Welcome to Edition 3 of Pipeline – Counter’s monthly snapshot of the tools and technologies our Tech Leads are exploring. Each month, we curate a concise collection of insights from the field, highlighting what we’re using, learning about, or monitoring. In under 100 words, we break down what you need to know.
China enters the AI wars
Spooking both the Tech industry and the US Stock Markets, the new Chinese Startup DeepSeek announced its latest reasoning model this month, which some believe rivals OpenAI’s o1 on certain benchmarks. Is it any good? We’re playing with it (very carefully!) so we will let you know.
Considering a new JS stack?
The Bun runtime took another step forward this month. With integration of cloud API’s like S3 and SQL (currently Postgres) built in, out of the box.
Agent evaluation
The new movement around AI adoption is the usage of agents. Specifically agents to perform tasks without human intervention. You might have seen examples where a developer uses an AI code assistant such as GitHub Copilot and the agent generates code for them. Another example might be using an agent to get up to date production information and help a consumer make an informed decision. Evaluating the performance of an agent can be complicated with various aspects to consider. Google Vertex have released a public preview of their managed evaluation service.
