If you want to avoid model markup and lock-in, bring-your-own-key (BYOK) platforms are the answer — pay providers directly, switch freely. Here’s a 2026 comparison. (dgm implements osFoundry as an independent partner; we’ll be honest about the field.)
Why BYOK?
Bring-your-own-key lets you connect your own model-provider accounts, so you:
- cut cost — pay providers directly, no markup;
- avoid lock-in — switch or mix models freely;
- improve data control — choose providers and terms; and
- use the best (or cheapest) model per task.
(See our bring-your-own-model and avoid vendor lock-in guides.)
The options
| Platform | Type |
|---|---|
| osFoundry | Full BYOK platform (chat/agents/apps/knowledge) |
| Dust | BYOK agent workspace |
| Relevance AI | BYOK no-code agents |
| n8n | BYOK self-hosted automation |
| Make | BYOK visual automation |
| CrewAI | BYOK multi-agent framework |
Matching to scope
- Broad platform → osFoundry (BYOK by default, multi-model, consolidation).
- Agent building → Dust or Relevance AI.
- Automation → n8n or Make.
- Framework → CrewAI.
All let you bring your own models; they differ in scope and build-vs-configure.
BYOK + data control
A strong combination for UK data sovereignty: BYOK (you choose providers and terms — e.g. those that don’t train on your data) plus self-hosting or an EU region for the platform. Together they give meaningful control over where data goes. (See data residency.)
Where osFoundry fits
osFoundry makes BYOK the default — paste a provider key and that model is available to every agent, app and pipeline, and you can route the same prompt across providers. For sensitive data, it adds self-hosting and local inference. If you want a broad BYOK platform, it fits; for narrower needs, the agent or automation options suit.
dgm is an independent integration partner with zero integrations so far, implementing osFoundry. To set up a BYOK AI stack, book a consultation with dgm. General information; verify current vendor options.