Faced with rising costs from proprietary labs, sophisticated enterprise clients are building internal evaluation and routing systems. This allows them to use cheaper, open-source models for less complex tasks, optimizing for both cost and performance.

AI agents generate orders of magnitude more data than humans. The high storage costs of traditional CRMs like Salesforce, which can be 1000x more expensive than modern databases like Postgres, create a strong financial incentive for companies to move data out, weakening the CRM's moat.

To offset declining seat-based revenue, some SaaS vendors are drastically increasing API prices. This strategy backfires with AI agents, which generate massive data volumes. The high costs create a powerful incentive for customers to migrate their data elsewhere, accelerating the vendor's own decline.

What was once a significant moat for SaaS companies—complex data migration—is collapsing. An AI agent, '10k', completed the core lift of a 10-year Marketo data migration, a project quoted at $100k and one year by a human agency, in a single hour for just $14.21 in compute costs.

While powerful, a custom AI research agent can incur monthly API fees of $25-$30. This cost limits adoption to professionals whose time savings justify the expense, temporarily shielding mainstream content consumption habits from widespread disruption by such tools.

Relying solely on premium models like Claude Opus can lead to unsustainable API costs ($1M/year projected). The solution is a hybrid approach: use powerful cloud models for complex tasks and cheaper, locally-hosted open-source models for routine operations.

Directly connecting an AI agent to a platform's API (e.g., Facebook Ads) is risky. API rate limits and pagination mean the agent might only analyze a fraction of your data, leading to flawed decisions. A data warehouse is essential to provide a complete, reliable dataset for the AI to analyze.

AI agents get ad accounts banned by spamming API 'read' calls. To avoid this, sync platform data to a data warehouse. The agent should perform all analysis on the warehoused data, using the live API strictly for 'write' operations like uploading new ads, thus avoiding rate limits.

To prevent AI agent usage costs from spiraling, GitHub expects the solution will be intelligent model routing. These systems will automatically select the most efficient and cost-effective AI model for a given task, such as using a cheap model for simple refactoring instead of a powerful, expensive one.