Webhound is no longer just a place to start reports and datasets. Starting today, the home page is a conversational agent that can start research, read your existing work, run code, organize everything in a workspace, and remember your preferences over time.
We're calling it Webhound Agent. Reports and datasets still work the same way under the hood. What changed is how you interact with them.
What it replaces
Before this, using Webhound for serious research meant a lot of manual coordination. Power users were using external agents—Manus, Claude Code, custom scripts—to call the Webhound API, store results in Dropbox or GitHub, then use those same agents to read and analyze the output. It worked, but it was fragile and tedious.
The Webhound Agent replaces that entire workflow. It lives on the home page, has access to everything in your account, and handles the full lifecycle: starting research, monitoring progress, reading reports, analyzing datasets, running Python, and organizing results into folders.
What it can do
- •Start research and extraction from chat. Describe what you need and the agent proposes a session with an editable budget, model, and target folder. You confirm before anything runs.
- •Read and analyze your existing work. Ask about any report or dataset you've already run. The agent reads the full document, searches across sessions, and answers with citations.
- •Run code. The agent writes and executes Python for data analysis, charts, transformations, and file generation—using your actual datasets, not made-up data.
- •Work in the background. Research and extraction sessions run while you keep chatting. You get notified when they finish, and the agent summarizes what completed.
- •Remember you. The agent learns your research interests, budget preferences, and working patterns across conversations. It gets more useful the more you use it.
The workspace
The right sidebar is now a workspace—a nested folder structure for organizing your reports and datasets. Navigate it like Finder: double-click into folders, use breadcrumbs to go back, drag sessions between folders. All your existing sessions are already there.
The agent has a "working directory" concept. Set a folder in the chat's top bar, and the agent scopes its context to that project—it knows what's in the folder, creates new work inside it, and adjusts its suggestions accordingly.
Multi-step pipelines
Describe a sequence of steps — "research the best CRMs, then extract 30 of them with pricing, then write a buying recommendation" — and the system runs them automatically in order. Each step builds on the prior step's output. You approve the full plan before anything starts, and you can stop the pipeline at any point.
One-shot mode is still there
If you prefer the direct approach—pick a type, set parameters, hit go—that's still available. Toggle between Agent and One-Shot from the home page. The agent is the default, but everything you had before still works the same way.
Try it out
- •Open Webhound. The agent chat is the new home page.
- •Ask it to research something. It'll propose a session with a budget you can edit before confirming.
- •Ask about your existing work: "summarize my last 3 reports" or "what did I find about X."
- •Organize your workspace: create folders, drag sessions, set a working directory.
Questions? Email team@webhound.ai.