AI by industry — Technology & SaaS
AI in technology and SaaS: margin, product and support
Software companies face the question twice: how AI changes their own cost base, and how it changes their product. The two decisions have different economics and should not be merged.
Where AI changes the economics
Engineering productivity
Code assistance, test generation and review support, measured on cycle time rather than acceptance rate.
Support deflection
Grounded answers over documentation with clean handover to a human.
In-product AI features
Capability that customers will pay for, priced against inference cost rather than bundled by default.
Go-to-market operations
Research, drafting and CRM hygiene for revenue teams.
What usually blocks deployment
- Inference cost eroding gross margin when features are bundled free
- Customer data boundaries and sub-processor disclosure
- Evaluation discipline — quality regressions ship silently without it
First moves
- Model inference cost per active user before pricing an AI feature
- Instrument an evaluation harness before the first release
- Separate the internal-efficiency case from the product case
Questions leaders ask
- Should we bundle AI features or charge for them?
- Model the inference cost per active user first. Bundling a variable cost into a flat subscription is the fastest way to erode gross margin at scale.
- How do we stop AI features degrading quietly?
- An evaluation harness with quality thresholds, run on every release. Without it, regressions reach customers before they reach a dashboard.
Score your readiness in technology & saas
Ten dimensions, about eight minutes, and a prioritized action list you can take into a board conversation.