AMP Free: This AI Coding Agent is FREE but is it WORTH IT?

Key Concepts

AMP (Sourcegraph's CLI coding agent): A command-line interface (CLI) and VS Code extension designed to assist developers with coding tasks using AI.
AMP Free: A newly introduced free tier for AMP, supported by advertisements and user data sharing for training.
AI Coding Agents: Tools that leverage artificial intelligence to help developers write, debug, and refactor code. Examples include Klein Claude, Gemini CLI, and others.
System Prompts: Predefined instructions or contexts given to an AI model to guide its behavior and responses.
OSS (Open Source Software) and Frontier Models: Open-source AI models and cutting-edge, experimental models.
Grok Code Fast: A specific AI model mentioned as potentially being used in AMP Free, known for its speed and proficiency in Godo development.
Training Data Sharing: The practice of using user interactions and generated code to improve AI models.
Advertisements in AI Interfaces: The integration of ads within AI-powered tools, a novel approach for coding agents.
Monetization Models: Strategies employed by AI companies to generate revenue, including subscriptions, API costs, and advertising.
Model Serving Costs: The significant expenses associated with running and maintaining AI models.

AMP Free Tier: A New Monetization Strategy

AMP, a CLI and VS Code extension by Sourcegraph that functions as an AI coding agent, has launched a free tier called "AMP Free." This new offering is a significant development, as it introduces advertisements into the AI coding agent interface, a strategy not widely adopted by competitors yet. The free tier is supported by ads and involves the sharing of user training data.

Features and Functionality of AMP Free

AMP Free combines system prompts, tools, and a selection of top open-source and frontier AI models. The availability of improved and more cost-effective AI models in the last six months has made this free tier possible. The proposition is to offer "plenty of good tokens at a discount," powered by a coding agent that can utilize them, monetized through advertisements.

Key aspects of AMP Free:

Free of Charge: Users can access the service without direct payment.
Supported by Ads: Advertisements are displayed within the interface to offset costs.
Training Data Sharing: User data is utilized for model training, a practice common among many AI tools like Quen Code and Gemini CLI.
Model Selection: AMP Free utilizes a combination of system prompts, tools, and selected OSS and frontier models. While the specific model for the free tier is not officially disclosed, threads suggest it might be Grok Code Fast, potentially alongside other models.

Monetization Model: Ads and Data Sharing

The introduction of ads in a coding agent is a notable departure from the current landscape. While other AI tools share training data, AMP is among the first to integrate advertisements directly into the coding assistant experience. The ads are described as short text from specific sponsors related to developer tools, not personalized based on user activity. This approach is seen as an inevitable monetization strategy for many AI companies.

Performance and Model Quality

The video creator tested AMP Free and found its performance to be "nothing extraordinary," describing it as "kind of fine." Based on the speed of responses, it is speculated that the agent primarily uses Grok Code Fast.

Observations on performance:

Speed: The agent is noted to be very fast, consistent with the performance of Grok Code Fast.
Interface: Both the terminal and VS Code extension interfaces have seen significant improvements and are considered "pretty good."
Model Specificity: The creator expresses a desire for greater transparency regarding the specific model being used, as the current vagueness is a drawback.
Grok Code Fast Strengths and Weaknesses: Grok Code Fast is highlighted as being exceptionally good for Godo development, almost like a fine-tuned model for it. However, it is not as effective for other technologies like movie tracker apps, Svelte, or Vue, which is considered disappointing.

Critical Analysis and Value Proposition

The core concept of AMP Free—exchanging ads and training data for free tokens—is appreciated. However, the current implementation is questioned due to the use of Grok Code Fast, a model already available for free on platforms like Kilo Code without ads.

Arguments against the current value:

Model Choice: Using Grok Code Fast, which is freely accessible elsewhere, diminishes the appeal of AMP Free, especially when ads are involved.
Lack of Premium Models: The creator suggests that the offering would be more compelling if it included access to superior models like GPT-5, CodeX, or Sonnet.
Monetization Viability: The creator doubts the long-term success of third-party companies like AMP monetizing through ads alone, given the high costs of model serving. Unlike blogs where hosting is cheap, running AI models is expensive. Ads might only be sufficient with a massive user base, which most such tools lack.

Potential for Broader Adoption and Future Outlook

The creator speculates that if companies with their own model infrastructure, such as OpenAI, Anthropic, or Google, were to adopt a similar ad-supported model, it could be more viable. These companies could leverage such strategies to monetize free users, a common practice in big tech. The key would be to ensure the ads are not overly aggressive.

Future considerations:

Model Infrastructure Ownership: Companies that control model serving infrastructure are better positioned to optimize ad revenue against per-request costs.
Data Utilization: The value of shared data for companies like OpenAI lies in improving their proprietary models, which they then monetize through enterprise and paid user tiers.
Third-Party Challenges: Third-party agents without model infrastructure may struggle to cover high model costs solely through advertising.

Conclusion and Recommendations

AMP Free presents an interesting concept for democratizing access to AI coding agents through an ad-supported model. However, in its current iteration, the value proposition is weak due to the use of a freely available model (Grok Code Fast) and the presence of ads. The creator believes that for AMP Free to become truly worthwhile, it needs to offer access to more advanced AI models. The broader adoption of ads in AI coding agents is seen as likely, but its success will depend on the execution and the underlying business models of the companies involved.

The creator encourages viewers to share their thoughts, subscribe to the channel, and consider supporting it through donations or channel memberships. They also reiterate their Twitter handle, "AI OK King," for updates.