Author moves from Microsoft to Salesforce, uses AI agents daily
Author notes a shift from a Microsoft Windows AI team role to a research engineer position at Salesforce AI Research, spending about $3,000/month on token usage. The daily workflow still relies on large language models for coding, research, summarization, transcription, voice, writing, art, music, and new video tasks, and the author has drafted a personal AI policy to guide usage.
Key points
- Author spends ~$3,000/month on AI tokens while working at Salesforce AI Research.
- Author uses 5–6 Claude Code agents to prototype features in about 20 minutes.
- Google AI summaries now dominate search results, often providing quick product specs.
A key change is the move to agentic coding: the author now manages 5–6 Claude Code (or OpenCode, Codex, Pi) agents that prototype features in roughly 20 minutes, test audio code on macOS and iOS, and even add noise‑cancelling to Mumble. The agentic approach replaces thousands of hand‑written code with a few hundred, but juggling multiple agents can be exhausting.
The author also reports that Google’s AI‑summarized search results now dominate the first page, offering quick specs for products like HP Reverb G2, though occasional inaccuracies occur. Claude is used for tutoring, fine‑tuning Mistral, and exploring legal documents, highlighting both the power and the limits of current LLMs.
The story so far
3 episodes →- Author moves from Microsoft to Salesforce, uses AI agents dailythis story
How I use AI (Oct 2026)
ben.stolovitz.com · 31 October 2026
Loading the full article…
This text was published by ben.stolovitz.com and written by Ben Stolovitz. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Hacker News discussion · 44 pointsnews.ycombinator.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Agents & Tools
All →- Boson AI’s Higgs Realtime scores higher than GPT and Gemini in interruption recovery · 1 src
- Vertafore makes Velocity Change Request Agent available in Surefyre · 1 src
- Google releases 85-task engineering benchmark and agent competition with Mercor · 1 src
- Apple opens Siri AI to developers via iOS 27.3 betas · 2 src
- Apple enables Siri AI actions in third-party apps via iOS 27.3 · 1 src
Comments
via GitHub Discussions