Maintenance for AI-built software

You built it. We make it something your department can run on. A named senior engineer takes over the maintenance of the tool you built with AI and reviews every change before it reaches your colleagues.

Software maintenance

We build with AI every day, so we know where AI-written code breaks: in permissions, in data and in calls to paid APIs. That is where the named engineer looks first.

$2,500
a month, the starting retainer
5
business days to a written Essential audit
1
named engineer on your tool, plus a backup

Who this is for

Department heads at established companies who built a tool with AI that a whole team, not only its builder, now uses every day.

  1. Situation 1

    You built it for your team, and now the whole department uses it

    You run operations, finance or sales, and you built an internal tool with Claude Code, Cursor, Lovable or Bolt to fix one process. It worked, more colleagues started using it, and every bug report now comes to you.

    What it settles: A named engineer learns the tool in the first 3 weeks and takes over the bug reports and the releases.

  2. Situation 2

    Keeping it running now eats the job you were hired for

    Every new user brings a request: a new report, a new field, another integration. You are the only person who can change the tool, so its backlog has become yours.

    What it settles: Small changes go into the monthly hours, and your colleagues ask the engineer directly.

  3. Situation 3

    You still want to build, and want every change checked before it ships

    You or a colleague keep adding features with AI. There is no staging environment and no way back after a bad deploy.

    What it settles: You keep building. The named engineer reviews each change on staging, and we own the release and the rollback.

Not for: a side project, or a tool nobody depends on yet: at that stake a freelancer costs you less than a retainer. A prototype that broke before anyone relied on it needs cleanup first (fixing AI-generated code that broke).

What the people who built these tools tell us

An operations leader built the company's internal tool alone, with an AI coding tool. It worked, the whole team came to depend on it, and its builder became the only person who could keep it running.

I'm trying to work myself out of a job.
An operations leader who built the company's internal tool with AI, on a call with us, September 2026
  • Monday at 7am, somebody's on a Slack message saying, this isn't working.
  • No sense check, no deployment, no staging environments, no second code reviews.
  • My tech debt is massive.
What the builder asked forWhat the retainer gives
People who use the tool talk to an engineer, without going through the builderA shared Slack channel with the named engineer
Nothing reaches the team without a second pair of eyesStaging and a review on every change
The builder keeps building with AIYour changes go through review on staging, and we run the release
The price of a retainer, against the price of a hirePublished tiers, compared with one developer salary below

What a tool with one maintainer costs

  • A bill many times the normal month

    A common failure: an AI-built dashboard skips a cached table and calls a paid maps API on every refresh. Nobody reviews the change, no spend alert is set, and the next invoice comes in many times the usual amount.

  • Your colleagues stop working

    When a tool the whole department uses goes down, the people who use it wait. The person who can fix it stops doing their own job.

  • A single person holds the tool

    The person who built the tool is the only one who can change it. On holiday, or after they move on, every change waits.

What the retainer covers

Software maintenance by 1 named engineer, for a tool your department built itself.

  • Ownership

    1 named senior engineer who knows the tool, a backup engineer who reads the same code, and a Solution Architect for architecture decisions.

  • Keep it running

    Error monitoring with Sentry, uptime checks, bug fixes, dependency and security updates, hosting and database care.

  • Safe changes

    A staging environment, code review on every change, and releases and rollbacks run by us.

  • Cost guardrails

    Spend limits and alerts on paid APIs (maps, SMS, AI models, hosting), set during onboarding.

  • Small improvements

    The reports, fields and screens your colleagues ask for, inside the monthly hours.

  • Response times

    Working hours are 9:00 to 18:00 Eastern European Time, Monday to Friday. Critical issues (the tool is down or data is at risk) get a response within 4 business hours, everything else by the next business day. Out-of-hours cover can be added to the contract.

  • Security and access

    An NDA before the audit. Engineers work through named accounts in your systems, with two-factor sign-in, and you can revoke any of them at once. Customer data stays in your systems, and staging runs on test or anonymised data.

  • Monthly report

    What broke, what shipped, the hours used and the risks found, in writing.

  • Knowledge that stays

    Documentation written as the work happens. Repositories, cloud accounts and credentials stay in your company’s name.

Who reviews AI changes

The cost of a mistake decides who signs off on a change.

ChangeExamplesWho makes itWho approves it for production
Small, cheap to get wrongCopy, a field, a label, a report filter, a dependency patchAI tools draft it, the named engineer checks and tests itThe named engineer
Business logicA scheduling rule, a pricing rule, a new screen your colleagues rely onThe named engineer, with AI assistanceThe named engineer and the person in your department who owns that rule
Expensive to get wrongThe data model, sign-in and permissions, payments, customer data, anything that calls a paid API or touches every userThe named engineerA Solution Architect
Changes you buildYou keep building in Claude Code, Cursor, Lovable or BoltYouThe named engineer reviews it on staging, plus a Solution Architect when it touches data, sign-in, payments or a paid API. We run the release

AI writes a lot of the code. A named engineer decides what reaches production and puts their name to it.

Your colleagues talk to the engineer directly

A shared Slack channel connects the people who use the tool with the engineer who fixes it. Bug reports stop going through you.

Talk to an engineer

How it starts

A call, the audit, 2 weeks of onboarding, then the monthly retainer.

