Buy a tool for this process, or build around our own knowledge?
Embedded vendor AI requires your data to live in their cloud, but agentic AI now enables building proprietary domain logic in-house at a fraction of historical costs, shifting the make option to a hybrid governance form.
The question
A vendor sells a tool that covers most of this process out of the box, but our differentiation comes from internal expertise the vendor's product does not encode. Do we buy the tool and accept the generic workflow, build around our own knowledge, or buy the commodity layer and build only the differentiated part?
Counsel's position
Buy the commodity layer and build only the differentiated part to protect expertise, maintain control, and optimize long-term cost and maintenance.
Verdict
The verdict: Buy the commodity layer and build only the differentiated part to protect expertise, maintain control, and optimize long-term cost and maintenance.
How the criteria decide
4 of 5 criteria resolved on cited evidence. 1 had none either way.
| Criterion | Favours | Evidence |
|---|---|---|
| differentiation | Buy commodity, build differentiated | Domain logic and proprietary data survive model generations while orchestration does not If it’s plumbing, buy it. Orchestration, hosting, standard retrieval. Your customers do not care whose CI/CD ran the deploy. |
| cost | Buy commodity, build differentiated | Domain logic and proprietary data survive model generations while orchestration does not If it’s plumbing, buy it. Orchestration, hosting, standard retrieval. Your customers do not care whose CI/CD ran the deploy. |
| time-to-market | No evidence either way | |
| control | Buy commodity, build differentiated | Embedded vendor AI requires your data to live in their cloud environment Every embedded AI capability ships with an unstated architectural prerequisite: your data must be where the AI can see it, in the shape it expects, under the governance the vendor enforces. Distributed AI workflows require continuous evaluation of trust metadata across administrative boundaries Every transition requires participants to determine whether the operational authority supporting the next stage of execution remains valid under current conditions. |
| maintenance burden | Buy commodity, build differentiated | Domain logic and proprietary data survive model generations while orchestration does not If it’s plumbing, buy it. Orchestration, hosting, standard retrieval. Your customers do not care whose CI/CD ran the deploy. Artificial Intelligence on Medium Routing decisions and workflow governance consume significant engineering effort in mature deployments Larger deployments often reveal that routing decisions, resource utilization, and workflow governance consume a significant portion of engineering effort. |
Embedded vendor AI requires your data to live in their cloud environment
You can compose a custom architecture for cross-application workflows to maintain control over your proprietary data and orchestration.
Domain logic and proprietary data survive model generations while orchestration does not
You can buy managed services for commodity components like orchestration while focusing your engineering effort on proprietary data and custom domain logic.
Agentic AI shifts the make option to a hybrid governance form
You can leverage agentic coding systems to build software in-house at a fraction of historical costs.
Routing decisions and workflow governance consume significant engineering effort in mature deployments
You can adopt control plane patterns to manage the operational variability of AI workflows when building your custom components.
Distributed AI workflows require continuous evaluation of trust metadata across administrative boundaries
You must implement shared coordination capabilities to preserve operational confidence as execution moves among participants in your custom architecture.
Read another verdict
- Which process should we point AI at first?
- Put one person in charge of AI — or is a Head of AI premature for us?
- Centralize AI strategy under CEO or distribute ownership?
- Adopt new AI ROI tools or refine existing methods?
- Invest in pre-build costing or post-deployment ROI tracking?
- Our documents are a mess. Clean them up before AI, or after?
- How do we measure the return on an AI workflow — and what baseline is honest?
- Our best people's know-how isn't written down — can AI even use it?