Platform · Governed & validated
Search is not truth. Validation is.
Storing knowledge is solved. Delivering it is solved. The part nobody has fixed is the one in between: deciding that an answer is true, and keeping it true. Not every answer has earned the right to be automated.
Reimbursement after referral
Up to 75% is reimbursed, after a referral from your GP.
Every channel that asked this question today got this answer.
What it costs when nobody owns it
Eleven months.
That is how long one wrong answer can live in a knowledge base nobody owns. Not because someone made a mistake, but because after publishing it, it stopped being anyone's job.
Reimbursement after referral
Fully reimbursed, no referral needed.
Every channel that asked this question today got this answer too.
The answer was not the problem. The silence around it was.
The life of one answer
An answer is not a file. It has a life.
Most knowledge bases only know two states: it exists, or it does not. Polly knows where every answer is in its life, and refuses to let an assistant use the ones that have not been through the gate.
Drafted
Written by whoever knows. Visible internally, invisible to every channel.
Owned
A named team carries it. Not the intranet, not "we". A team with a name.
Reviewed
Checked against the policy it comes from, and dated. This is the gate.
From here AI may use itLive everywhere
One version, delivered to every channel, every agent and every assistant at once.
Due again
Every answer has an expiry date. When it passes, the owner hears about it, not the customer.
Search tools ask whether an answer exists.This asks whether it still holds.
The same question
Nothing moves. Only what it is worth.
One answer, three states. The words on the page stay where they are. What changes is whether anyone owns it, when it was last checked, and whether an assistant is allowed to use it.
Someone wrote it down. That is all. No owner, no review date, and no channel may serve it. It is invisible until a team puts their name on it. A named team carries it, it was checked against the policy it comes from, and the next check is already scheduled. This is the only state in which an assistant may use it. Same owner, same wording, same place. But the review date passed eleven months ago. It is still being served, and nobody has looked at it since.
Asked 214 times this month
Reimbursement after referral
Covered up to 75%, I think, if the GP referred them. Someone should check the €1,000 threshold. Covered up to 75% of the contracted rate, with a referral from the general practitioner. Prior approval is required above €1,000. Covered up to 75% of the contracted rate, with a referral from the general practitioner. Prior approval is required above €1,000.
An assistant may not use this answer yet. Every channel that asked this question today got this answer. Every channel that asked this question today got this answer too.
Scattered knowledge does not become trustworthy by being easier to find. It becomes trustworthy the moment somebody puts their name on it.
The part of the work that never shows up in a demo, and decides everything that comes after it.
Louwman · one of the organizations running on Polly
Who this is actually for
Somebody is on the phone right now.
She has ninety seconds. The customer wants to know whether the repair covers a replacement device. She finds three answers: one from the intranet, one a colleague pasted into the team chat last spring, and one the assistant just produced with complete confidence. Two of the three are wrong, and nothing on her screen says which two.
So she picks the one that sounds most official, and hopes. Everything on this page exists for that ninety seconds. Not for the audit, not for the dashboard. For her, on that call, with the customer waiting.
Somebody is in the field right now.
He is standing at a street cabinet with a procedure to follow and a work instruction that changed in June. He opens it on his phone, in the app he already uses for the job, and it comes from the same source the people on the phone work from. Same question, same validated answer, whichever end of the organization it is asked from.
Built by people who did that job.
Polly did not start in a category playbook. It started in contact-center projects across Europe in 2015, where the same thing kept happening: good people giving confident answers nobody had checked. We built the fix we kept wishing existed.
Meet the team
Why it matters
Three questions most teams cannot answer.
Not from an auditor. From your own people, on the day an answer turns out to have been wrong for eleven months.
Who decided this was correct?
In PollyA named owner sits on the answer itself, not in a spreadsheet next to it. You can see it from the channel it was delivered to.
When was it last checked?
In PollyA date, and the next one. An answer that is overdue is visibly overdue, to the people who own it.
What did it say last March?
In PollyVersion history and an audit trail. What changed, when, and who changed it.
A wrong answer is a bad day.A wrong answer nobody owns is a liability.
In practice
Governance that people actually use.
Governance fails when it becomes a second job. This is built to run inside the work that is already happening.
Ownership by name
Every answer belongs to a team. No orphans, and no answer that everybody assumed somebody else was watching.
Review cycles with a date
Not "we should look at this some time". A next review, on the answer, that comes back to the owner.
Versions and audit trail
Every change kept, with who and when. The question "what did we tell people in March" has an answer.
Feedback from the floor
The agent who spots that an answer no longer matches reality flags it where they are, and it lands with the owner.
Change once, everywhere
A validated change reaches the chatbot, the agent screen and the help center in the same moment. No republishing round.
Only the validated feeds AI
Drafts and expired answers stay out of reach of every assistant. That is the difference between a search index and a foundation.
This is also where the law lands. The EU AI Act asks who is accountable for what an automated system tells people, NIS2 asks how you keep it secure, and the GDPR asks what you did with the data behind it. All three are answered by the same thing: an owner, a date and a trail. The harder requirements sit next door, on security and compliance.
Bring one answer you are not sure about.
In thirty minutes we put it through the whole life: owner, review, gate, delivery. On your own knowledge, not on a demo set.