# Augmented Engineering

> AI-native software engineering, run by Jeevan Thandi from Shoreham-by-Sea in the UK.
> Augmented Engineering builds production software using coding agents alongside
> fifteen years of conventional engineering experience, which compresses delivery
> from months to days. It is a trading name of Question Data Ltd.

Engagements run from a one-day audit through to full production builds the client
owns outright. The distinguishing claim is speed with real infrastructure: first
working versions ship in days, on real payments, real data and a stack that scales
into production rather than a mock-up.

Contact: jeevan@augmented.engineering

## Services

- [Augment](https://www.augmented.engineering/#services): Fractional engineering direction inside an existing team, on a retainer at a flexible day rate. A couple of days a week covers what a full week of conventional engineering would.
- [Audit](https://www.augmented.engineering/#services): A second opinion on the stack, the process or whatever is slowing the team down. One day under NDA, returning an honest read plus a short plan. Day or hourly rate.
- [Automate](https://www.augmented.engineering/#services): Manual business processes rebuilt as custom agents that answer the phone and handle repetitive work in the background. Fixed fee per process.
- [Coach](https://www.augmented.engineering/#services): Hands-on 1:1 AI engineering coaching, so people can do the work themselves. Also at coworktocode.com.
- [Design](https://www.augmented.engineering/#services): Design, specification and code as one chain rather than three handoffs — the design tool is wired directly into the agents that write the software. Fixed fee, value priced.
- [Prototype](https://www.augmented.engineering/#services): A working interface on real data, for users, a board or an investor, in a stack that can scale into production. Fixed fee, 1 to 3 days.
- [Build](https://www.augmented.engineering/#services): Production software owned outright rather than rented monthly. Fixed scope, phased but fast.

## Case studies

- [Unstuck, for Matt Bagwell](https://www.augmented.engineering/work/unstuck/): A coaching and publishing platform built for author and coach Matt Bagwell. No written specification existed, so a recorded conversation became the specification. A first working version ran 3 days later against a conventional estimate of 3 months: real Stripe checkouts, real video rooms via the Daily API, live coaching sessions with automatic transcription, a Life Audit assessment tool, and a member app plus admin workspace. The client's words: "You have literally moved my business forward by at least 3 months, maybe more, in 3 days."
- [shoreham.tech](https://www.augmented.engineering/work/shoreham-tech/): A volunteer-run tech community for Shoreham-by-Sea, built and run as an in-house project. Free coding and AI training, a wiki written from first principles, and digital tools built for the town.
- [SlopGuard](https://www.augmented.engineering/work/slopguard/): A code quality guardrail for teams shipping with coding agents. Coding agents iterate by changing code and running tests, so where tests are missing the loop has no feedback and the agent is guessing. SlopGuard scores every method on complexity against missing coverage and produces one number to gate in CI. Live at slopguard.dev.

## Notes

- [Designing with agents: Paper and Claude Code](https://www.augmented.engineering/notes/paper-claude-code/): Paper is a design tool with an MCP server, so Claude Code can read the canvas, draw new artboards and export assets. One design-system reference artboard (mark, colour tokens, type scale, components, motion, voice) is read by agents before they design anything, which is how one reference produced the page designs, the social cards and the shipped article template itself. Paper's design-to-code skill then turns an approved artboard into production code in the project's conventions.
- [6 habits of the AI-native operator](https://www.augmented.engineering/notes/ai-native-operator/): What AI-native actually means — agents as the default executor, with human direction, context and judgement — and the 6 working habits behind it, each with named tools: record and transcribe everything (Voice Memos, Wispr Flow, Granola, Whisper), choose tools an agent can drive over MCP, CLI or API (Paper, AgentMail), learn a multi-pane terminal (iTerm2, tmux), package repetition into skills, write specs backed by quality gates (SlopGuard, Playwright), and treat software as disposable.

## Elsewhere

- [LinkedIn](https://www.linkedin.com/in/jeevanthandi/): Jeevan Thandi.
- [GitHub](https://github.com/JeevanThandi): Jeevan Thandi.
- [Cowork to Code](https://coworktocode.com): 1:1 AI engineering coaching.