01
Full application systems
Flutter and Firebase applications taken from spec to a live store listing: data model, authentication, payments, release builds, and submission to the App Store and Google Play.
Applications, AI & automation
Software engineering that puts quality and security first.
Zetesis is a founder-led software engineering agency building applications, custom AI systems, and business automation. We work primarily through private referrals and take on new engagements where there is a strong technical and working fit.
Our approach to security & qualityBuilt to work together.
The screens and interactions people use, connected to the logic and data behind them.
- Engagement
- Fixed scope, quoted per project, rather than a retainer.
- Intake
- Project work primarily through private referrals. New inquiries reviewed for fit.
- Work
- Remote, async by default.
- Confidentiality
- NDA signed before project scoping. Client work is never named on this site.
- New
- Sustain early access and Headcount setup calls need no referral or NDA.
001 Services
Three service lines, each with a defined deliverable.
These are starting points. Every project is scoped and quoted once the requirements are clear.
An application, from scope to release.
- What we discuss
- Who will use it, the core tasks it needs to support, and the systems it connects to.
- Possible deliverables
- A mobile or web application, authentication, a data model, payments, and release preparation.
- Bring to the first meeting
- Your main user journey, any existing designs or software, and the constraints around launch.
An AI capability with clear boundaries.
- What we discuss
- The task, the data available, who may access it, and what should happen when the model gets something wrong.
- Possible deliverables
- Document retrieval, structured extraction, or an agent workflow, with evaluation and fallback behavior.
- Bring to the first meeting
- A description of your data, examples of useful outputs, and your privacy and access requirements. Confidential material can wait until an NDA is in place.
Related engineering: Why a system prompt is not a security boundary
How an AI project worksConnect the steps in your workflow.
- What we discuss
- The current process, the tools involved, and where manual work or repeated errors occur.
- Possible deliverables
- An integration between systems, scheduled processing, an intake tool, or an internal dashboard.
- Bring to the first meeting
- A walkthrough of the process, its inputs and outputs, and the exceptions someone currently handles by hand.
Start with the problem you need solved.
- What we discuss
- What happens today, what needs to change, and the constraints a solution needs to respect.
- Where it leads
- A conversation about the technical options and which service fits. The quote follows once the scope is clear.
- Bring to the first meeting
- A plain-language description, examples of the current process, and any deadline or budget constraints you already know.
02
Custom AI systems
Retrieval, extraction, and agent workflows built to hold up on real data rather than a demo set, with evaluation and a fallback path for the cases the model gets wrong.
03
Business automation tooling
Internal tools that remove a manual step: intake forms, scheduled jobs, integrations between systems that do not talk to each other, and dashboards over the data you already hold.
// estimates only, final scope set per project
002 New services
Two new services, open to anyone.
Project work comes through private referrals, with an NDA signed before scoping. Sustain and Headcount need neither, so anyone can join Sustain's early access or book a Headcount setup call.
Zetesis Sustain
Early accessSustain
Put your app's maintenance on autopilot.
An AI maintenance agent for custom apps. Sustain watches your app and the services it relies on. It prepares and tests routine fixes, and tracks app store deadlines weeks ahead. You decide how much it can ship.
Autonomy, within your limits
- Monitor Check the app
- Prepare Stage a fix
- You decide Review and approve
- Flutter and Firebase at launch
- Read-only on the Watch plan
- Every fix can be undone
Zetesis Headcount
Available nowHeadcount
Your entire company in one place.
AI agents for the jobs a one-person company would hire for. We set them up in one workspace, with a channel for each job, and can keep them running for you. Anything that reaches a customer or spends money waits for your OK.
Five hires, one person
An org chart fills in as five agents join under you, one for each job. The team grows to six. The number of people stays at one.
