WorkDNA only when actual guarded execution ships.

  1. Trust and guardrails are now table stakes

Human-in-the-loop, policy checks, claim matrices, bounded copy, and audit logs are not overhead. They are the product surface for sensitive domains. This matters for WorkDNA/OperDNA/SmeDNA as much as for health/finance.

  1. Visual graph UX is useful but not sufficient

The constellation kit improves “wow” and cross-app reuse. But market leaders still win by activation, data, integrations, and trust. A graph skin alone should not move D7 or D4 unless it creates real workflow value.

  1. Thai vertical trust remains a defendable wedge

FlowAccount’s Thai SME footprint, Hume’s emotion model depth, Tractive’s hardware/data moat, and Power BI’s enterprise integration show the same pattern: DNAfinity can win where Thai workflows + ecosystem data compound, but must prove external adoption.


3. App-by-App Analysis

3.1 WorkDNA — Workforce Intelligence

Score: 4.3 · Competitors: Eightfold, Gloat, Lattice, 15Five, Darwinbox

WorkDNA gets the only major upward move in this scorecard. The autonomy-kit is no longer just design: backend/app/autonomy_kit, autonomy models/router, plan/check/approve/execute, policy/kill-switch, L3/L4 language, executors for workforce actions, and frontend autonomy console are all visible. This justifies D2=5 under the rubric because AI can now move from insight/draft into guarded execution.

Still not D7>1: no external pilot or enterprise trust proof. Eightfold/Gloat still dominate data/deployment.

Next move: ship one WorkDNA pilot where an AI staffing/rebalance recommendation is approved, executed, audited, and reviewed by a real manager.

3.2 SmeDNA — Thai SME / POS / AI Co-Founder

Score: 4.1 · Competitors: FlowAccount, PEAK, StoreHub, Loyverse, Xero/QuickBooks

SmeDNA stays high. The 07-02 evidence for supervised procurement execution still stands, and the 07-04 work adds graph/guide polish rather than a scoring threshold jump. It remains commercially strong because Thai SME workflows, procurement, tax/accounting, POS/gold/shop depth, and supervised action fit a real buyer.

Gap: FlowAccount/PEAK retain the trust/accountant/user-base advantage. SmeDNA needs a named pilot more than another vertical surface.

Next move: package one Thai SME demo into a measurable 14-day pilot: stock/cashflow/reorder/promo approval loop.

3.3 AppForge — AI App Factory / Platform Engine

Score: 4.0 · Competitors: v0, Lovable, Bolt, Cursor, Replit Agent

Machine-B AppForge is accessible in this session. I verified 11 commits since 2026-07-02 on feat/kit-sync-foreign-apps: Kit Sync for foreign apps, constellation-graph kit, ROI/reforge hardening, and i18n/admin polish. This supports the platform thesis, but does not cross a new scoring threshold beyond the already-credited OIDC/AppForge ID, reverse-engineer, gap-analysis, and kit moat from 07-02.

Gap: public developer traction, external quality benchmarks, and implemented Human DNA Engine tables/endpoints. Kit distribution is more real now; canonical Human DNA graph is still architecture, not code.

Next move: formalize a kit registry with version, owner, adoption status, and regression tests per consuming app.

3.4 BloomDNA — Learning Intelligence

Score: 3.9 · Competitors: Khanmigo, MagicSchool AI, Thai EdTech platforms

BloomDNA gains role-scoped constellation universes and production seed-demo evidence. The privacy design is sensible: classroom-level/k-anon views rather than persisting sensitive child graphs as source of truth.

Still no score movement because the app already had D1/D3 strength, and this work does not yet add school pilot, LMS integration, or measured learning outcomes.

Next move: one Thai classroom pilot with teacher workflow and parent growth card metrics.

3.5 MU-AI Core — Life Intelligence / Human DNA Producer

Score: 3.9 · Competitors: Co-Star, The Pattern, Nebula, Replika, BetterUp

Machine-B MU-AI is accessible in this session. I verified 4 commits since 2026-07-02: Saju visual mount, constellation kit port, and self-graph controls. This improves presentation and kit reuse, but does not add Core-specific engine depth, daily retention, external traction, or a live Human DNA producer API.

Gap: daily retention, share loop, external users, and live cross-app sync.

Next move: Destiny Snapshot + compatibility invite + consented cross-app Human DNA updates.

3.6 CareDNA — Health / Wellness

Score: 3.7 · Competitors: Apple Health, Oura, WHOOP, Function Health, MyFitnessPal

Self-Check Kit v1 is the best new CareDNA evidence: vision acuity, color vision, smell, wellness self-assessment, server-side recomputation, PHI encryption, claim matrix gate, and safe copy. It improves confidence inside the current band.

I do not raise D1 to 5 because the app still lacks clinical validation, wearable/lab moat, expert review, and external trust. Health scoring must stay conservative.

