Do most companies build or buy their AI solutions?

Across 2,208 classified AI deployments in the AI Use Case Hub, 70% are built custom (pro-code), 10% adopt an off-the-shelf assistant or copilot (buy), 13% compose a solution with low-code/no-code tooling, and 7% combine more than one approach.

Implementation

Build vs Buy vs Compose

Across 2,208 classified AI deployments, 70% are built custom (pro-code), 10% buy off-the-shelf, and 13% compose with low-code. Compare the build approach for every use-case type.

Use case types

165

Cases classified

2,208

68% of 3,231

Build

70%

custom / pro-code

Buy

10%

off-the-shelf

The whole corpus

How AI use cases get builtBuild, Buy & Compose — what they mean
Build — Custom, pro-code engineering — bespoke pipelines and SDKs (e.g. Bedrock, Vertex AI).
Buy — An off-the-shelf assistant or copilot, adopted as the main solution.
Compose — Low-code / no-code assembly (e.g. Copilot Studio, Power Platform, n8n).
Mixed — A combination of more than one of the above.

Of 2,208 deployments where the build approach is documented (68% of 3,231 cases), 70% build custom, 10% buy off-the-shelf, 13% compose with low-code, and 7% mix approaches.

70%

13%

Build— custom / pro-code engineeringBuy— off-the-shelf assistant / copilotCompose— low-code / no-code assemblyMixed— more than one approach

Min. cases

Sort

Customer service automation182cases·65% classified

61%

13%

17%

Build 61%Buy 13%Compose 17%Mixed 9%

AI platform150cases·77% classified

72%

14%

Build 72%Buy 14%Compose 9%Mixed 5%

Predictive maintenance145cases·63% classified

89%

Build 89%Buy 7%Compose 2%Mixed 2%

Claims automation114cases·63% classified

79%

17%

Build 79%Buy 4%Compose 17%Mixed 0%

Agriculture optimization101cases·50% classified

88%

Build 88%Buy 4%Compose 4%Mixed 4%

Risk assessment95cases·58% classified

71%

15%

Build 71%Buy 11%Compose 15%Mixed 4%

Workflow automation95cases·76% classified

Build 53%Buy 11%Compose 25%Mixed 11%

Patient engagement77cases·57% classified

75%

16%

Build 75%Buy 5%Compose 16%Mixed 5%

Automotive operations automation75cases·83% classified

69%

13%

Build 69%Buy 13%Compose 13%Mixed 5%

Clinical documentation75cases·77% classified

50%

34%

Build 50%Buy 34%Compose 9%Mixed 7%

Compliance automation74cases·72% classified

Build 49%Buy 13%Compose 26%Mixed 11%

Intelligent document processing74cases·84% classified

82%

15%

Build 82%Buy 0%Compose 15%Mixed 3%

Document automation72cases·81% classified

81%

Build 81%Buy 7%Compose 10%Mixed 2%

Fraud detection71cases·66% classified

94%

Build 94%Buy 2%Compose 4%Mixed 0%

AI development platform66cases·50% classified

88%

Build 88%Buy 6%Compose 6%Mixed 0%

Business process automation66cases·79% classified

29%

54%

12%

Build 29%Buy 6%Compose 54%Mixed 12%

Customer personalization53cases·79% classified

79%

14%

Build 79%Buy 14%Compose 0%Mixed 7%

Energy operations automation53cases·60% classified

Build 56%Buy 13%Compose 19%Mixed 13%

Customer support automation51cases·76% classified

46%

13%

33%

Build 46%Buy 13%Compose 33%Mixed 8%

Cloud migration44cases·39% classified

82%

12%

Build 82%Buy 0%Compose 12%Mixed 6%

Manufacturing quality monitoring42cases·62% classified

69%

19%

Build 69%Buy 19%Compose 4%Mixed 8%

Industrial inspection41cases·73% classified

93%

Build 93%Buy 0%Compose 7%Mixed 0%

Shopping recommendations36cases·61% classified

68%

14%

Build 68%Buy 14%Compose 14%Mixed 5%

Medical document automation35cases·77% classified

93%

Build 93%Buy 0%Compose 7%Mixed 0%

Supply chain optimization34cases·35% classified

50%

42%

Build 50%Buy 0%Compose 42%Mixed 8%

Customer experience analytics32cases·78% classified

92%

Build 92%Buy 0%Compose 0%Mixed 8%

Code assistant31cases·81% classified

56%

16%

20%

Build 56%Buy 16%Compose 8%Mixed 20%

Legal document automation31cases·77% classified

71%

13%

Build 71%Buy 13%Compose 4%Mixed 13%

Retail analytics platform31cases·68% classified

100%

Build 100%Buy 0%Compose 0%Mixed 0%

Student support31cases·65% classified

Build 40%Buy 20%Compose 15%Mixed 25%

Conversational support30cases·80% classified

75%

13%

Build 75%Buy 4%Compose 8%Mixed 13%

IT operations30cases·63% classified

79%

16%

Build 79%Buy 0%Compose 16%Mixed 5%

Healthcare workflow automation29cases·86% classified

64%

28%

Build 64%Buy 8%Compose 28%Mixed 0%

Workflow orchestration29cases·83% classified

75%

21%

Build 75%Buy 0%Compose 21%Mixed 4%

Medical imaging28cases·79% classified

95%

Build 95%Buy 5%Compose 0%Mixed 0%

Onboarding automation26cases·100% classified

15%

54%

31%

Build 15%Buy 54%Compose 0%Mixed 31%

Energy optimization25cases·44% classified

82%

Build 82%Buy 0%Compose 9%Mixed 9%

Legal document summarization25cases·84% classified

57%

33%

Build 57%Buy 33%Compose 0%Mixed 10%

Contact center modernization24cases·54% classified

54%

15%

23%

Build 54%Buy 8%Compose 15%Mixed 23%

Public sector automation22cases·45% classified

20%

70%

Build 20%Buy 10%Compose 70%Mixed 0%

ⓘ How this is measured

Each documented deployment is classified by how it was built: Build (custom / pro-code engineering — bespoke pipelines, SDKs such as Azure AI Foundry, Vertex AI or Amazon Bedrock), Buy (an off-the-shelf assistant or copilot adopted as the main solution), Compose (low-code / no-code tooling such as Copilot Studio, Power Platform or n8n), or Mixed when more than one applies.

Percentages are shown over classified cases. Deployments whose write-up doesn’t describe the build approach are left unclassified and reported as the “% classified” coverage figure, not forced into a bucket.

Back to Insights Hub