Skip to content

Blog

AI velocity without engineering discipline fires slop

A framework sorts tech roles into four archetypes, and it reads like a raid party. AI rollouts without engineering discipline add damage output with no tank and no healer. Agentic engineering balances the party.

Written by

Specification and review

Published: 7 April 2026 Updated: 2 min read

Someone called AI-powered generalists “slop cannons.” The name is brutal, and accurate.

A framework just surfaced that reduces all tech roles to four archetypes. If you’ve ever played World of Warcraft, you already understand the party composition.

The four archetypes

The slop cannon (DPS). AI engineer, designer, PM. A high-velocity generalist who ships fast, breaks things and moves on. AI amplifies their damage output, and no one checks if they’re pulling aggro.

The stitcher (tank). Security, SRE, infrastructure. Absorbs the disorder the DPS creates, and exists because of the slop cannon: with this much output, the framework argues, you need very good people stitching it together.

The connector (utility). Deals with humans and buffs the party: coordination, keeping everyone informed, making everyone else more effective.

The governor (healer). The brake on an organization that keeps accelerating. Keeps the raid alive and fixes what gets through. The grown-up who decides what gets built and what doesn’t.

A party of all DPS wipes on the first boss

Every World of Warcraft player knows that a party of all DPS wipes on the first boss.

That’s exactly what most organizations are building. AI tool rollouts without engineering discipline create more DPS: faster damage output, no tank, no healer. Total party wipe.

  • More code shipped, less code that survives a month
  • More PRs merged, more incidents triggered
  • More features built, fewer customers satisfied
  • More velocity reported, less value delivered

The fix is the harness, not the person

The slop cannon isn’t a person problem. It’s a party composition problem. Same developer, different harness, completely different output.

  • Context engineering makes raw DPS directed DPS.
  • Quality gates are your tank, absorbing the damage before it reaches production.
  • Specifications before prompts are your healer, keeping the raid alive.
  • Shared conventions are your utility buffs, making the whole party stronger.

Engineer the DPS itself

As AI creates more DPS, demand for tanks and healers explodes. More damage output creates more disorder, which creates more need for someone to absorb it.

Unless you engineer the DPS itself.

That’s the difference between vibe coding and agentic engineering. One is a slop cannon pulling aggro with no tank. The other is a balanced raid clearing content.

Is your org running a balanced party, or wiping on the first boss with five slop cannons?

Further reading

  1. The costly mistake is hovering over your agent Writing 20 lines by hand is no longer the overcorrection to worry about. Staying in the loop so tight that the agent never gets room to work costs far more. 1 min readShort post
  2. Agent-written code breaks what code review was for Code review was never mainly about catching bugs. It kept the team's mental model aligned, and agent-written code breaks that loop before anyone has built the replacement. 2 min readShort post
  3. AI writes your tests, but not the one for the right problem A suite at 94% coverage, every test green, for a feature no user completed. The test that checks the business need depends on context only humans can supply. 2 min readShort post

All 23 posts on Specification and review

ClientsClient work

Test the constraint in your system

The Factory Readiness Assessment connects this research to your value stream, your data, and the next decision.

Training on this topic: Agentic Engineering for Developers and QA
See the assessment

Want to check one team first? The self-check: twelve yes or no questions, three minutes, instant result, no account. Take the self-check