Next move: a non-clinical wellness pilot and explicit validation plan before stronger claims.

3.7 FinDNA — Personal Finance

Score: 3.7 · Competitors: Cleo, YNAB, Copilot Money, Finnomena

FinDNA gets a wealth-network graph polish and trust taxonomy hardening. It remains strong in Thai finance/ecosystem framing, but the market is moving toward money agents and live connectors.

Gap: bank/statement ingestion, PromptPay/payment action, SEC/broker perimeter, and real user trust.

Next move: approval-gated “Fin Autopilot” using the autonomy-kit pattern, starting with safe internal ledger actions.

3.8 MU-AI Game / GENVERSE

Score: 3.6 · Competitors: Finch, Habitica, Co-Star-like gamified apps

No commits in this window. Keep frozen. The prior AR/3D/gameplay depth remains useful, but no new AI-native or traction evidence appears.

3.9 WisdomDNA — Faith / Interfaith AI

Score: 3.6 · Competitors: Hallow, Pray.com, general LLMs

No commits in this window. Keep frozen. The neutral multi-tradition/citation thesis remains promising, but corpus/expert/community trust are still the gating factors.

3.10 DataDNA — Decision Intelligence

Score: 3.4 · Competitors: Power BI Copilot/agent skills, ThoughtSpot, Tableau, Hex

DataDNA has the largest technical UX work after WorkDNA: constellation graph views V2-V11, graph-view endpoints, telemetry, visual/e2e coverage, and a real matched bug fix affecting answer-card rendering. However the current repo is on feat/graph-views-p1-vision with uncommitted work, so I hold the score conservative.

If merged/deployed cleanly, DataDNA likely deserves D3 3 -> 4 and overall 3.6 in the next run.

Next move: merge/deploy the query bug fix first, then score the graph system after production validation.

3.11 OperDNA — AI COO / Ops Automation

Score: 3.3 · Competitors: Lindy, Zapier Agents, n8n, Relay.app, Make

OperDNA is the second real mover. The 07-02 correction was based on “mock execution” and no estate registry. 07-04 directly addresses both: config/apps.yaml, estate registry loader, monitoring/news executors, SSRF guard, arq scheduler, HubEvent writes, and claims matrix that distinguishes real vs simulated domains.

D1 goes 3 -> 4, D2 goes 3 -> 4, D5 goes 3 -> 4. I do not give D2=5 because current real actions are mostly read-only/reversible monitoring/news; email/files/scheduling remain mock or incomplete.

Next move: add one state-changing but reversible executor with approval and rollback, then D2 can be debated again.

3.12 EmotiQ — Emotion Intelligence

Score: 3.3 · Competitors: Hume AI, Observe.AI, Uniphore, Cogito/Verint

The emotion graph constellation port improves presentation. But Hume’s emotion-aware voice stack and model/data moat remain far deeper. No backend model validation or real-time coaching proof changed in this window.

3.13 InfluDNA — Creator OS

Score: 3.3 · Competitors: Jasper, OpusClip, HeyGen, Buffer, Canva

Admin DB provider key and claim/funnel work are useful, but publishing/persona/foresight still need real social APIs and analytics. Keep score flat.

3.14 TradeDNA — Investment Intelligence

Score: 3.3 · Competitors: eToro, Composer, Robinhood frontier, Finnomena

Instrumentation and dashboard surfaces improve evidence, not category position. No broker execution, no legal trust proof, and no live trading integration. Keep flat.

3.15 PetDNA — Pet Life OS

Score: 3.1 · Competitors: Tractive, Furbo, Petcube, Sylvester AI, Petkit

The 9-species threshold/recommendation work is real and already reflected in the prior band. Tractive’s 2026 tracker/health intelligence push underlines why PetDNA still lacks hardware/data/vet moat. Keep flat.

3.16 MUAI-Verse — Web3 Identity / Protocol

Score: 2.9 · Competitors: Galxe, verifiable credential / soulbound identity protocols

No commits. Keep frozen. Strategic but not a current lead horse.


4. Portfolio Synthesis

4.1 Agentic Reality Map

Level Definition Apps
5 · Guarded state-changing execution AI/system can execute domain actions with policy, approval/auto tier, audit, and kill-switch WorkDNA, SmeDNA
4 · Real bounded automation Real read/reversible executors or structured HITL action with scheduler/policy OperDNA, FinDNA, DataDNA
3 · Insight / draft / coach AI explains, drafts, or visualizes; user performs action MU-AI Core, BloomDNA, WisdomDNA, CareDNA, TradeDNA, EmotiQ, PetDNA, MU-AI Game
2 · Mock / simulated execution Agent language exists but real action is not proven InfluDNA publish/social areas; remaining OperDNA email/files/scheduling domains

4.2 What This Round Proves

  1. Autonomy-kit reuse is real: SmeDNA-style procurement guardrails influenced WorkDNA and OperDNA.
  2. Constellation-kit reuse is real: WorkDNA-style graph UX appears across FinDNA, BloomDNA, EmotiQ, DataDNA.
  3. Honest-claims discipline is improving: CareDNA and OperDNA explicitly constrain claims instead of inflating.
  4. Traction is still untouched: no scorecard movement in D7.

