--- type: source source_kind: x post title: Thread by @matt_slotnick source url: https://x.com/matt_slotnick/status/2026697640726114699 author: - "[[@matt_slotnick]]" published: 2026-02-25 captured: 2026-02-26T17:29:11-05:00 status: - used description: tags: - reference topics_auto: entities_auto: related_auto: distilled_to: "[[26-02-26 System of Record + Agents = Future of Software]]" used_in: --- [[26-02-26 XYZ]]## AI Distillation[[26-02-26 System of Record + Agents = Future of Software]] --- clipped_from: https://x.com/matt_slotnick/status/2026697640726114699 clipped_date: 2026-02-26T17:29:11-05:00 title: Thread by @matt_slotnick tags: article, summary, AI --- **Thesis:** The core argument is that future software will split into systems of record for managing work and agents for doing work, each with different business models. Systems are reliable foundations with predictable contracts, while agents are pay-per-use, low-margin, and high-volume. Growth will come mainly from agents, forcing software providers to adapt. This collaboration between systems and agents will drive productivity. **Key Mechanisms / Ideas:** - Systems of record (foundations, manage work, reliable contracts) - Software agents (workers, pay-per-work, elastic scaling) - Business models (per seat/data fees vs. outcome-based/token markup) - Economic law for agent pricing (yet to be named) - Collaboration between systems and agents - Integration with human workers and governance - Build vs buy decisions - Incentives alignment across layers **Why This Is Important:** Strategically, this matters as it forecasts a shift in software architecture and economics, impacting how companies invest in and deploy technology. Intellectually, it challenges existing paradigms by emphasizing AI-driven automation and new pricing models, guiding future innovation and competitive dynamics. **Open Questions:** - How to align incentives and governance between systems and agents - Defining the economic law for agent pricing and its constraints - Impact on build vs buy decisions in this new landscape - Validity of the prediction amid skepticism (e.g., \\"this will age like milk\\") - Ensuring reliability and compliance as agents scale ## Full Content **Matt Slotnick** @matt\_slotnick [2026-02-25](https://x.com/matt_slotnick/status/2026697640726114699) the future of software is simple. you will have systems of record that manage work, and you will have agents that do work. both are software, but they will necessarily have very different business models. the largest vendors will provide both --- **Matt Slotnick** @matt\_slotnick [2026-02-25](https://x.com/matt_slotnick/status/2026697905994870977) systems of record are foundations. they manage work, and direct workers (both human and agent), and make sense to be larger, more predictable contracts. because you need them to be there and you need them to be reliable. you pay for that --- **Matt Slotnick** @matt\_slotnick [2026-02-25](https://x.com/matt_slotnick/status/2026698415967711457) software agents are workers. they do the work. and you'll pay only for the work that's done. these will be lower margin, high volume, and elastic. you'll pay for them as a markup on tokens, and that markup will forever be constrained by a yet to be named economic law --- **Matt Slotnick** @matt\_slotnick [2026-02-25](https://x.com/matt_slotnick/status/2026699355156259128) to understand where software is going, you need to internalize that these two categories of SOFTWARE will work in concert but that growth will come predominantly from agents if as a software provider that unsettles you, you need to dramatically rethink everything you're doing --- **sur4js** @sur4js [2026-02-25](https://x.com/sur4js/status/2026795947104366779) this will age like milk --- **Matt Slotnick** @matt\_slotnick [2026-02-25](https://x.com/matt_slotnick/status/2026798810639511593) we'll see --- **Tom** @latentsaint [2026-02-25](https://x.com/latentsaint/status/2026767056276685126) What happens to build vs buy decisions in this world? --- **Capital Accounts** @CapAccounts [2026-02-25](https://x.com/CapAccounts/status/2026698250192040264) \*\*\*And you will have high performing FTEs deeply engrained in integration, execution, and review --- **Cole Benefield** @floinkus [2026-02-25](https://x.com/floinkus/status/2026699526447284538) nailed it!!! we realized this and decided to pursue both cause you also get some special advantages when you control both pieces --- **Brad Buss** @bgbuss [2026-02-25](https://x.com/bgbuss/status/2026702695080538114) Agree with you and there will be agents that operate on TOP of the systems of record not just INSIDE them. Observing how the software vendors deal with this reality will be interesting. --- **Joshua Poddoku** @JoshuaPoddoku [2026-02-25](https://x.com/JoshuaPoddoku/status/2026732677328404772) Systems and agents collaborating will unlock real world productivity --- **Ghost Gaps Intelligence** @ZeroCompWhop [2026-02-26](https://x.com/ZeroCompWhop/status/2026919873813602537) Matt’s point about different business models is the trillion dollar question of 2026. The "System of Record" (Salesforce, ERPs) will likely stick to per seat or data storage fees because they provide the "truth" and compliance. But the "Agents" will move toward outcome based --- **Kyle Tran** @KyleTranAI [2026-02-25](https://x.com/KyleTranAI/status/2026730736292642991) Agree with the thread, the real bottleneck is aligning incentives and governance across layers to keep margins while scaling agent work --- **Vendor Rip** @vendorrip [2026-02-26](https://x.com/vendorrip/status/2026981140318261590) This is the RIP stack --- **Tom** @latentsaint [2026-02-25](https://x.com/latentsaint/status/2026755022759837738) I agree with this view profoundly --- **Muldoon's Market Mayhem** @placigyunv6567 [2026-02-26](https://x.com/placigyunv6567/status/2026934059491013033)