FAQ
Questions, answered straight.
No dodging, no 'contact sales to find out.' Winsen does three jobs at once, so people compare us to three kinds of tool. Here's the honest answer to each, with the receipts.
TL;DRWinsen is a workspace, a company brain, and a team of AI employees in one. That puts us next to copilots, memory layers, and AI-employee tools all at once. Below: how we differ from each, plus the straight answers on cost, ownership, and security.
Three jobs. One product.
Most tools do one of these. Winsen does all three, because they only work together.
KNOWS YOUR COMPANY
The company brain.
A living, sourced model of how you work, built from real work. Memory layers make you fill them. This one fills itself.
next to Glean · mem0 · Zep
DOES THE WORK
Walle, your teammate.
It drafts, sends, schedules, and logs, approval-first. Copilots answer and summarize. Walle acts, then waits for your yes.
next to Copilot · Gemini · Glean Assistant
HIRES AS A TEAM
AI employees by role.
Specialists who already know the job and your company. Automations follow a flowchart. Employees use judgment.
next to Zapier · 11x · Lindy
The basics
The short version, before the comparisons.
What is Winsen, in one breath?+
A workspace with a company brain underneath it. The workspace gives your team mail, calendar, tasks, drive, CRM, and pages in one place, and a colleague to manage it all. The brain is a living, sourced model of how your company actually works, built automatically from that work. Walle and the AI employees run on the brain and do the boring half of the job, approval-first.
So is it an assistant, a memory tool, or AI employees?+
All three, on purpose, because they only work together. An assistant with no memory of your company is a stranger. A memory layer that can't act is a filing cabinet. AI employees without a shared brain each start from zero. Winsen is the brain that fills itself, the teammate that acts on it, and the specialists you hire by role.
Why is it invite-only?+
We sit with every team we bring on so the brain is set up right and the AI actually works before you rely on it. Quality over land-grab. Ask for an invite and we will sort out access.
What does it cost?+
We are in early deployment, so pricing is not public yet, and you get three months free when you are invited in. When paid plans land: the workspace is per seat for your humans, AI employees are priced per employee, and you never pay for empty seats or work that did not happen. No custom enterprise quote process just to find out the number.
Can I just try it?+
Yes. The live demo is the real product with seed data, no signup. Click around the workspace, the brain, and the AI employees.
Knowing your company
The company brain, and how it differs from memory layers and search like Glean, mem0, and Zep.
bridge.winsen.ai
The Brain
847 facts91% avg confidence14 sources
winsen.mrr = $134,212
Stripe live · 6h
98%observed
tejas.signoff = "– Tejas"
you set · from 412 sent mail
pinned
winsen.icp = Series A SaaS, 15–60 people
52 deals analyzed · 5d
72%inferred
acme.next_renewal = Jul 14 · $148K ARR
HubSpot + Gmail · 38m
96%observed
winsen.runway = 11.4 months
Mercury + Ramp · 2h
100%observed
Click any fact → source, history, who's using it.
Is this just another wiki or vector database?+
No. A wiki is a pile of pages someone has to write and nobody updates. A vector database is a bucket of chunks you query and hope. The brain builds itself from the work already happening, keeps a source on every fact, and stays current because the work keeps happening. You are not maintaining it. It maintains itself.
Do we have to enter, tag, or import anything?+
No data entry, no tagging, no setup project. You connect the tools you already use and the brain assembles itself from real work: the emails, the tasks, the deals, the docs. The thing every memory product quietly needs, you typing facts into an API, is the thing we refuse to make you do.
Where do the facts come from, and how do I know they're right?+
Every fact points back to where it came from: the message, the document, the deal, the event. Click any claim and see the receipt. If the source changes, the fact changes with it. Nothing in the brain is a confident summary you have to take on faith.
How is this different from Glean or enterprise search?+
Search finds you a document. The brain knows how your company works and then does something about it. Winsen is the workspace, the brain, and the AI employees in one place, so the brain is wired into AI that acts, not a read-only answer box bolted onto your other tools.
Won't the brain go stale the moment our company changes?+
That is the exact failure mode we built against. A wiki goes stale because it waits for someone to remember to update it. The brain does not, because it is fed by the work itself. When the deal moves or a decision gets made, the brain already knows, because it watched it happen.
Doing the work
Walle acts, it doesn't just answer. How it differs from copilots like Microsoft Copilot, Gemini, and Glean Assistant.
bridge.winsen.ai
Walle, where did the Acme renewal land?
Acme renews Jul 14 at $148K ARR. Heads up: Mira's been quiet four days and health dropped to 54.