4.3 Roadmap: ดี / ว้าว / ฟิน

Level Goal Priority moves
ดี Make top apps pilot-ready WorkDNA pilot packet, SmeDNA SME pilot, OperDNA self-hosted ops pilot, DataDNA deploy bug fix
ว้าว 30-second first value WorkDNA autonomy demo, Core Destiny Snapshot, SmeDNA weekly AI action, DataDNA graph answer trace
ฟิน Cross-app compounding moat AppForge kit registry, Human DNA event ledger, autonomy policy kit, consented app data contracts

Next 30 days

  1. Get one named WorkDNA or SmeDNA pilot.
  2. Merge/deploy DataDNA graph/query fix or split the critical bug fix out first.
  3. Turn OperDNA Phase 0 into a daily founder ops loop with authenticated app health events.
  4. Make AppForge the official kit registry for autonomy + constellation.
  5. Keep D7 strict; do not let internal demos masquerade as traction.

5. Founder + AI Efficiency

Person-month comparisons are directional scope estimates, not facts about competitor teams.

The short-window signal is efficiency of reuse, not raw commit volume. In two days, the estate reused two patterns:

This is the kind of leverage a 1-founder + AI estate needs. The caution is that reuse must become governed versioned product infrastructure, not manual copy/paste.

Area 07-04 evidence Market-equivalent read
WorkDNA 12 commits, real autonomy framework/executors Enterprise HR agents are now the market frontier; WorkDNA finally enters that tier technically, not commercially
OperDNA 2 commits but high leverage: registry, scheduler, real executors Moves from demo shell toward ops platform, still far behind Zapier/Lindy/n8n integrations
DataDNA 17 commits on graph views branch Strong UX/evidence work; not counted upward until merged/deployed
CareDNA Self-Check Kit + claim matrix Good governance; still no clinical/user validation
Portfolio kit redistribution across apps Confirms AppForge ecosystem thesis but does not replace traction

6. Δ vs Claude 07-02

Per prompt, this table compares Codex 07-04 against the 07-02 Claude calibration anchor. I also checked Claude 07-04 after my evidence pass; numeric conclusions converge with the updated Claude 07-04 report.
App Claude 07-02 Codex 07-04 Δ Verdict Reason
WorkDNA 4.1 4.3 +0.2 Up Guarded autonomy moved from planned/bounded to real plan/check/approve/execute with workforce executors. D2 crosses 4 -> 5.
SmeDNA 4.1 4.1 0.0 Hold 07-02 supervised procurement/autonomy score holds; 07-04 changes do not cross a new threshold.
AppForge 4.0 4.0 0.0 Hold Machine-B direct check shows 11 commits, but they reinforce existing kit/platform moat rather than add a new scoring threshold.
BloomDNA 3.9 3.9 0.0 Hold Role-scoped graph work improves UX evidence but not pilot/outcome traction.
MU-AI Core 3.9 3.9 0.0 Hold Machine-B direct check shows UI/kit reuse, not new Core engine depth or retention.
CareDNA 3.7 3.7 0.0 Hold Self-Check Kit is useful and bounded; no clinical/user validation yet.
FinDNA 3.7 3.7 0.0 Hold Wealth graph/i18n hardening does not close bank/agentic-money gap.
MU-AI Game 3.6 3.6 0.0 Hold No new threshold-crossing evidence.
WisdomDNA 3.6 3.6 0.0 Hold No new threshold-crossing evidence.
DataDNA 3.4 3.4 0.0 Hold Significant graph/bug-fix work, but branch/unmerged/deploy state keeps score conservative.
OperDNA 2.9 3.3 +0.4 Up Real registry + scheduler + monitoring/news executors address the 07-02 mock/no-registry correction.
EmotiQ 3.3 3.3 0.0 Hold Graph port does not change model validation or live coaching gap.
InfluDNA 3.3 3.3 0.0 Hold Provider/settings/funnel work does not fix simulated publishing.
TradeDNA 3.3 3.3 0.0 Hold Instrumentation only; broker/trust gap unchanged.
PetDNA 3.1 3.1 0.0 Hold Health thresholds already within current band; no hardware/vet moat.
MUAI-Verse 2.9 2.9 0.0 Hold No new threshold-crossing evidence.

Consensus implication: only WorkDNA and OperDNA numerically move vs Claude 07-02. The operational debate is still simple: count DataDNA after merge/deploy, and convert WorkDNA/OperDNA agentic progress into real pilots.


7. Sources

Method and local evidence

Competitor web refresh


8. Caveats