AI時代の開発生産性は、個人技からチーム設計へ

Developers Summit 2026 Summer・IT Strategy Summit 2026での発表資料です。
https://event.shoeisha.jp/devsumi/20260716

Transcript

1. AI࣌୅ͷ։ൃੜ࢈ੑ͸ɺ ݸਓٕ͔ΒνʔϜઃܭ΁ CodeRabbit த௡઒ಞ࢘ @Developers Summit 2026 Summer #DevSumi
2. 𝕏 @goofmint fb.me/goofmint த௡઒ ಞ࢘ CodeRabbit σϕϩούʔΞυϘέΠτ ࣗݾ঺հ Ұൠࣾஂ๏ਓDevRel ୅දཧࣄ
3. CodeRabbitʹ͍֮ͭͯ͑ͯ΄͍͠3ͭͷ͜ͱ 1.AIίʔυϨϏϡʔαʔϏε 2.ఆֹɾOSS͸ແྉ 3.ΩϟϥΫλʔͷ໊લ͸Hoppy coderabbit.ai/ja
4. ΞδΣϯμ 1.AIࢧԉ։ൃ͕޿͕ΔதͰͷมԽ 2.AIࢧԉ։ൃͷ՝୊ 3.։ൃੜ࢈ੑͱ͸ 4.։ൃ૊৫ͰAIΛ࢖͏ͨΊʹ 5.CodeRabbitͰͷAIར༻ྫ
5. AIࢧԉ։ൃ͕޿͕ΔதͰͷมԽ
6. AIࢧԉ։ൃ͕޿͕ΔதͰͷมԽ 1.γνζϯσϕϩούʔͷ૿Ճ 2.GitHubͳͲͷΞΫηε૿ 3.ηΩϡϦςΟ/඼࣭Ψόφϯεधཁ૿
7. γνζϯσϕϩούʔʢࢢຽ։ൃऀʣͷ૿Ճ 1.اۀͷ63%͕಺੡Խɺ38%͕ࢢຽ։ൃʹऔΓ૊Έ※1 2.ੜ੒AIಋೖࡁΈاۀ͸57.7%※2 3.ϩʔίʔυɾϊʔίʔυπʔϧͷಋೖ͕51.0%ʹ※2 ※1 ΞϓϦέʔγϣϯ։ൃʹؔ͢Δௐࠪ | IT Foresight ※2 ໺ଜ૯߹ݚڀॴɺ೔ຊاۀΛର৅ʹʮIT׆༻࣮ଶௐࠪʢ2025೥ʣʯΛ࣮ࢪ | χϡʔεϦϦʔε | ໺ଜ૯߹ݚڀॴ(NRI)
8. GitHubͷΞΫηε૿ 1.೔ຊͷGitHubར༻ऀ਺͕450ສਓ௒ ※1
 աڈ12͔݄Ͱ100ສਓ૿ɻ2030೥ʹ1,170ສਓ௒Λ༧ଌ 2.PR͸લ೥ൺ29%૿ʢάϩʔόϧʣ※1 3.ެ։commitͷ4%͕Claude Codeʢ2026೥2݄ʣ※2
 2026೥຤ʹ͸20%Ҏ্ʹୡ͢Δ༧ଌ΋ ※1 2025೥൛ OctoverseϨϙʔτ ※2 Claude Code is the In fl ection Point
9. ηΩϡϦςΟ/඼࣭Ψόφϯεधཁ૿ 1.શςετέʔεͷ45%ʹ੬ऑੑؚ͕·ΕΔ※1
 ৽ϞσϧΛར༻ͯ͠΋վળ͞Εͣ 2.AI͕ར༻͢Δ֎෦ϥΠϒϥϦͷ49%ʹط஌੬ऑੑ※2
