{"slug": "we-spent-months-building-ai-agents-then-we-deleted-them", "title": "We spent months building AI agents. Then we deleted them", "summary": "Runnit has removed its fleet of specialized AI agents and rebuilt its architecture around a single intelligence called Ru that loads capabilities on demand, after finding that newer large language models with larger context windows can naturally switch between tasks without needing separate agents. The company spent months building multiple agents for planning, research, scheduling, and writing before concluding that the multi-agent design was based on outdated model limitations. Runnit's founder argues that progress often means removing complexity rather than adding it.", "body_md": "Over the last few months, we've spent a lot of time building AI agents into Runnit.\n\nNot one or two of them. Lots of them.\n\nEach had a specific responsibility. One focused on planning, another on research, another on scheduling, another on writing. They all had carefully crafted prompts, distinct behaviours and their own area of expertise.\n\nAt the time, it felt like the right architecture.\n\nThe original LLMs we were building on (which seemed like only yesterday) had much smaller context windows (64k-128k), so breaking work into specialised agents wasn't just cleaner, it was necessary to produce even semi-accurate results. Each agent only needed to understand one problem, which meant it could use its available context more effectively.\n\nAs the models improved though, we slowly realised we were still designing around yesterday's limitations.\n\nThe more we tested newer models with larger contexts, the more we found they could naturally switch between different kinds of work without needing to become a completely different assistant. They could plan, write, research and reason within the same conversation, provided they had the right information available at the right time.\n\nThat forced us to ask a simple question.\n\nDo these really need to be separate agents?\n\nIn the end, our answer was no. So we removed them.\n\nInstead of maintaining a growing collection of specialised agents, we rebuilt the architecture around a single intelligence ([introducing Ru](https://runnit.io/articles/introducing-ru-runnits-ai-operator)) that loads capabilities when it needs them.\n\nResearch isn't an agent.\n\nScheduling isn't an agent.\n\nWriting isn't an agent.\n\nThey're capabilities and directives. They're pieces of expertise that can be loaded for a task and then unloaded again when the work is finished.\n\nThe change sounds small, but it simplified almost everything.\n\nThere's one consistent understanding of your business. One conversation. One memory. One place for organisational knowledge to live.\n\nWhen speed matters, that same intelligence can simply create parallel instances of itself, each loading the directives required for its own piece of work before bringing everything back together. From the outside it still works in parallel, but underneath it's the same intelligence solving different parts of the problem.\n\nWe also found ourselves deleting a surprising amount of complexity. We no longer had to maintain dozens of individual prompts, coordinate hand-offs between separate agents, or worry about different agents gradually drifting apart as they evolved.\n\nNone of this means specialised agents are obsolete. There are still good reasons to isolate work, use different models or separate permissions.\n\nBut I do think many of us are carrying architectural decisions that were made for models that no longer exist.\n\nIf we were starting Runnit again today, I don't think we'd begin by asking, \"What agents should we build?\"\n\nI'd start with a different question.\n\n\"What capabilities does a single intelligence need?\"\n\nSometimes progress isn't adding another layer.\n\nSometimes it's realising you don't need it anymore.", "url": "https://wpnews.pro/news/we-spent-months-building-ai-agents-then-we-deleted-them", "canonical_source": "https://runnit.io/articles/we-spent-months-building-ai-agents-into-runnit-then-we-deleted-them", "published_at": "2026-07-21 12:10:05+00:00", "updated_at": "2026-07-21 12:22:40.315982+00:00", "lang": "en", "topics": ["ai-agents", "large-language-models", "ai-products"], "entities": ["Runnit", "Ru"], "alternates": {"html": "https://wpnews.pro/news/we-spent-months-building-ai-agents-then-we-deleted-them", "markdown": "https://wpnews.pro/news/we-spent-months-building-ai-agents-then-we-deleted-them.md", "text": "https://wpnews.pro/news/we-spent-months-building-ai-agents-then-we-deleted-them.txt", "jsonld": "https://wpnews.pro/news/we-spent-months-building-ai-agents-then-we-deleted-them.jsonld"}}