  1. Step 1

    Discovery call

    When
    Week 0
    Who from our side
    Solution Architect
    What happens
    We look at the tool, who in the department uses it and what has broken, and say whether a retainer fits
    Your time
    45 minutes, free
    You receive
    A written yes or no, and the reason
  2. Step 2

    Software code audit

    When
    Week 1 (Essential audit)
    Who from our side
    Solution Architect and a senior engineer
    What happens
    We read the code your AI tool wrote, the hosting, the data and the paid APIs, and map what each failure would cost your department
    Your time
    About 5 hours
    You receive
    A prioritized risk register and what has to change before a retainer starts
  3. Step 3

    Onboarding

    When
    Weeks 2 to 3
    Who from our side
    The named engineer
    What happens
    Your code in a repository in your company’s account (Lovable and Bolt projects included), access in your accounts, a staging environment, monitoring, API spend limits and the first documentation
    Your time
    1 call to hand over access
    You receive
    Alerts that reach the engineer before they reach you
  4. Step 4

    Monthly retainer

    When
    Ongoing
    Who from our side
    The named engineer and the backup
    What happens
    Fixes, updates, reviewed changes and releases. Your colleagues report issues in the shared channel
    Your time
    Reporting issues
    You receive
    A tool a named engineer looks after
  5. Step 5

    Monthly review

    When
    End of each month
    Who from our side
    The named engineer and the Solution Architect
    What happens
    The report, the hours used, the risks found, and the tier sized up or down
    Your time
    30 minutes
    You receive
    The written report and next month's priorities

What it costs

Priced by the hours a month the tool needs. Any tool we didn't build starts with the audit.

TierForHours a monthPrice
Discovery callWhether a retainer fits your tool, with a Solution Architect45 minutesFree
Software code auditAny tool we didn’t build, before a retainer startsFixed scopeEssential $5,000, Standard $9,900
Keep it runningA stable tool your department must not loseAbout 40From $2,500 a month
Keep building safelyYou or your colleagues still build with AI and need review, staging and releasesAbout 80From $5,000 a month
Part-time teamRegular new features on top of maintenance: engineer, QA engineer, part-time project manager, Solution ArchitectAbout 160From $10,000 a month
For comparison: 1 US software developer1 person, with no backup and no second reviewerFull time$135,980 a year median, about $11,300 a month before benefits (BLS, May 2025)

Terms: a fixed monthly fee, not billed by the hour. 3-month minimum, then 30 days notice. Unused hours roll over for 1 month. Extra hours only with your approval, at your tier's hourly rate. All engineering time counts against the tier's hours, whether the named engineer, the backup or the Solution Architect does it; the monthly review call does not.

How we already treat AI-written code

We have no published retainer for a department-built tool yet. These are the rules our own AI-built work runs on, and every maintenance change on your tool gets them too.

  • Tests come with the code

    An internal product we built with AI from 1 requirements document has 82 database models, 343 API endpoints and about 454 automated tests, written alongside the code.

  • A security pass on AI-written code

    An automated security check runs over AI-generated frontend code on every generation. It has caught critical vulnerabilities before they shipped, including key-stealing injection paths.

  • An engineer on top of every AI output

    A first pass is generated quickly, then an engineer who owns performance and structure rebuilds it with real data fetching, so it holds up in production.

When other companies want your tool too

Some internal tools solve a problem other companies have. Then we turn the tool you built into a product you can sell: one SaaS system shared by many customer companies, or a white-label copy for each, keeping what works and rebuilding what can't serve many customers. Revenue Boosters, an amusement operator in Kentucky, needed new route management software for its own collectors after its old software failed, and wanted to sell it to other operators. We built the MVP in 3.5 months with that sale in mind. How that move works: turning an internal tool into a SaaS product, and our SaaS development services.

What can go wrong, and what we do

  • The backup engineer already reads the same code and sits in the monthly review, so the work goes on without a restart.

FAQ

A named senior engineer owns the tool: monitoring, bug fixes, dependency and security updates, hosting care, the small changes your colleagues ask for, and a review of every change before production. It covers tools built with Claude Code, Cursor, Lovable or Bolt. You get a written report every month.

If your colleagues use it every day, yes: when it goes down, their working day stops with it. The audit shows what a failure would cost your department before you commit to anything.

Working hours are 9:00 to 18:00 Eastern European Time, Monday to Friday. Critical issues, when the tool is down or data is at risk, get a response within 4 business hours; everything else gets one by the next business day. Out-of-hours cover can be added to the contract.

Retainers start from $2,500 a month for about 40 hours, with tiers from $5,000 for about 80 hours and from $10,000 for about 160. Any tool we didn’t build starts with a software code audit, from $5,000. Extra hours need your approval first and cost your tier’s hourly rate.

Hire when the tool needs a full-time engineer for years and your company can manage one. A retainer fits when the tool has to stay up but the work is part-time: you get a named engineer, a backup and a reviewer for less than 1 salary, and nobody to replace when someone resigns.

Yes. A shared Slack channel connects the people who use the tool with the named engineer. They report issues there, and you read the summary in the monthly report.

Yes, with the AI tools you use now. During onboarding your code moves to a repository in your company’s account; for Lovable and Bolt projects we connect that repository to the tool. Your changes then go through review on staging, and we run the release.

Yes. AI tools draft small fixes and dependency patches, and the named engineer checks and tests each one. Changes to data, sign-in, payments and paid APIs are written by the engineer and approved by a Solution Architect.

45 minutes with a Solution Architect, free. You get a written yes or no on whether a retainer fits the tool you built, and why.

Talk to the architect who would plan your retainer

Book a free discovery call

Schedule a call or send us a message

Pick a date that works for you and tell us about the tool you built.

Rodion Salnik

Rodion Salnik

CTO and Co-founder at Brocoders

Pick a date that works for you to see available times to meet with me and discuss your project needs. Looking forward to meeting you!

Book a call