 ࡢࠓͷαϓϥΠνΣʔϯ߈ܸʹܨ͕Δ໰୊ 3.AI͸ਓͷ1.7ഒɺਂࠁͳ໰୊ΛؚΉPRΛ࡞੒※3 ※1 GenAI and Code Security: What You Need to Know ※2 State Of Dependency Management 2025 | Application Security |… ※3 State of the AI vs. Human Code Generation Report
10. AIࢧԉ։ൃͷ՝୊
11. AIࢧԉ։ൃͷ՝୊ 1.ίʔυϨϏϡʔ 2.AI׆༻εΩϧͷόϥπΩ 3.ROIʢඅ༻ରޮՌʣ
12. ίʔυϨϏϡʔ 1.AIੜ੒ίʔυ૿ͰɺϨϏϡʔ͕ϘτϧωοΫ※1 2.ίʔυϨϏϡʔ୲౰ͷ໿9ׂ͕ɺAIੜ੒ίʔυ ͷීٴͰʮϨϏϡΞʔෛ୲૿ʯΛ࣮ײ※2 3.ϨϏϡʔͷ଴ͪ࣌ؒ΍ରԠ͕࣌ؒ௕Ҿ͖ɺόά ͷൃੜ཰͸54%૿Ճ※3 ※1 [2605.17548] Rethinking Code Review in the Age of AI: A Vision for Agentic Code Review ※2 ʲAIੜ੒ίʔυͷϨϏϡΞʔෛ୲ʹؔ͢ΔௐࠪʳϨϏϡʔ୲౰ऀͷ໿9ׂ͕ʮෛ୲૿ʯΛ࣮ײ ※3 AI Impact on Engineering Productivity: 2026 Report Data
13. ։ൃνʔϜମ੍ ΤϯδχΞ γχΞΤϯδχΞ PM
14. ΤϯδχΞ >> ϨϏϡʔ୲౰ 1.։ൃʮྔʯ͸૿͑ͯ΋ɺϨϏϡʔͰࢭ·Δ 2.ϨϏϡʔ͸ίʔυΛݟΔ͚ͩͷ࡞ۀͰ͸ͳ͍ 3.มߋͷഎܠɺۀ຿ཁ݅ɺઃܭҙਤɺطଘίʔυ ΁ͷӨڹΛཧղ͢Δʢೝ஌ෛՙʣ
15. AI׆༻εΩϧͷόϥπΩ 1.ద੾ͳίϯςΩετΛ༩͑ΒΕ͍ͯΔ͔൱͔※1 2.ݸਓͷMCPαʔόʔɺεΩϧͷઃఆʹґଘ 3.ֶशݩσʔλͷ໰୊
 OSSɺίϛϡχςΟͰެ։͞ΕΔίʔυ͸඼࣭͕ෆ҆ఆ ※1 The Most Common Security Vulnerabilities in AI-Generated Code | Blog | Endor Labs ஌ݟΛݸਓʹด͡ΔͷͰ͸ͳ͘ɺνʔϜͰڞ༗͢Δ
16. ROIʢੜ࢈ੑ͕ຊ౰ʹ޲্͍ͯ͠Δͷ͔ʁʣ 1.AIར༻Ͱ24%ߴ଎Խ͢Δͱ༧ଌ͍͕ͯͨ͠ɺ19%௿ Լ͍ͯͨ͠※1 2.CopilotಋೖͰόά਺͕41%૿ՃɻPRαΠΫϧλΠϜ ͸มԽͳ͠※2 3.ίΞ։ൃऀͷϨϏϡʔෛՙ6.5%૿ʹΑΓɺνʔϜͷ ։ൃੜ࢈ੑ͕௿Լ※3 ※1 Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity - METR ※2 Does GenAI Improve Software Developer Productivity? ※3 [2510.10165] AI-Assisted Programming Decreases the Productivity of Experienced Developers by Increasing the Technical Debt and Maintenance Burden
