How It Works
From prompt
to production.
Not a suggestion. Not a draft. A deployment. Here's exactly what happens when you send a prompt to AfterBOTS.
Step 1 — Connect your repo
The Index
AfterBOTS connects to your GitHub repository and builds a deep index. Not just file names — actual understanding. Method signatures, class hierarchies, dependency graphs, naming conventions, error handling patterns, test coverage. We understand your codebase the way a senior engineer does after six months on the team.
→ Method-level chunking means we understand individual functions, not just files. No 200-file limit. No context window approximations. Real comprehension.
Step 2 — Task classification
Routi
Your prompt hits Routi first. Routi classifies the task type, assesses complexity from 1 to 5, and assembles the exact pipeline needed — no more, no less. A one-line bug fix does not get the full 10-agent pipeline. A major architectural change gets everyone.
→ Classification happens in under 2 seconds. The right team for the right task, every single time.
Step 3 — Requirements
The Spec
Reqi reads the prompt, reads the codebase, and writes a structured specification. Acceptance criteria. Edge cases. API contracts. Breaking change detection. By the time Codi starts writing code, there is no ambiguity.
→ Specs are attached to every implementation. Your team always knows exactly what was intended and why.
Step 4 — Implementation
The Build
Codi writes the code. Not scaffold code. Surgical, production-ready implementations that match your existing patterns, naming conventions, dependency versions, and error handling style. The code looks like it was written by someone who's been on your team for years.
→ Unchanged sections are marked clearly. You see exactly what changed and why.
Step 5 — QA and Security
The Review
Testi writes tests against the implementation. Every acceptance criterion becomes a test case. Boundary conditions, error paths, edge cases — all tested. Then Secu reviews for OWASP Top 10, injection vectors, auth flaws, and sensitive data exposure. A PASS from both means the code has been attacked and held.
→ Coverage reports, security findings with severity ratings, and prescriptive fixes — all included in every output.
Step 6 — Deployment
The Deploy
Mendi writes the deployment runbook. Environment variable documentation, migration scripts, CI/CD config updates, verification steps, rollback procedures. Nothing ships without a rollback plan.
→ Every deployment includes a post-deploy verification checklist. Mendi knows what a bad deploy looks like.
Complexity Levels
Routi assigns every task a complexity level. The level determines which agents activate.
| Level | Label | Time |
|---|---|---|
| 1 | Trivial | ~15 seconds |
| 2 | Simple | ~30 seconds |
| 3 | Moderate | ~60 seconds |
| 4 | Significant | ~2 minutes |
| 5 | Complex | ~3 minutes |
Ready to see it in action?
Get early access and run your first pipeline in minutes.