I drafted a check-in and held the renewal. Want to see it?
from acme.next_renewalfrom AI CSMapproval-first
Ask Walle anything…⌘⇧Space
Does it actually do things, or just chat at me?+
It does things. Walle sends the email, updates the CRM, moves the deal, books the meeting. The difference from the other action agents is that Winsen is approval-first, so nothing that matters happens without your sign-off. You stay the one who hits send.
How is this different from Microsoft Copilot or Google Gemini?+
Copilot and Gemini bolt an assistant onto the tools you already pay for, and they mostly draft and summarize. Winsen is the workspace and a team that does the work, and it acts instead of just answering. It also builds a model of how your company actually runs, which a copilot tab in someone else's suite cannot.
Does it really know my company, or is it a generic bot?+
It knows your company because it builds a sourced model of how you work, pulled automatically from the tools you already use. Every answer can point back to where it came from. It is not a chatbot guessing from public training data and a pasted-in paragraph of context.
Will it hallucinate and embarrass me?+
It answers from your sourced brain rather than making things up, every consequential action waits for your approval, and everything it does is logged. The category is full of confident bots that sent off-brand nonsense. Winsen would rather show you the receipt and wait for your yes.
Hiring the team
AI employees by role, and how they differ from automations and AI-workforce tools like Zapier, 11x, and Lindy.
Rita
AI SDR
Combed 22 fintech ops leaders overnight. Top fit: Mae Chen, Lattice Ops (0.91 match). Skipped 6 already in your pipeline. 14 drafts waiting for your read.
218
Sent · 30d
11.4%
Reply rate
17
Meetings
How is an AI employee different from a Zapier automation?+
A Zap does exactly what you wired it to do, in the exact order, forever. A Winsen employee has a job, not a flowchart. Rita reads the company brain, works out what the situation needs, and handles it the way a person in that seat would. You hire the role, you do not draw the diagram.
Do I have to build it, train it, or prompt-engineer it?+
You hire it. No canvas, no node graph, no prompt you have to get exactly right. The specialist already knows its craft and already knows your company, because it reads the shared brain. Setup is 'Rita, take outbound,' not three weeks of plumbing.
Will it do things without my approval?+
Not the things that matter. Winsen is approval-first: anything consequential waits for a human, every time, until you say otherwise. Routine, low-stakes work can be made autonomous once you are comfortable, and employees earn more autonomy the way a new hire does, shadow then supervised then trusted.
What does an AI employee cost versus hiring a person?+
A fraction of a loaded salary, and you can staff the whole team instead of one seat. The bigger difference is what stays: a person leaves and takes the context with them, an AI employee leaves the context in the brain you own. You are staffing a department that shares one memory.
Security and ownership
The boring stuff we got right before we shipped.
Do you train models on our data?+
No. Your data runs your company's brain, it does not train anyone's model, and we do not train across customers. The brain is scoped to your company, full stop.
Who owns the company brain?+
You do. It is customer-owned and exportable, a model of your company and not ours. If you ever leave, you take the brain with you. Most of this market locks that knowledge inside the vendor and hands you nothing on the way out. We think that is a weird thing to do to a customer.
Can the AI see things it shouldn't?+
No. Walle and the AI employees inherit the exact permissions of the person they work for. If you cannot see a doc, neither can your AI. There is no separate AI access tier and no backdoor.
Can I delete everything?+
Yes, for real. Export your brain, then delete it. Customer-owned means you are in control, including the exit.
Where's the compliance paperwork?+
Encrypted at rest and in transit, permission-aware retrieval, and a full audit trail on every action. SOC 2 Type II is in progress, and we would rather tell you exactly where we are than imply we are further along.
Using it anywhere
Winsen inside the agents and apps you already use, over MCP.
Can I use Winsen inside Claude Code or other agents?+
Yes. Winsen Bridge is an MCP service that exposes your whole workspace and the company brain to any agent or app that speaks MCP, including Claude Code, Claude Desktop, and Cursor. Your other agents get the same grounded access to mail, calendar, tasks, CRM, drive, and the brain.
What does the Bridge actually expose?+
Read access to the workspace and the brain (search, recall, who-knows-what, deal and account history) and the ability to take action (draft, schedule, log, update). The same tools Walle uses, available to whatever agent you point at the Bridge.
Is it still approval-first through the Bridge?+
Yes. Permissions and the approval gate travel with the Bridge. An agent in your terminal can draft and propose all day, but the consequential actions still wait for your yes, and every call is logged.
