enterprise software

Bilingual

Here Comes (Forward Deployed) Everybody
2 months ago
- Salesforce's Headless 360 shifts implementation labor to customers, making them assemble software from APIs rather than receiving finished products.
- Major enterprise vendors are expected to follow Salesforce's lead, requiring companies to have in-house roles to configure generic AI and APIs for specific workflows.
- The 'Pit Crew' role is proposed for each department (marketing, finance, etc.) to translate AI capability into actionable tools, analogous to a racing pit crew tuning the car.
- Two visions emerge: 'Substitution' (fewer jobs, same work with AI) and 'Multiplication' (more jobs as productivity drives demand expansion).
- Historical parallels include Piggly Wiggly's self-service grocery model and Clay Shirky's thesis on cheap coordination enabling ad-hoc groups; now cheap building enables atomized specialized work.
- The Multiplication outcome requires systematic records and practices beyond just hiring more Pit Crew members.
Non-performant software and its incentives
2 months ago
- Re-listening to Software Unscripted #78 highlights Casey Muratori's belief that performant software has a market opportunity because most software is garbage, and users value performance in games.
- The author agrees modern software performance is poor, citing Discord using 400MB RAM idle, and blames dependency stacks prioritizing developer experience over efficiency.
- Muratori's claim that market forces drive performance in enterprise software is contested: enterprise contracts are sticky, switching costs high, and decision-makers are often not end-users.
- Enterprise software buying is risk-averse (e.g., 'nobody got fired for IBM' effect); vendors with better demos and sales win over those focused on performance.
- Performance improvement is a cultural failing in the industry, but turning the tide is harder than simply building a performant product due to structural market barriers.
Microsoft 365 Copilot adoption is under 4.5%, only 1% use it weekly
3 months ago
- Microsoft 365 Copilot has low adoption, with less than 5% of commercial customers paying for it, and only 20-30% of those users open it weekly.
- The paid Microsoft 365 Copilot add-on costs $21-$30 per user per month on top of existing subscriptions, offering priority AI access, work data integration, and agents like Researcher.
- Microsoft is adding third-party models like Claude to Copilot's paid tier, indicating low confidence in its own AI model, while relying on deep app integration as a selling point.
- Copilot faces reputation issues due to forced integrations into apps like Notepad and File Explorer, leading to user backlash and the nickname 'Microslop'.
- Microsoft is restructuring Copilot amid competition from Gemini and Claude, with web market share around 1% and GitHub Copilot losing mindshare to rivals like Cursor.

enterprise software

Here Comes (Forward Deployed) Everybody

Non-performant software and its incentives

Microsoft 365 Copilot adoption is under 4.5%, only 1% use it weekly