--- title: "Reconstructed Slides: The State of AI Code Quality: Hype vs Reality — Itamar Friedman, Qodo" category: "slides" video_id: "rgjF5o2Qjsc" sourceLabels: ["Cropped public YouTube video frames", "Local OpenCV slide-region detection", "Local RapidOCR"] --- # Reconstructed Slides: The State of AI Code Quality: Hype vs Reality — Itamar Friedman, Qodo ## Source Video [The State of AI Code Quality: Hype vs Reality — Itamar Friedman, Qodo](https://www.youtube.com/watch?v=rgjF5o2Qjsc) ## Method This deck is reconstructed from the existing video frame captures by detecting likely slide regions with OpenCV, cropping/upscaling those regions, deduplicating similar crops, and OCRing the cropped slide images locally. It is a cleaner companion to the full-stage frame deck. ## Reconstructed Slides ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-001.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-001.html) - AI slide classifier: `title_card` confidence `0.99` - Text source: agent_vision. Slide text: > The State of AI Code Quality > Hype vs. Reality > Itamar Friedman > Qodo CEO and Co-Founder ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-002.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-002.html) - AI slide classifier: `content_slide` confidence `0.99` - Text source: advanced OCR `rapidocr-live/bright-screen/contrast`. - OCR decision: ready — Dense small text and embedded screenshot content are better handled by OCR. Slide text: > 3 Cloud Outages in 3 Weeks > Companies are pushing to move faster... OFFLINE CLOUD > Companies are reporting Al-generated code grows: 10% → 30% → 50% ERROR 503 > Related? > each commit is generated or shapedby Al 65%of developers say atleast a quarterof > nowAl-influenced. 15%saythatmorethan80%oftheircodeis > AIE/LEAD REALITY THE STATE OF AI CODE QUALITY: HY > Google DeepMind PRESENTED BY ITAMARFRIEDMAN/ Co-Founder&CEO opob ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-003.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-003.html) - AI slide classifier: `content_slide` confidence `0.98` - Text source: advanced OCR `rapidocr-live/right-72/opencv-adaptive`. - OCR decision: ready — Dense code/prompt screenshot text is best recovered with OCR. Slide text: > @6 > C2 (m nMrmo O 田 > QowH > ppo1-pncy dutcription > reaoe show.'l, Red, Glcb, Grtp, Ls, Tmt Ershigz atf.7, Boshtgh tatur1, Bsthgk bg-1, Bsshigk show.1, Bauxot Corgiere a tocurty revien ot the perding changrs > B > Rrriew the corghle drt sbove, Thi's conlars Mi cede chunges in de PR > Dotenas, Th's hs not a gthera code revina tocus ONY on soourty inoicationn nemy sodtd by ths Pe, Do rot conmant cn eisdng Eeourty corcerra. CBJECTrve: Perlorn a socurty-iooused code rn'ew to koentty HKH CONFioENCE securty Yanerbres tt coud he res eplozioon > CATCAL INSTAUCTONS: > 2. Ayoo Noise: Sxp theorecca ssues, stye concens, er low-tmoxct rrdngs. MrivizE FAlsE POsrTMes: Oety nhg bsues where youre >Box conrdent of scted eapkrscuty A. Focus Ot mpAcT: Prorhr+ Turesbhaes thst coud lad to unuthorred socess, dra brtchm, or synen coeepr L ExcluSoNs: Do NoT report the fclow'no ssue typet9 > Secrets or senstre data stared on dhsk (uhese are handled by othot oroctsets) > ITY: HYPE VS. REALITY THE STATE OF AI > AMAR FRIEDMAN / Co-Founder & CEO opob ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-004.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-004.html) - AI slide classifier: `content_slide` confidence `0.99` - Text source: advanced OCR `rapidocr-live/bright-screen/opencv-adaptive`. - OCR decision: ready — Dense agenda text and embedded report screenshot are better handled by OCR. Slide text: > godo Part l: State of Al codlng adoptlon > Agenda > 0 Sources: @Qodo,@sonar.@Faros codequality/reviewrelatedreport: > Of pull requests, and billions of lines of code. Sample size: thousands of developers, millions > AIE/LEAD Katelyn Lesse. ANTHROPIC Dan Shpper EVERY AsafBord Northwestern Mutual Michele Catasta Preplit MeKinsey Natasha Maniar &Company > Google DeepMind Presented Sy OP\C CURSOR Cognition FACTORY Graphite:e04j au ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-005.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-005.html) - AI slide classifier: `content_slide` confidence `0.98` - Text source: agent_vision. Slide text: > Engineering the future of AI > Productivity > Investment > Learning System Focused on Quality > With Agentic Quality > Agentic Code Gen > Code Gen ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-006.