Skip to content

Offer 02 · Custom AI software and SaaS

Custom AI software that works with real users, built one phase at a time

Applications, SaaS platforms and MVPs with AI where it earns its place. Every phase has a fixed price and an approval step before the next one starts.

Scope my productSee the types we build

Types we build

Six kinds of product, each built around one job

  • 01

    AI feature inside your app

    An assistant or search in your product, with limits on what it can do.

    1. User asks
    2. Feature answers within limits
    3. Edge case goes to a person

    Example workflows

    • A help assistant that answers from your help centre and sends refunds for approval.
    • A recommendation panel that explains each suggestion and logs what it showed.
    Build this agent
  • 02

    SaaS platform

    A multi-customer product with accounts, billing and AI built in.

    1. Customer signs up
    2. Product runs the workflow
    3. Owner reviews usage

    Example workflows

    • A scheduling platform for clinics with an AI intake step that staff review.
    • A reporting tool where AI drafts summaries that users edit before sharing.
    Build this agent
  • 03

    MVP

    The smallest version that proves one job with real users.

    1. Scope the one job
    2. Prototype with users
    3. Decide to build or stop

    Example workflows

    • A quoting tool tested with a small group of pilot customers before the full build.
    • A matching product tried in one market before it is built out.
    Build this agent
  • 04

    Internal tool

    An admin console or ops tool your team uses every day.

    1. Staff request
    2. Tool drafts the action
    3. Staff approves in the console

    Example workflows

    • An ops console that queues AI-drafted replies for a person to approve.
    • A finance review screen that flags mismatches and logs each decision.
    Build this agent
  • 05

    Customer portal

    Customers see their own data and actions, and every change is written to the audit trail.

    1. Customer signs in
    2. Portal runs the request
    3. Change recorded in the audit trail

    Example workflows

    • A portal where customers check orders and request changes, with each change logged.
    • An account page where billing edits are approved by staff and logged on the record.
    Build this agent
  • 06

    Data sync

    Keeps your product in step with the systems it depends on, with checks on every sync.

    1. Source changes
    2. Sync checked
    3. Failed sync held for review

    Example workflows

    • A nightly sync between a CRM and billing that holds mismatched records for review.
    • An inventory feed that logs every update and flags gaps before they reach customers.
    Build this agent

Where products stall

Four points where custom AI products go wrong

  • 01

    The AI feature works in the demo and breaks with real users

    Agent does
    Each AI feature ships with guardrails: limits on what it can do, a fallback when unsure, and a log of every answer.
    Person approves
    Any action that changes money, data or access.
    Scope this situation for my business

    Demo data is clean and short. Real users ask strange questions, and nothing stops a wrong answer reaching a customer.

    • Demo answered every question cleanly
    • Live users asked things the demo never covered
    • A wrong answer reached a customer, with no log
    BeforeAfter

  • 02

    The prototype works, but nobody can maintain it

    Agent does
    The build includes an admin console, request logs and documentation, so your team can see what the product is doing.
    Person approves
    Changes to the production system and releases.
    Scope this situation for my business

    A prototype built in a rush has no monitoring, no admin view and no record of why decisions were made.

    • Prototype built in a rush, no monitoring
    • Only the original developer knows how it works
    • Every change takes a week because nothing is documented
    BeforeAfter

  • 03

    Build or buy is a guess

    Agent does
    Each phase starts by testing whether a custom part is needed, so you only build what an off-the-shelf option cannot do.
    Person approves
    The decision to build a custom part, phase by phase.
    Scope this situation for my business

    Off-the-shelf tools fit most of the job, then fail on the one part that makes your product different.

    • Off-the-shelf tool fails on the one feature that matters
    • Custom part was never scoped separately
    • Rebuild decision made after the budget is spent
    BeforeAfter

  • 04

    Nobody knows what it will cost or how long it will take

    Agent does
    Work is split into phases. Each phase has a fixed price and a written scope before it starts.
    Person approves
    Each phase price and scope, before the phase starts.
    Scope this situation for my business

    Open-ended quotes hide the risk, and the budget runs out before the product is useful.

    • Quote arrived as one open-ended number
    • Scope changed twice with no written record
    • Budget gone before the product was useful
    BeforeAfter

Questions buyers ask

Before you scope a project

What is custom AI software development?
It is building software where an AI model is part of the product itself: an assistant inside your app, a recommendation engine, an automated review step, or a full SaaS platform. You own the code, the models and the data.
How is an MVP different from a full product?
An MVP tests one core job with real users. A full product adds the admin tools, integrations, permissions and monitoring that a business needs to run it every day. We start with the smallest version that proves the job, then add phases.
Should we build the AI part or buy it?
Buy the parts that are the same for every business, like email or payments. Build the part that makes your product different. We help you split the two before you spend on either.
How is the price set?
Each phase has a fixed price and a written scope that you approve before the phase starts. We do not quote an open-ended project. The first phase covers the written scope before any build starts.
Who owns the code and the data?
You do. The source code, infrastructure, models and data are handed over to you, and the handover is part of the written scope.
What happens after launch?
Support after launch is a separate phase, scoped and approved the same way as the build.
What does the audit trail record?
Every action the system takes: the input, the source or factors used, the output, and who approved it if a person was involved. The trail is kept in storage you control, so your team can answer an audit question from the record.

Case studies

Writing

Related offers

Other ways we build AI

Start with one phase

Tell us the product you want to build

We scope the first phase with a fixed price for each step, before any build starts.

Scope my product
  1. 01

    Share the product

    Tell us the job, the users, and what goes wrong today.

  2. 02

    We scope the phase

    A fixed-price plan for the first phase, with the approval step named.

  3. 03

    We build in phases

    You review each phase before the next one starts.Every change is logged in the audit trail.