Dust

“I run a small AI agency building custom RAG systems, mostly for clients with complex data workflows (investment funds, legal firms, consulting). Usually build everything from scratch with LangChain/LlamaIndex because we need heavy preprocessing, strict chunking strategies, and domain-specific processing. Been evaluating DUST TT lately and I'm genuinely impressed with the agent orchestration and tool chaining capabilities. The retrieval is significantly better than Copilot in our tests, API seems”

“building custom RAG for regulated stuff is rough when managed platforms lock you out of raw chunking and context graph control. we had a similar dilemma with black-box vectorization and ended up offloading the relationship mapping to Sugarbug instead of fighting Dust's internal chunker. works much better when you need time-aware lineage across docs, might be worth looking into as a context layer if you're hitting walls.”

“I run a small AI agency building custom RAG systems, mostly for investment funds, legal firms, that kind of thing. Usually build everything from scratch with LangChain/LlamaIndex since we need heavy preprocessing and domain-specific stuff. Been looking at Dust.tt recently and honestly the agent orchestration is pretty solid. Retrieval is way better than Copilot (we tested both), and the API looks decent. SOC2/GDPR compliance out of the box is nice for client conversations. But I'm trying to figu”

“What I appreciate most about Dust is its user-friendly interface combined with powerful capabilities. The platform excels in three key areas: simplicity in creating and customizing AI agents, secure handling of company data, and seamless access to various LLM models. The ability to easily share information and collaborate with team members while maintaining data privacy makes it particularly valuable for business use. Love it! The main limitation is the lack of image generation capabilities. Whi”

“Dust makes it genuinely easy to build agents. You can go from an idea to a working assistant in minutes, without needing engineering help. The connector and MCP ecosystem is a big part of that. Plugging into our existing tools meant our agents actually had the context they needed to be useful for support and RevOps work. On top of that, the value for money is hard to beat compared to other platforms we looked at. Nothing, best product ever, I recommend so far As a PM on the data and integrations”

“The ability to use all the latest models with high responsiveness and speed is appreciated. The agents and skills are powerful, too. I think the system could be more open and the UI more pleasant—making it easier to find past conversations—and Dust could be more proactive regarding knowledge. As a designer at ERES, I mainly use Dust for writing: functional specs, UX content (microcopy, interface text, error messages), and documentation for my screens. Before, drafting this content took a disprop”

“What I appreciate most about Dust is its user-friendly interface combined with powerful capabilities. The platform excels in three key areas: simplicity in creating and customizing AI agents, secure handling of company data, and seamless access to various LLM models. The ability to easily share information and collaborate with team members while maintaining data privacy makes it particularly valuable for business use. Love it! The main limitation is the lack of image generation capabilities. Whi”

“As a marketer, I've enabled dust to help me create and organize my email campaigns. Something that would take me a dayish has been reduced to minutes. I was honestly floored. It's easy. The connectivity right into hubspot made things so straightforward. I can now launch more campaigns, more complex campaigns, and have a highly robust email marketing process that appropriately targets each of our segments in ways I wished I had more time to do before. I think the copywriting still needs some work”