17. ։ൃੜ࢈ੑͱ͸ʁ
18. ։ൃੜ࢈ੑͱ͸ ଎౓ ඼࣭ ϨϏϡʔ ϑϩʔ ମݧ
19. ։ൃੜ࢈ੑͱ͸ ଎౓ ඼࣭ ϨϏϡʔ ϑϩʔ ମݧ ଎౓ 1.ϦʔυλΠϜ 2.PRϚʔδ࣌ؒ 3.σϓϩΠස౓ ϘτϧωοΫ೺Ѳʹར༻
20. ։ൃੜ࢈ੑͱ͸ ଎౓ ඼࣭ ϨϏϡʔ ϑϩʔ ମݧ ඼࣭ 1.ຊ൪ো֐݅਺ 2.όά࠶ൃ཰ 3.มߋࣦഊ཰ 4.ςετࣦഊ཰ 5.ηΩϡϦςΟݕग़݅਺ ଎͍͚ͲɺΤϥʔ͕ଟ͍Ͱ͸ବ໨
21. ։ൃੜ࢈ੑͱ͸ ଎౓ ඼࣭ ϨϏϡʔ ϑϩʔ ମݧ ϨϏϡʔ 1.ϨϏϡʔ଴ͪ࣌ؒ 2.ϨϏϡʔॴཁ࣌ؒ 3.ίϝϯτղܾ࣌ؒ 4.PRαΠζ 5.ࠩ͠໭͠཰ ϨϏϡʔͷϘτϧωοΫঢ়ଶΛՄࢹԽ
22. ։ൃੜ࢈ੑͱ͸ ଎౓ ඼࣭ ϨϏϡʔ ϑϩʔ ମݧ ϑϩʔ 1.࡞ۀத࣌ؒ / ଴ͪ࣌ؒ 2.ձٞ࣌ؒ 3.ϒϩοΫ࣌ؒ 4.ίϯςΩετεΠον ճ਺ ࣮૷ΑΓ଴ػɾௐ੔ɾׂΓࠐΈʹ ࣌ؒΛୣΘΕ͍ͯͳ͍͔
23. ։ൃੜ࢈ੑͱ͸ ଎౓ ඼࣭ ϨϏϡʔ ϑϩʔ ମݧ ମݧ 1.։ൃऀຬ଍౓ 2.ूத࣌ؒ 3.ೝ஌ෛՙ 4.Ξϯέʔτ ຬ଍౓ɾޮ཰ɾϑϩʔ͸ॏཁͳ؍఺
24. ։ൃ૊৫ͰAIΛ࢖͏ͨΊʹ
25. ։ൃ૊৫ͰAIΛ࢖͏ͨΊʹ 1.ద੾ͳίϯςΩετઃܭ 2.શһͷݟΒΕΔ৔ॴͰAIͱ΍ΓऔΓ 3.඼࣭ͱ҆શੑͷ୲อ
26. ద੾ͳίϯςΩετઃܭ 1.AIʹԿΛ೚ͤΔ͔Λ໌֬ʹ͢Δ 2.AIʹԿͷ৘ใΛ༩͑Δ͔ΛܾΊΔ 3.ਓ͸ԿΛ൑அ͠ɺ੹೚Λ͔࣋ͭΛܾΊΔ
27. શһͷݟΒΕΔ৔ॴͰAIΛར༻ 1.ྑ͍ϓϩϯϓτ΍ࣦഊΛެ։ɺڞ༗ͯ͠஌ݟΛ ஝ੵ͢Δ 2.ීஈ࢖͍ͬͯΔπʔϧͷதʹू໿͢Δ 3.৽͍͠ϝϯόʔʹରͯ͠ɺաڈͷҙਤΛڞ༗͢ Δ
28. ඼࣭ͱ҆શੑͷ୲อ 1.ίϯςΩετ͚ͩͰ๷͛ΔΘ͚Ͱ͸ͳ͍ 2.ίʔυϨϏϡʔ΍ςετɺCI/CDͳͲ඼࣭ήʔ τͰͷAI׆༻ 3.ϩάͱղੳɺݖݶ؅ཧ
29. CodeRabbitͰͷAIར༻ྫ
30. CodeRabbitͰͷAIར༻ྫ νϟοτ Issue ϨϏϡʔ
31. νϟοτʢSlackʣ 1.CodeRabbitͱ͍͏SlackΤʔδΣϯτ 2.Datadog/Sentry/Salesforce/Google Driveͱ઀ଓ 3.SlackͰର࿩ͨ͠ΓɺPylonʹૹΒΕ͖ͯͨ໰͍ ߹ΘͤΛݩʹௐࠪɺIssue࡞੒ɺPR࡞੒·Ͱߦ ͏
32. νϟοτʢSlackʣ 1.ීஈ࢖͍ͬͯΔπʔϧͰಈ͘ 2.օͷݟ͑Δ৔ॴͰର࿩͢Δ 3.઀ଓ͢ΔαʔϏε͸؅ཧऀ͕ઃఆ 4.໰͍߹Θͤ΍ΞϥʔτͳͲΛى఺ʹಈ࡞
