Back to blog
This site is an independent third-party technical services platform offering aggregated access to multiple model APIs. It is not affiliated with, authorized by, or partnered with Anthropic, OpenAI, Google, or any other model provider.
Key Takeaways
-
One developer raised an important point: the gap between people who “know how to use AI” can be enormous, large enough to require an objective framework for measuring skill.
-
They proposed a five-level hierarchy—the lower the number, the more advanced the capability:
- Level 5: Chat · Search · Consultation
- Level 4: Research · Analysis · Slide Creation
- Level 3: Strategy · Markdown Documentation · Git Management · Business Automation
- Level 2: Development · Operations · Product · Marketing
- Level 1: Harnesses · CLIs · Building Your Own Supporting Tools
-
The value of this framework is not labeling people. It is showing a clear progression path: from treating AI as a search box to treating it as programmable infrastructure.
A Closer Look
From Level 5 to Level 4: From Questions to Deliverables
At the entry level, people treat AI as an upgraded search engine and conversational partner: asking questions, looking up information, and requesting suggestions. Most people stop here.
At the next level, users begin asking AI to deliver concrete outcomes: conduct research, analyze data, or create a slide deck. The difference is that you are no longer merely chatting with AI; you are integrating it into a production process.
Level 3: AI Starts Taking Over the Workflow
At Level 3, AI is no longer a standalone tool. It becomes embedded in your workflow: helping structure strategy, organize Markdown documentation, manage Git commits, and automate repetitive business operations.
The key shift is that you begin using AI to handle processes, not just tasks. This requires breaking down requirements, providing the model with the right context, and connecting the results into a pipeline.
Level 2: AI Becomes Part of the Business
At the next level, AI participates directly in core business functions such as development, operations, product, and marketing.
People at this level are usually already using APIs rather than only web interfaces, because business-grade use requires calls that are stable, programmable, and scalable.
Level 1: Building Your Own Tools
The most advanced level involves building your own harnesses—execution frameworks that support AI capabilities—CLIs, and supporting tools.
At this point, AI is no longer a product that you simply use. It becomes raw material that you orchestrate, package, and extend. You build tools for your own use cases that other people have not yet created.
What This Means for Developers
This framework highlights a practical reality: the further upward you move, the closer you get to the underlying model capabilities, and the more demanding your integration requirements become.
- At Levels 5 and 4, a web interface is usually enough.
- From Level 3 onward, you will almost certainly need APIs and automation scripts.
- At Levels 1 and 2, when building harnesses, CLIs, and business systems, you need a stable, programmable foundation that lets you switch models freely.
There is also a commonly overlooked pitfall: if you hard-code your tools to one provider’s SDK from the beginning, switching models later—or assigning different models to different tasks—will require rewriting authentication and calling logic.
The higher your level, the more expensive that technical debt becomes.
In Practice: Give Your Tools a Model-Switchable Foundation
Whether you are building a CLI or business automation, a practical approach is to standardize the integration layer around an OpenAI SDK-compatible interface and make the model a replaceable parameter.
Code0 is a multi-model API gateway that provides access to more than 300 leading models—including Claude, GPT, Gemini, and DeepSeek—through a single key. It is compatible with the OpenAI SDK and requires no major integration changes:
from openai import OpenAI
client = OpenAI(
base_url="https://hk.code0.ai/v1",
api_key="sk-your-key", # Get your key from console.code0.ai
)
def ask(model, prompt):
resp = client.chat.completions.create(
model=model,
messages=[{"role": "user", "content": prompt}],
)
return resp.choices[0].message.content
# One codebase, different models for different tasks—
# this is the foundation your tools need.
print(ask("claude-opus-4-8", "Review this architecture design.")) # Deep reasoning
print(ask("gpt-5.4", "Come up with five titles for this copy.")) # Creative ideation
print(ask("deepseek-v3", "Classify these logs by error type.")) # Cost-effective batch processing
By abstracting the model into a function parameter, your harness or CLI automatically gains the ability to route different tasks to different models. You do not need to restructure the tool whenever you switch providers.
Usage is pay-as-you-go, and failed requests are not charged. Available models and billing rules are subject to the information shown in the Code0 console.
Conclusion
This five-level framework is ultimately saying one thing:
Knowing how to use AI is a path from being a user to becoming a builder.
The further you progress, the more you need to treat AI as programmable infrastructure. And the first lesson of infrastructure is not to lock yourself into a single model.
Give your tools a switchable integration layer, and you will have room to climb from Level 3 toward Level 1.
