Devrecated Solutions01
Cycle time
Request accepted to change in production.
Reported against baseline
Devrecated Solutions
Agentic Engineering as a Service.Built for growing companies.
Are you or your team Devrecated? We’re here to help. Our products automate software engineering legwork so you can focus on your business. See below how we simplify, harden, audit, and automate software with agentic AI.
What we do
We do the work of a software agency, with agents doing most of it.
Clients bring us bugs, small changes and whole new apps. Agents write the code and the tests, and our team reviews the result before it ships. Our output grows with compute, so a new client doesn't mean a new hire. Each client's rules and past decisions are written into the system, so nothing is lost when someone leaves.
Shorten checkout to one step
- Requester
- product@acme
- Surface
- Dealer portal · checkout
- Intake
- Typed · stack skills attached
People who use your product open tickets for features and fixes. The agent reads the board, claims the work, and starts from a typed intake — not a chat paste.
Products
AutoDevelop at the center, with tools you can buy on their own.
AutoDevelop is our core product and has its own site. It comes with support hours, where our team builds for you, and depending on the package it includes our other products. DocPuppet and LayoffAnalyzer are available on their own and are an easy place to start. RiskAuditor, VibeKiller and MobileNinja are in preview.
Also in the catalog
DocPuppet
Release evidence for a running application.
Point it at a deployed application and its intended behavior. It plans the capture, writes stills and an optional video tour, writes how-tos and a changelog, and returns a verdict on whether the product matches the behavior the developer intended.
Stills and a WebM tour captured with PlaywrightHow-tos and a changelog built from real commit subjectsPass, fail, or uncertain verdict with CI-usable exit codes
DocPuppet detailsLayoffAnalyzer
A 0–100 read on delivery efficiency.
Scores a repository's delivery over rolling windows against a house midpoint of 50. One algorithm, shared with Autodevelop — a score, its history, and the inputs behind it, with no dollar amounts.
Rolling-window score per repository and developerWeekly compare job against your connected GitHub AppScore history and inputs — never a cost-per-developer figure
LayoffAnalyzer detailsMobileNinja
Store deploys without the certificate tax.
Automates shipping iOS and Android builds so a small team can reach the stores without living in certificates, listing forms, and review cycles. Included with Autodevelop when it ships.
iOS and Android release path in one placeLess certificate, listing, and review frictionIncluded with Autodevelop; standalone product coming soon
MobileNinja detailsRiskAuditor
One auditor for every code movement.
Scores commits, pull requests, releases, reverts, and ranges — findings, a score, and remediations from a single algorithm. Surfaces as MCP advice, a CLI report, and a GitHub Check that can deny the ship.
One score for commit, PR, release, revert, or rangeFindings and remediations, not a second house metricIncluded with Autodevelop; standalone product coming soon
RiskAuditor detailsExploitGenie
Authorized penetration testing with OpenHat.
Describe a scoped wish against systems you own or have written permission to test. When the wish is in scope, ExploitGenie produces the exploit code, explains how it was built, and explains how to remediate.
Scoped wishes only — customer-owned or permitted systemsExploit code plus build and remediation explanationAdd-on built with OpenHat; not included in Autodevelop
ExploitGenie detailsVibeKiller
Screen tweaks from people who are not engineers.
Operators describe copy, layout, and behavior changes. A session live patch can show the tweak before deploy; the durable change is a draft pull request opened from the ticket widget. Live database included; application source is not.
Allowlisted live patch for copy, style, and visibilityDurable change ships as a draft pull request$100/mo for one app, $50/mo each additional
VibeKiller details
01 / 06
Proof
What it has done so far.
- Optic Truck Works
- With AutoDevelop, we delivered more for Optic Truck Works in two to four weeks than in over a year with AutoDevelop's pre-alpha.
- Demo: Acme Fleet dealer portal
- Acme Fleet is a working dealer portal. We built the first version in under an hour, and it has had only a few hours of work since.
- Demo: healthcare admin portal
- We stood up a healthcare admin portal with AI vision in under an hour.
- Our own company
- Our marketing and content run on scheduled AI routines that we review, and AutoDevelop works through its own bug list every night.
How an engagement runs
Working with us, in four steps.
Every AutoDevelop subscription includes support hours. Your team can run it directly, or we run it for you.
01
Book a call
Tell us what is slowing the business down. We'll tell you which product fits, or whether AutoDevelop with support hours makes more sense.
02
Baseline
We measure cycle time, change failure rate and code turnover in your repository before anything changes.
03
Set up
AutoDevelop learns your stack, your rules and your backlog. Our kits plug into Cursor, Claude, opencode and Grok Bot, so it fits the tools your team already uses.
04
Ship and report
Changes are checked against your coding standards and the ticket's acceptance criteria before review. You get a regular report measured against your baseline.
Why now
A year ago this wasn't possible.
- Running AI got cheap
- GPT-3.5-level performance cost about $20 per million tokens in late 2022 and about 7 cents by late 2024. Source: Stanford AI Index 2025.
- Agents finish real work
- Agents can now work through a real ticket from start to finish and check their own output along the way.
- They still don't know your business
- Generic agents start every session knowing nothing about your system. AutoDevelop keeps your rules, decisions and past work, so it gets better the longer it runs.
What we report
Numbers you can check in your own repository.
Output volume is easy to inflate, so on its own it tells you very little. We baseline these three numbers in your code when an engagement starts and report against that baseline.
02
Change failure rate
Share of production changes that need a fix, rollback or patch.
Failed changes
Reported against baseline
03
Code turnover
Share of new code rewritten or deleted within 30 days, 90 days and 12 months. High turnover means the architecture isn't holding.
30 days, 90 days, 12 months
OpenHat Security is our partner company for security work.
Start here
Tell us what you'd want built.
Book a call and we'll walk through your business and where agents can take work off your team. If you'd like to see it first, try a change in the Acme Fleet sandbox.