Meet Hunch. A Different Kind of AI Workspace

Most AI tools fall into familiar categories. They're either chatbots, automation platforms, or writing assistants built around large models.

Hunch doesn't fit into any of those. After sitting in on their session during the IT Press Tour in California, it became clear this platform is built for something else entirely. Hunch provides people with a space to break down complex thinking and then lets AI support that process with precision.

It isn't about throwing prompts into a black box and hoping for something useful. This is a canvas where you structure ideas, test what works, and turn rough workflows into reliable tools that others can build on.

Built for Thought, Not Just Output

David Wilson, co-founder of Hunch, opened with a striking observation.

"For 40 years, computers helped automate factory lines. But only recently have models been able to automate knowledge tasks in any meaningful way."

The point wasn't to pitch AI as a replacement for workers. It was a reflection on how much of today's work is done in front of a screen and how little of that effort taps into real capability. David continued,

"Even though this kind of work can be automated, most tools still make it painful. There's a gap between what's possible and what's usable."

That's what Hunch tries to solve. You describe your work the way you would explain it to a colleague. The system interprets, tests, and refines the approach. It's not magic. It's more like a set of thinking tools that help people delegate the repetitive parts while staying focused on what matters.

Everyone talks about productivity. But Hunch focuses on something more frustrating: the constant flood of tasks that drain attention and waste time. David referred to this as the busywork tax.

"So much of what we do on a computer is just moving content around. Formatting. Copying. Updating systems that don't talk to each other."

That mental overhead builds up. It wears people down. It turns smart roles into maintenance work. And worse, it prevents teams from completing tasks they know are worth the effort, such as writing better content, improving documentation, or following up with leads more thoughtfully.

Most current tools try to address this in one of three ways:

1. Rigid automation platforms that break the moment something changes

2. Prompt hacks where people manually paste into a model for each micro-task

3. Early AI agents that still require hours of configuration

Hunch steps into that gap. It takes your natural instructions and breaks them into a system that works without constant supervision.

Plain Language Becomes Process

One of the most impressive features of Hunch is its ability to handle task definition effectively. Instead of writing scripts or stitching together services with glue code, you describe what needs to happen in everyday language. The system turns that into what it calls a playbook.

Playbooks aren't flowcharts. They're readable documents that outline each part of the task in order. You can adjust steps, change models, or add constraints as needed. And because it's all in plain text, the entire structure is easy to understand and share. You need to track pricing changes across competitor websites.

In a traditional setup, you'd be dealing with brittle scrapers, constant maintenance, and the risk of failure if anything changes. Hunch interprets your goal and adapts as needed. If a link breaks, it looks for other signals. If a brand is added, it learns and applies that insight across the workflow.

This is not a fixed pipeline. It's a system that grows with the user. For example, the podcast producer template can be used to generate show notes (incl. timestamps & chapter summaries), title ideas, promotional tweets, and draft blog posts from a podcast audio or video file.

A Tool for Real Teams

You can use Hunch Solo. Many people do. But the platform shines when teams start using it together. Any process built by one person can be saved, reused, or extended by others.

What begins as a one-off automation becomes a shared asset. David explained this simply.

"If someone on your team writes a great prompt to summarise calls or extract insights, that shouldn't just live in their notebook. With Hunch, that becomes a tool others can run immediately."

Instead of each person building from scratch, teams accumulate capability. Every improvement, every new playbook becomes part of a growing system of collective knowledge.

It Started with Curiosity

The team behind Hunch includes veterans of Cape Networks, a company acquired by Aruba (now part of HPE). Their last product focused on reducing alert fatigue for network engineers. That same attention to clarity and user experience also runs through Hunch.

"We started this just to see what each model could do," David told us. "And we built what we needed to test them properly. That eventually became the product."

One of their early experiments went viral. A year-end LinkedIn review tool built with Hunch picked up hundreds of thousands of users. It pulled data from the public and created a highlight reel, complete with quotes and summaries. What started as a playful experiment turned into a case study in scalable AI workflows.

When asked about model performance, David offered a refreshingly blunt opinion.

"Most benchmarks are meaningless. They're about chat experience, not task accuracy."

Instead of comparing models through artificial tests, Hunch uses something more grounded. It selects models based on the nature of the task. Writing jobs might go to Claude. Image generation might use a diffusion model. If something breaks, the platform learns and adjusts.

Users can always override the model choice. But in most cases, the built-in logic gets it right. That flexibility also allows Hunch to introduce newer, more capable models over time without requiring users to modify their workflows.

Real Use Cases, Not Hypotheticals

One global advertising agency uses Hunch to prepare microsites for client pitches. It scrapes product pages, analyses brand language, and builds sample journeys. These aren't mockups. They're real, interactive sites. And they go from idea to finished draft in a matter of hours.

Another customer uses Hunch to review every sales and support call. The system extracts questions, feature requests, and common objections. Those insights feed a shared tracker, which then triggers other workflows, including content creation, FAQ updates, and knowledge base revisions. The entire loop runs through Hunch.

If Hunch continues on this path, it won't just compete with other AI tools. It couldl redefine how teams think about working with AI. Instead of treating these systems as one-off assistants, users will treat them as collaborators who can reliably, transparently, and repeatably carry out parts of their jobs.

What would you do with that kind of support? Would you finally build the workflows you've had in your head for months? Would your team start capturing more of the value they create every day?

I'd love to know have you tried building structured workflows with AI yet? What's worked for you, and what hasn't? Let's compare notes.

I will be speaking with the team at Hunch on the Tech Talks Daily Podcast in the next few weeks. If you have any questions you would like me to ask, please let me know, and you can also be a part of the conversation.