How we use AI
AI-assisted software development, with senior engineers accountable
An honest account of how AI fits into our engineering: where it speeds us up, where it doesn't, and the controls that keep quality and your data safe.
What AI-assisted development means here
AI-assisted software development is the use of AI coding and analysis tools by engineers to draft, test, refactor, and document software faster, while humans remain responsible for design decisions, review, and quality. At Tech Insane, AI is a tool used by senior engineers, never a replacement for them.
Where AI helps, and where it doesn't
| Stage | How we use AI | What stays human |
|---|---|---|
| Discovery | Summarizing interviews and documents; drafting requirement lists | Understanding the business, deciding what matters |
| Architecture | Exploring options, drafting decision records | Every architectural decision and trade-off |
| Build | Scaffolding, boilerplate, refactors, migrations, first drafts of code | Design, review, and accountability for every change |
| Testing | Generating test cases and edge cases; test data | Test strategy; exploratory testing on real devices |
| Documentation | Drafting docs, changelogs, and handover notes | Accuracy review |
| Security | Assisted code scanning and dependency review | Threat modeling and final security sign-off |
Quality controls
- The engineer who submits a change is accountable for all of it, AI-assisted or not.
- Every change is reviewed by a second senior engineer.
- Automated tests, linting and static analysis run in CI on every pull request.
- AI-generated code gets extra scrutiny for licensing, security, and hidden complexity.
- AI features in your product are measured with evaluation suites before each release.
Data and security controls
- Only approved AI tools, used under business or enterprise terms that exclude training on your data.
- No secrets, credentials, or personal data pasted into AI tools.
- Your policy wins: if AI tools are not allowed on your codebase, we switch them off for your project.
- Tool list and data-handling terms available on request for your security review.
[VERIFY: list the specific AI tools you use and confirm each one's data-retention and training terms]
AI in the products we build
Building AI into your product is a separate discipline from using AI to write code. For product features, such as agents and automation or chatbots and voice bots, we treat the model as a component that can be wrong, and engineer around that: evaluation sets built from real cases, guardrails and schema validation, human approval for risky actions, logging for audit, and cost monitoring per task.
Questions about AI in our process
Does AI-assisted development mean lower quality code?
Not with the right controls. AI-assisted code goes through the same review, tests, and security scanning as any other code. The engineer who opens the pull request is accountable for every line, whoever or whatever typed it.
Will our code or data be used to train AI models?
No. We only use AI tools under business or enterprise terms that exclude training on customer data, and only with your agreement. If your policy prohibits AI tools, we disable them on your project.
How much faster is AI-assisted development?
It depends on the work. Gains are typically largest on routine work such as boilerplate, tests, migrations, and documentation, and smallest on novel architecture and complex debugging. We report delivery metrics monthly, so you see the effect on your own project rather than a marketing number. [VERIFY: add your measured range once you have data.]
Do you pass the savings on?
We invest them in quality and scope: more tests, better documentation, and more features delivered within the same retainer.
Want to see AI-assisted delivery on your project?
A 30-minute strategy call with our founder. We will ask sharp questions, tell you honestly whether we are the right team, and outline how we would approach it. NDA available before the call.