Why We're Building Another AI Workspace
The AI workspace market is already crowded.
Buzz.xyz, Grok's bots, and incumbents like Slack are all adding agents. The direction is similar too: understand what people said, and turn it into work that actually happens.
So why did we step into a market this crowded?
I've long believed imagination is one of the strongest forces behind human progress.
People picture things that don't exist yet, and then make them real. We never used our brains only to survive. We imagine buildings, products, companies, technologies the world has never seen.
And then we execute.
Value appears the moment imagination is executed.
Markhub started there.
We began by trying to fix a designer's problem
I worked as a designer for a long time.
The thing that bothered me most was client feedback.
Feedback never arrives in one place. Some of it is said in a meeting. Some comes by email. Some is left in a messenger. Sometimes a client just calls.
The hard part comes after.
The designer has to gather that feedback, fold it into the work, produce the result, and share it back through yet another channel.
Miss one piece and the work gets redone. Schedules slip. Costs appear.
So at first I thought about it simply.
"Let's build a tool where clients and designers can exchange feedback properly."
That was the first Markhub.
It failed.
Looking back, the problem was real, but solving it didn't require a new tool. A careful person with good habits could get most of the way there.
So we tried a second hypothesis: put design-specific feedback features on top of chat.
That failed too.
What we were actually building was a slightly different Slack for design teams.
Two failures left us with one discovery
The product experiments failed, but one question kept surviving them.
What makes a designer work?
And from the other side:
What is the client actually paying for?
The deeper we dug, the simpler the answer got.
Request and result.
What a client wants to know is this:
"What became of the thing I asked for?"
"Did my requirements actually make it in?"
"When do I get it?"
And the designer is thinking:
"What has to happen to fulfill this request?"
"How long will it take?"
"What follow-up work does this create?"
"And how much value does this produce?"
Even the smallest freelance design project is more complicated than it looks. Analyze the requirements, research references, draft, review, take feedback, share revisions, get approval, deliver.
There's a request at the front and a result at the end. Everything in between is work.
This was never only about design
Drop that structure onto another industry and it gets more interesting.
Say a pharmaceutical company makes one product.
Plan the product, research ingredients, define the formulation, build a prototype. Name it, brand it, manufacture it, package it, market it, ship it, get it onto retail shelves.
Off the top of my head, that's close to ten stages.
Every stage produces quotes, spreadsheets, plans, reports, design files. All of them get revised.
By the time one product reaches the world, counting every version of every file, there may have been hundreds of updates.
And each one has an author, a reviewer, a decision maker.
Someone requests. Someone executes. Someone checks. Someone approves.
Completely different industries, remarkably similar shape.
Request → work → result → feedback → work again.
And one thing is present at every step.
Conversation.
Work starts in conversation
"Can you look into this?"
"Change this part like so."
"Let's go with this option."
"Redo the quote based on these numbers."
"The client approved. Start production."
The conversations we have every day at work aren't just text.
They carry requirements, knowledge, decisions, and things that need doing.
Until now, a person had to move after the conversation ended.
The meeting ends, someone writes the notes.
Someone reads the messages and copies out the tasks.
Someone updates the spreadsheet.
Someone writes the document.
Someone files the ticket for engineering.
Someone passes the revisions to the designer.
And then someone checks that all of it happened.
Which is where we arrived at a third hypothesis.
Could the conversation itself move the work?
So we collect conversations
AI changed the answer to that question, because a person no longer has to carry out every repetitive step.
From a conversation you can pull out the requirements, find the decision, create the task, attach the context it needs, and hand the doable parts to AI.
So Markhub today isn't a collaboration tool that stores conversations, and isn't a note-taker that summarizes meetings.
We're building a hub that collects the conversations happening across an organization and connects the knowledge and decisions inside them to real execution.
People focus on the conversation.
Markhub finds the work inside it.
Then it routes what belongs to people to people, and what AI can handle to AI.
Where it helps, it connects to a local agent so the work carries through into engineering, design, and documents.
What we want isn't more stored knowledge.
It's stored knowledge that turns into results.
Conversation isn't data. It's the energy for execution.
We're not collecting conversation in order to record everything.
The language that comes out of people's mouths and hands carries things that haven't become real yet.
"I want to build a product like this."
"What if we solved it this way?"
"Let's change this part."
"Let's have this ready by next week."
None of that is a result yet.
But executed, it becomes a product, a design, code, a contract, revenue.
I think the most important force behind AI is still human knowledge and imagination.
However strong AI gets, what to build, which problem to solve, and what result we want still starts with a person.
So what Markhub wants to collect isn't chat logs.
It's the moments right before human imagination and knowledge get executed.
And we want the distance between those moments and the actual result to be as short as possible.
When you talk, the work moves.
That's why we stepped back into a crowded AI workspace market.