Software engineering consulting
ComplxAI provides software engineering consulting for Australian businesses and founders: architecture reviews, codebase and security audits, pre-raise technical reviews, delivery rescue and AI-readiness assessments, each ending in a costed plan we can deliver.
- Findings and code are yours from the first paid stage
- The reviewers are the engineers who would build it
- Every stage priced in writing, exit at any boundary
For anyone deciding about software they did not write
A founder about to raise. A CTO whose platform slows with every customer. A business that inherited a system from a vendor. Each needs an independent view of what is there and what to do first.
Know what you want built? See custom software development. Choosing AI use cases? Start with AI consulting. Reviews follow the stages in how we work and can stop at the report or continue into delivery.
A good fit if…
- About to raise, with technical risk to put in writing before investors ask
- A project is late and nobody can say when it ships
- A codebase inherited from a vendor or a departed engineer
- AI features wanted, platform readiness unknown
What our software engineering consulting covers
Architecture reviews
Structure, boundaries, data flow and scaling limits, ending in a target design you can build towards.
Pre-raise technical reviews
For founders about to raise: code ownership, key-person risk, security and scalability, written so investor questions are answered before they are asked. We have supported clients through VC raises.
Codebase and security audits
Tests, dependencies, secrets and access, sorted into what is dangerous, what is expensive and what to leave alone.
Delivery rescue
Assess the stalled build, reset scope and cadence, then coach the team or take over delivery in 12-week blocks.
Engineering practices
CI/CD, testing strategy, observability and release management, so the team ships on a schedule rather than around incidents.
Modernisation roadmaps
What to replace first, what to keep, how to migrate data and how to keep the old system running meanwhile.
AI-readiness assessment
Whether an existing product can carry AI features: data access, permissions, latency, cost. Leads into AI feature development or AI engineering.
What a review examines
Six layers in your repositories and cloud account, each reported with what to fix first. The infrastructure and security layers draw on our cloud consulting work.
- Codestructure and qualityModule boundariesDependency graphTest valueError handling
- Datastorage and movementSchema and migrationsIndexes and query plansBackups and restoresPII inventory
- Infrastructurewhere it runsAWS account layoutTerraform/CDK coverageScaling modelCost by service
- Securitywho can do whatAuthenticationAuthorisation modelSecrets managementAudit logging
- Operationskeeping it runningObservabilityIncident historyRunbooksRecovery objectives
- Deliveryhow change happensCI/CD pipelineCode reviewRelease cadenceKey-person risk
What you receive
Written for whoever has to read them: a board, an investor or your own team. Exposed credentials are raised with you directly, not written into the report.
- Written findings
- Plain-language summary, then a technical appendix with file references and evidence from your accounts.
- Prioritised risk register
- Each risk rated by likelihood, impact and cost to fix, in recommended order.
- Target architecture
- Where the system should be, with the path from the current state.
- Delivery plan with costs
- Work sequenced into priced 12-week blocks you can fund, take in-house or put to another vendor.
How an engagement runs
Two free conversations, then stages priced in writing.
- ScopeA free intro call and pre-discovery session to agree what is reviewed and which decision it supports.
- ReviewTwo to four weeks from a fixed fee: read-only access, interviews, evidence from your accounts.
- Findings and planFindings, risk register, target architecture and costed plan, walked through with the people who act on it.
- Optional deliveryThe same engineers carry out the plan in 12-week blocks, with working software every couple of weeks.
Where this work shows up
Pre-raise technical review
A founder wants investor questions answered before they are asked. We review the platform, write findings in term-sheet language and fix what would become conditions.
Scaling pains in a SaaS
Slow pages and failed jobs in a multi-tenant product. The review finds the database contention and the missing queue; fixes land before the next customer cohort. See SaaS development.
Replacing a legacy system
Software its original developer no longer supports. A modernisation roadmap decides what to keep, what to rebuild, and in what order.
Taking over from a previous vendor
A repository, some credentials, a short handover email. We audit what was delivered, secure the accounts and stabilise the system before new work starts.
Preparing a platform for AI features
Data locked in a monolith, permissions enforced in the UI. The assessment fixes the foundations, then hands into adding AI features to an existing product.
How we handle your code
This work often runs under NDA, and the working rules below are agreed before access is granted. In rescue and delivery, changes go through CI and a person approves anything with consequences.
The working rules
- We regularly work under NDA and will sign yours before access is granted
- Read-only, least-privilege credentials you issue and revoke when the engagement ends
- Code stays in your repositories, never in tools you have not approved
Advice from the people who would build it
We build what we recommend
The engineers writing the findings would implement them, so nobody proposes a rewrite they would not want to do.
Independent and small
A small senior Australian team that does the work itself, with no platform or licence to resell.
Priced per stage, yours from day one
No mandatory retainers. The report, the plan and any code we write belong to you the moment a paid stage starts.
Systems we have shipped
A custom SaaS platform for Modedu, web and mobile apps for Luxpip, invoice processing for DisabilityAssessments. See our case studies.
Where a review usually leads
Custom software
Systems built to the target architecture.
Custom software developmentCloud engineering
Infrastructure-as-code, cost and reliability.
Cloud consultingAI consulting
Use cases scored by impact, effort and risk.
AI consulting in AustraliaAI engineering
Retrieval, evaluation and observability in production.
Production AI engineeringSaaS development
Tenancy, billing and admin surfaces built to grow.
Custom SaaS platformsAI feature development
Features for a product that has passed its assessment.
Adding AI features to an existing productQuestions people ask before they book
The rest is for the free intro call.
Book a free intro callHow long does a review take?
Two to four weeks, from a fixed fee put in writing before the review begins. A tighter timetable narrows the scope, not the depth.
What do we receive?
A plain-language summary with a technical appendix, a prioritised risk register, a target architecture and a costed delivery plan, all yours to share.
Do you then build what you recommend?
Yes, if you want us to. The same engineers deliver it in 12-week blocks, priced separately and in writing, with an exit at every boundary. Or take the plan in-house or to another provider.
Do you sign NDAs?
Yes. We regularly work under NDA, which is why several engagements on our case studies page appear only in outline. Send yours before pre-discovery.
Bring the repository, the deck or the deadline
Two free conversations to scope the review, then a fixed price in writing. If the system should be left alone, we will say so.