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-006.html) - AI slide classifier: `content_slide` confidence `0.99` - Text source: agent_vision. Slide text: > The Quality Crisis > How it Begins > 1. AI code gen tools adoption is growing > 2. AI productivity doesn't guarantee quality. > 3. Teams lack frameworks for AI-era quality assurance. ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-008.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-008.html) - AI slide classifier: `content_slide` confidence `0.98` - Text source: agent_vision. Slide text: > The Catch and Crisis in Vibe Coding > 1 More code > 2 More PRs > 3 Not less bugs/issues per line of code. (If not more). > 4 Lines of codes are written faster than a developer can review. ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-009.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-009.html) - AI slide classifier: `content_slide` confidence `0.97` - Text source: advanced OCR `rapidocr-live/right-72/opencv-adaptive`. - OCR decision: ready — dense multi-column slide with small text and bullet lists Slide text: > amsFace uality Issues That Modern Engineering qodo > Code-Level Problems Process:Level challenges > Functional Learning 0 Krowledge gaps n A-generated solulions > Non-functional Verification > O Secunty vdnerabihities n Algenerated code O Insufficitnt test coverage for new code O Poor error handing and edge case covcrage O (rconsistent coding standards and patteins Techncal debt accumulation Guardrails 0 Lack of Al output vakdation 0 Difficulty trackirg code provenance 0 Rushed or ncompkete code revevrs 0 tnsufficent guardrals for Al tools > O Integration and compatibility issues O (nadequate documentation Standards 0 Inconsistent quslity standards across teams > HQodo Stute of Al Code Cussty Report & SorrSourco Stste ot Code > ATE OF AI CODE QUALITY: HYPE VS. REALI > AMAR FRIEDMAN / Co-Founder & CEO opob ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-011.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-011.html) - AI slide classifier: `content_slide` confidence `0.96` - Text source: agent_vision. Slide text: > Code review as a gateway & gatekeeper to improve and govern code-level & process-level quality > Remembers, code review isn't only for code quality, it is also in order for the team to own and learn the code together ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-012.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-012.html) - AI slide classifier: `content_slide` confidence `0.99` - Text source: agent_vision. Slide text: > Why is AI Code Review Important? > More productive, higher quality > 2x > When using AI code review there are double quality gain > 47% > More productive when using AI code review ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-013.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-013.html) - AI slide classifier: `content_slide` confidence `0.99` - Text source: agent_vision. Slide text: > Qodo Code Review Stats > Guardrail that detects critical issues, boosts productivity > Sample size: ~1m PRs > 17% > Of PRs contained issue with high severity (9-10) > 80% > When AI review tool is enabled, 80% of PRs don't have any human comment or review ![[assets/reconstructed-slides/rgjF5o2Qjsc/slide-014.jpg]] - Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/rgjF5o2Qjsc/slide-014.html) - AI slide classifier: `content_slide` confidence `0.97` - Text source: advanced OCR `rapidocr-live/right-72/contrast`. - OCR decision: ready — Three-column slide with smaller bullet text; OCR is likely cheaper and more reliable than manual transcription in this pass. Slide text: > The Path Forward > Competitive Advantage Quality is a Al is a Tool, Not a Solution Iterate Tomorrow Start Today. > Establish Quaity Baselines.Multi·agent System.Your Context Engine > ShiftReviews Leftof Commit Code& TooksGovernance Al CodeReview > Document&Embed BestPractices Build Your Quality-specific MCPs.A QualityWorkflows > .Monior > opob > ATE OF AI CODE QUALITY: HYPE VS. REALI > AMARFRIEDMAN/ Co-Founder&CEO opob ### Hidden Non-Slide Evidence - [`slide-007.jpg`](/assets/reconstructed-slides/rgjF5o2Qjsc/slide-007.jpg) — `speaker_stage` confidence `0.99`; speaker on stage, not a readable slide - [`slide-010.jpg`](/assets/reconstructed-slides/rgjF5o2Qjsc/slide-010.jpg) — `speaker_stage` confidence `0.99`; speaker stage shot, no readable slide content - [`slide-015.jpg`](/assets/reconstructed-slides/rgjF5o2Qjsc/slide-015.jpg) — `speaker_stage` confidence `0.99`; Speaker-on-stage wide shot with audience; no readable slide content. Classification audit: `raw/sources/slide-ai-classification/reconstructed/rgjF5o2Qjsc/audit.json`