YouOwner
- Inbox agentFront desk
- Bookings agentScheduler
- Invoices agentBilling
- Newsletter agentWriter
- Numbers agentAnalyst
- $2,500 setup, paid once
- Monthly plans from $250
- Up to 5 agents at setup
Engineering, in practice
What a prompt cannot protect.
Sandbox testing caught a public-facing service with access to private documents, days before deployment. The fix changed what its credentials could read.
Compare the access
Inherited credentials
The read was still permitted.
The service could access the private collection even though its prompt described a public-facing role.
Separately scoped credentials
The data store denies the read.
The public service has no read permission on the private collection. A change in the model's behavior cannot grant that access.
Simplified illustration. The linked reconstruction uses synthetic documents.
003 How a project runs
From the first conversation to a system you own.
Follow the work from scope to handover. Each stage has something concrete for you to review.
01 / 04
Agreed scope
Scroll to follow. Select a stage to jump.
01 / Scope
Make the decisions
before the build.
We first establish whether the project and working relationship are a fit. With an NDA in place, we discuss requirements, constraints, and technical options. The quote follows once the scope is clear.
We agree on milestones and walk through account setup, access, and what will transfer at handover.
You reviewA written scope, a fixed quote, and a milestone plan.
02 / Build
Connect the parts.
Review the work.
The interface, application logic, data, and integrations come together around the agreed scope. You review and approve completed milestones before their payments are due.
Project conversations are included. We arrange follow-up calls when a question or decision needs one.
You reviewThe working deliverable for each agreed milestone.
03 / Test
Check what happens
when things go wrong.
Testing is matched to the project: the core flows, access rules, and how integrations behave when something fails. The agreed work is reviewed before release.
Access and permissions
Who may read the data or perform an action? The check needs to reach the permission boundary, including requests that should be refused.
See a permissions issue caught before launchFailure handling
What happens if an external service is unavailable, a request times out, or someone repeats an action? The expected response belongs in the design and the checks for that integration.
Acceptance checks
Does the deliverable do what was agreed? Review the core user journeys, important edge cases, and any limitations against the project requirements.
You reviewThe agreed deliverable against its requirements.
How we approach security & quality04 / Handover
Your code.
Your system to run.
After final payment, ownership of the custom project code and source repository transfers to you. Infrastructure is set up in your name where practical, with remaining transfers planned from the start.
We walk through setup and handover so you know what you own and how to take the project forward.
You receiveThe source repository, agreed account transfers, and a handover walkthrough.
Payments, ownership & support004 Stack
The toolchain the work actually runs on.
Depth in a focused set of tools, backed by comfort across the mainstream stack when a project calls for it.
app
Flutter
iOS, Android, and web from one Dart codebase
React
web frontends where a native or Flutter app is not the right fit
data
Firebase
Auth, Firestore, Cloud Functions, Storage
payments
Stripe
checkout, subscriptions, and webhook handling
cloud
Cloudflare
DNS, edge Workers, bot protection
AWS
infra outside the Cloudflare and Firebase footprint
systems
Rust / C
performance-critical components, chosen per project
005 About
Direct from scope to delivery.
Zetesis builds and ships production systems: mobile and web applications, AI features that have to survive contact with real data, and internal tooling that takes manual steps out of a business.
Zetesis began as independent software engineering work and grew through private referrals. It is developing into an agency around those existing client relationships.
Zetesis is founder-led. You work directly with the person responsible for scoping and building your system. Scope is agreed before work starts, and each engagement runs to a defined deliverable.
Quality and security matter in every system we build. We check that it does the agreed job, control access to data and actions, and review how it handles failures.
New inquiries are considered for project fit and an ongoing working relationship. Every engagement stays confidential. Our writeups explain engineering decisions without identifying clients or showing their products.
006 Contact
Introduce your project.
Tell us what you want to build and your timing. A high-level introduction is enough; keep confidential project material out of this message.
We review your project, assess whether we can work well together, and arrange a conversation where there is a fit. An NDA comes before confidential scoping.