33. IssueʢGitHubʣ 1.IssueΛجʹ࣮૷ܭըΛ࡞੒ʢCodeRabbit Planʣ 2.ݱࡏͷίʔυϕʔεɺաڈͷIssueɺυΩϡϝϯ τͳͲΛࢀরͯ͠ܭըΛ࡞੒ 3.୭Ͱ΋ίϝϯτ͠ɺ࣮૷ܭըΛվળͰ͖Δ
34. CodeRabbit PlanͰ࣮૷ܭըΛ࡞੒
35. IssueʢGitHubʣ 1.୭Ͱ΋ݟΒΕΔɺίϝϯτͰ͖Δ 2.ޙͰࢀরɺৼΓฦΓ΋Ͱ͖Δ 3.ίϯςΩετ࿙Ε͕ൃੜ͠ͳ͍
36. ϨϏϡʔʢCodeRabbitʣ 1.CLI΍VS Codeػೳ֦ுͰϩʔΧϧϨϏϡʔ 2.PRϨϏϡʔ
37. 1. ϨϏϡΠʔɿϨϏϡʔΛड͚Δଆ AIίʔσΟϯάΤʔδΣϯτͷϨϏϡʔػೳ͸ͪ͜Β 2.ϨϏϡΞʔɿϨϏϡʔ͢Δଆ ɹ ϨϏϡʔ͸2ͭͷଆ໘͕͋Δ
38. 1. ϨϏϡΠʔɿϨϏϡʔΛड͚Δଆ AIίʔσΟϯάΤʔδΣϯτͷϨϏϡʔػೳ͸ͪ͜Β 2.ϨϏϡΞʔɿϨϏϡʔ͢Δଆ ↑ͪ͜Β͕଎͘ͳΔͱ։ൃੜ࢈ੑʹޮ͘🐰 ϨϏϡʔ͸2ͭͷଆ໘͕͋Δ
39. ࠶ܝ 1. AIར༻Ͱ24%ߴ଎Խ͢Δͱ༧ଌ͍͕ͯͨ͠ɺ19%௿Լͯ͠ ͍ͨ※1 2. AIੜ੒ίʔυ૿ͰɺϨϏϡʔ͕ϘτϧωοΫ※2 3. ίΞ։ൃऀͷϨϏϡʔෛՙ6.5%૿ʹΑΓɺνʔϜͷੜ࢈ ੑ͕௿Լ※3 4. ίʔυϨϏϡʔ୲౰ͷ໿9ׂ͕ɺAIੜ੒ίʔυͷීٴͰ ʮϨϏϡΞʔෛ୲૿ʯΛ࣮ײ※4 ※1 Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity - METR ※2 [2605.17548] Rethinking Code Review in the Age of AI: A Vision for Agentic Code Review ※3 [2510.10165] AI-Assisted Programming Decreases the Productivity of Experienced Developers by Increasing the Technical Debt and Maintenance Burden ※4 ʲAIੜ੒ίʔυͷϨϏϡΞʔෛ୲ʹؔ͢ΔௐࠪʳϨϏϡʔ୲౰ऀͷ໿9ׂ͕ʮෛ୲૿ʯΛ࣮ײ
40. 1. Α͋͘Δϛεɺ໰୊͸ࣄલʹղܾ 40+ͷLinter/SASTɺWebΫΤϦʔͰ࠷৽৘ใ 2.ਓ͕ݟΔΑ͏ʹϨϏϡʔ IssueɺυΩϡϝϯτɺաڈͷίϝϯτͰίϯςΩετΛ֦ॆ 3.ਓ͕PRΛཧղ͢Δͷʹඞཁͳ৘ใΛఏࣔ ֓ཁɺมߋϑΝΠϧάϧʔϓɺγʔέϯεਤ CodeRabbitͷ৔߹ ϨϏϡΞʔͷ޻਺Λ൒෼ʹ
41. ROI ࣌ؒ φϨοδϕʔε ඼࣭ ϚʔδલνΣοΫ
42. ·ͱΊ 🥕 AIར༻͸ݸਓٕ͔ΒνʔϜઃܭ΁ 🩴 ։ൃੜ࢈ੑ޲্ͷ伴͸ϨϏϡʔ ⛵ ΩϟϥΫλʔͷ໊લ͸Hoppy ͰνʔϜͷAI։ൃੜ࢈ੑΛߴΊΑ͏ʂ