AI workshops and trainings don't work.
Rebasing does.
The classroom model is dead, because group averages don't ship code. We coach every engineer at their own level, inside their own codebase, until AI-native is just how they work.
Trusted by teams at
The "AI Slop" Trap
Vibe code is AI-generated code produced without structure, tests, or context discipline: unpredictable, unmaintainable, and expensive to clean up. Most teams ship it by default.
More Time Debugging Than Coding
Developers spend more time fixing AI-generated hallucinations than they would have spent writing the code themselves.
Hallucination Rabbit Holes
AI confidently invents non-existent APIs and libraries. Teams chase ghosts for days, completely destroying sprint predictability.
Context Rot
Unstructured AI generation turns your codebase into a black box. No one understands how it works anymore.
How Rebasing Works
A Cognitive Rebase is a structured transformation of how an engineering team thinks about and works with AI: every engineer is profiled, then coached inside their own codebase through three engineering disciplines. We don't just prompt; we engineer.
Executable Specs (TDD)
Turning business logic into deterministic anchors that guide AI precisely. The AI writes code to satisfy tests, not the other way around.
Context Isolation
Using Hexagonal Design to limit the blast radius of AI changes. Your codebase is structured so agents only need a tiny, relevant slice of context.
Small, Verifiable Steps
High-frequency feedback loops to prevent context drift and hallucinations. Microscopic, instantly-verifiable steps AI handles with near 100% accuracy.
Delivery model
Three pillars. One outcome:
an AI-native team.
Team-wide onboarding
Every engineer starts on the same page, with their context already mapped.
Personalized micro-coaching
Clone your leadership, or use our AI avatar coach. Either way, every engineer learns the tracks at their own level.
Context Bankruptcy
The Claude MD Effect
The Productivity Illusion
Weekly upskilling reports
You see exactly who is adapting, who is stuck, and where to intervene; every week.
Team-wide onboarding
Every engineer starts on the same page, with their context already mapped.
Personalized micro-coaching
Clone your leadership, or use our AI avatar coach. Either way, every engineer learns the tracks at their own level.
Context Bankruptcy
The Claude MD Effect
The Productivity Illusion
Weekly upskilling reports
You see exactly who is adapting, who is stuck, and where to intervene; every week.
Beautifully presenting the holistic way of thinking that needs to be adopted when developing software with AI.
Not for everyone.
IT Services &
Nearshore Teams
Your clients are already asking if your team is AI-ready. Now you can prove it, with engineers who ship AI-augmented code that holds up to enterprise review.
FinTech / Banking
Tech Centers
AI-generated code in regulated systems has to be deterministic and auditable. Ours is, and we'll teach your team to keep it that way.
→ If your team ships code that touches production,
→ and AI is part of the workflow,
→ we should talk.
Two things most won't say out loud.
Individual contributors should not be reading code anymore.
The engineers who will define the next decade are the ones directing AI; not the ones debugging what it wrote.
CTOs should not leave their teams to natural selection.
Some engineers will adapt. Most won't; not without structure, visibility, and a deliberate path forward. That's your job to provide.
See the Method in Action
Short clips from live sessions; real methodology, no fluff.
Practitioners, Not Theorists
They've trained engineering teams on disciplined AI development across European time zones.
Tsvetan Tsvetanov
Co-Owner @ Camplight
Product-oriented software engineer with 12 years of experience. AI-Native engineer who continuously refines systems and principles for software development without writing code manually. Has worked on enterprise search systems and large-scale microservice architectures powering transit systems in major cities.
LinkedIn
Martin Martinov
Co-Owner @ Camplight
Programming since 1996 on 8-bit machines, giving him perspective across the entire technological evolution up to modern Kubernetes architectures. Sees legacy systems as foundations to build on, not burdens. A pragmatist over hype, with experience in critical sectors like healthcare and finance. Helps teams deploy intelligent automation that reliably takes over routine work.
LinkedIn🇧🇬 Based in Bulgaria. Delivered in English or Bulgarian. 🇬🇧
Recent Programs Delivered via 
Grow
Live programs with real engineers, real codebases, and measurable results.
Beyond Vibe Coding: Disciplined AI Development
Learn to direct AI with structured methodology instead of manually debugging generated code. Build a complete project from scratch using TDD + AI.
Transform Legacy Systems with Agentic AI
Use AI as a partner for understanding, documenting, and safely transforming legacy code using the Test-First AI Modernization (TFAM) Framework.
The Cost of AI Slop
Calculate how much undisciplined AI usage is costing your engineering team annually.
Adjust the sliders, then unlock your estimate
Grounded in 2025–2026 industry research
of developers frustrated by AI code that's "almost right"
Stack Overflow 2025 (n=49K)
more issues in AI-coauthored pull requests vs human-written code
GitHub 470-repo analysis
of developers don't fully trust AI-generated code accuracy
Sonar State of Code 2026 (n=1,149)
increase in duplicated code blocks since AI coding tools went mainstream
GitClear 2025 (211M lines)
Engagement Models
Choose the depth of transformation your team needs.
Frequently Asked Questions
Everything you need to know about Cognitive Rebase and working with us.
Methodology & Technicals
Engagement & Delivery
Logistics & Pricing
Book a 30-Minute Fit Call
A quick, no-pressure chat to see if we're the right fit for your team.