Companies that sell what teams need to train, serve, evaluate and monitor models. It has no schema of its own: companies here are mapped on the Artificial intelligence schema, and what it does not cover is recorded as gaps.
© 2026 System Atlas · Built by MIOSA
Free to map your own company · Members $300 a yearHow automatable each primitive is, estimated from the 3 organizations mapped in public on the Artificial intelligence schema, which this industry inherits: the share of those with the primitive that run it on software. Estimate from public data: the share of mapped organizations that run this on software. Not a recommendation.
| Primitive | Layer | Automation | Peers on software | Software parts named most |
|---|---|---|---|---|
Pipeline generation Demand: winning customers | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Quotes and order forms Demand: winning customers | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Model and product design Supply: making or delivering the offer | Process | MediumLayer only | 3 of 3 mapped; too few to read | Company Identity (1), Computers (1), MaluDb Application Framework (1) |
Sprint and capacity planning Supply: making or delivering the offer | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Compute and data purchasing Supply: making or delivering the offer | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Model access and service Supply: making or delivering the offer | Process | MediumLayer only | 2 of 3 mapped; too few to read | Docker Host deployments (1), Kinetic Cloud GPU environments (1), PostgreSQL 17 and Oracle 19c database services (1) |
Output quality review Supply: making or delivering the offer | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Code, infrastructure and licences Supply: making or delivering the offer | Data | MediumLayer only | 0 of 3 mapped; too few to read | — |
Provisioning Supply: making or delivering the offer | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Uptime and incidents Supply: making or delivering the offer | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Compute capacity Supply: making or delivering the offer | Data | MediumLayer only | 0 of 3 mapped; too few to read | — |
Training data acquisition and licensing Supply: making or delivering the offer | Data | MediumLayer only | 0 of 3 mapped; too few to read | — |
Pretraining and fine-tuning runs Supply: making or delivering the offer | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Evaluation and red-teaming Supply: making or delivering the offer | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Developer and enterprise onboarding Customers: keeping them | Process | MediumLayer only | 1 of 3 mapped; too few to read | — |
Customer data Customers: keeping them | Data | MediumLayer only | 1 of 3 mapped; too few to read | — |
Renewals Customers: keeping them | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Churn and expansion Customers: keeping them | Data | MediumLayer only | 0 of 3 mapped; too few to read | — |
Invoicing Money: getting paid and paying | Process | MediumLayer only | 1 of 3 mapped; too few to read | — |
Collections and receivables Money: getting paid and paying | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Payments and payables Money: getting paid and paying | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Payroll Money: getting paid and paying | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Cash and treasury Money: getting paid and paying | Data | MediumLayer only | 1 of 3 mapped; too few to read | — |
Unit economics Money: getting paid and paying | Data | MediumLayer only | 0 of 3 mapped; too few to read | — |
Hiring People: who does the work | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Onboarding and training People: who does the work | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Scheduling and capacity People: who does the work | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Metrics and board reporting Systems: tools and data | Data | MediumLayer only | 2 of 3 mapped; too few to read | Sentry error tracking (1) |
Data management Systems: tools and data | Data | MediumLayer only | 2 of 3 mapped; too few to read | — |
Legal and contracts administration Control: rules the company runs under | Data | MediumLayer only | 2 of 3 mapped; too few to read | — |
Management cadence Direction: deciding where it goes | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
Learning and improvement Direction: deciding where it goes | Process | MediumLayer only | 0 of 3 mapped; too few to read | — |
The rule, the same for every industry: with 5 or more organizations having the primitive, 50% or more of them on software is high, 20% or more medium, below that low. With fewer, the level rests on the primitive's layer alone and says "Layer only". Software parts are counted once per organization; a detected technology is named by its product. See the blueprint for which organizations run each.
Counts of what public research has found and sourced, each part with its source on the organization’s profile. They say what is known, not how much of the industry exists.
Focused research on the organizations with the fewest mapped parts in machine learning and ai developer tooling; what it finds is public for everyone.