01 · Solutions

AI-driven development

Two things live here. What we build for you: assistants that answer from your own documents with citations, extraction from paperwork with a human queue for anything uncertain, and workflow agents that log every action — each shipped with an eval suite you keep. And how we build: AI drafts the scaffolding, tests, and documentation so our engineers spend their hours on your business logic, which is why the same scope costs 30-40% fewer hours. If policy rules out generated code, the hand-crafted track uses no AI tooling at all and comes with a written attestation.

RAGAgentsEvalsGuardrails
  • Working prototype in the first week on most projects
  • AI features with citations, permissions, and an eval suite you keep
  • A fully manual delivery track when you need one
Get a written scope
  • Fixed written scope and price
  • NDA before project details
  • You own the code and IP
  • Reply within one business day

How we build with AI

Six stages, each with a human gate

01

Scope with AI, decide with people

We use AI to explore the problem space fast — competitor teardowns, data-model options, edge cases nobody listed. The architecture decision is made by an engineer and written down.

Gate: Architect signs the design doc

02

Generate the skeleton

Project setup, data models, CRUD, API contracts, and typed clients are generated against our own templates rather than typed out for the hundredth time.

Gate: Lead engineer reviews the scaffold

03

Build the parts that matter by hand

Business logic, pricing rules, permissions, anything with money or safety attached — written and reasoned about by a senior engineer.

Gate: Peer review, no self-merges

04

Draft tests with AI, harden them with QA

AI proposes the happy paths and obvious edges; QA adds the cases only someone who has watched real users would think of.

Gate: Coverage delta must not fall

05

Review every diff, generated or not

One checklist for all code: tests, dependency diff, licence scan, threat model impact, and a named human approver on the merge.

Gate: Named reviewer recorded per merge

06

Ship, watch, and iterate

CI/CD, observability, and a weekly demo. Post-launch, AI helps triage logs and draft fixes — the fix still ships through the same gate.

Gate: Release sign-off by delivery lead

Guardrails

What we commit to in writing

Your code and data stay yours

We use business-tier tooling with training disabled. Client code and data are never contributed to a vendor training set, and secrets and personal data are redacted before any prompt.

A human is accountable for every line

Generated code carries no special status. It passes the same tests, review, and security checks, and a named engineer approves the merge.

Licence and provenance checks

Dependency licences are scanned on every build and we keep a generation log per merge, so procurement and legal have an answer when they ask.

You can opt out entirely

Choose the hand-crafted track and no generative tooling touches your project. You get a written attestation of a fully human-authored codebase.

AI in your product

Capabilities we build for clients

Retrieval assistants (RAG)

Answers grounded in your own documents, with citations and permissions that respect who is asking.

pgvectorAzure OpenAICitations

Document extraction

Invoices, contracts, and scanned paperwork turned into structured data, with a confidence threshold that routes uncertain cases to a human queue.

OCRStructured outputHuman-in-the-loop

Workflow agents

Multi-step automations that call your systems, log every action, and stop at the approval step you define.

Tool useAudit logApprovals

Classification & triage

Routing for support tickets, leads, and claims — measured against a labelled set rather than a hunch.

EvalsConfidence scores

Search that understands intent

Semantic search over catalogues and knowledge bases, blended with the keyword filters your users already trust.

Hybrid searchRanking

Evaluation & monitoring

An eval suite built before the feature, plus dashboards for quality, latency, and cost per request after launch.

Eval setsObservabilityCost control

With AI or without

The same team, two delivery tracks

Who writes the code

With AI
Engineers direct AI, then review and refactor every diff
Without AI
MZA engineers, every line, no generation

Typical timeline

With AI
2-10 weeks for most scopes
Without AI
4-16 weeks for the same scope

Indicative cost

With AI
Baseline
Without AI
Roughly 1.4x the AI track

Hourly rate

With AI
$10 / hour blended
Without AI
$10 / hour blended

Human review

With AI
Mandatory named reviewer on every merge
Without AI
Standard peer review on every merge

Test coverage target

With AI
Same target, tests drafted by AI and hardened by QA
Without AI
Same target, tests written by QA

IP provenance

With AI
Generation log per merge, licence scan on dependencies
Without AI
Clean-room attestation, contributor-level records

Your data in AI tools

With AI
Never used for vendor training; redaction before any prompt
Without AI
Not applicable — no AI tooling used

Best when

With AI
Speed and budget matter and you want AI features in the product
Without AI
Policy, procurement, or licensing rules out generated code

When we say no

AI is not always the answer

We would rather lose the line item than ship a model where plain code belongs. These are the cases where we will tell you so on the first call.

  • The problem is a rules engine wearing an AI costume — deterministic code is cheaper and testable.
  • There is no labelled data or evaluation set, and no appetite to build one.
  • A wrong answer has no human in front of it and no way to appeal.
  • The value is a demo for a board meeting rather than a workflow someone will use on Monday.

FAQ

Straight answers before you sign

The questions every buyer asks us — on cost, ownership, and how the work actually runs.

No. We use business-tier tooling with training disabled, and secrets and personal data are redacted before any prompt. Client code and data are never contributed to a vendor training set.

Tell us what you want to build

Book a 20-minute call or send a short brief. You get a written scope and a fixed price — we reply within one business day.

  • Fixed written scope and price
  • NDA before project details
  • You own the code and IP
  • Reply within one business day