Praxeum demo: your AI co-workers, on your own AI

Slack
Google Drive
Notion
Confluence
Jira
Gmail

48,120messages, docs, and tickets read

refund-exceptionsdraft

onboarding-runbookdraft

tone-of-voicedraft

Understanding your company data

Praxeum starts with what your team already does: your messages, docs, and tickets. Read-only, and only the spaces you choose.

What it learns becomes skills: small, reviewable procedures your agents can follow. Each one starts as a draft your team approves before any agent uses it.

Finding the patterns

Your policies and documented procedures are the foundation. On top of them, Praxeum finds how the work actually happens: the recurring steps, the exceptions, the rules that never reached a doc.

FoundationPolicies & documented procedures

Refund exceptionsseen 12× · 4 people · 3 wks
Customer onboardingseen 9× · 3 people · 2 wks
Billing exceptionsseen 6× · 2 people · 4 wks

'No, do it this way.' The corrections are where the real procedure lives.

Opening a pattern

/refundrefund-exceptions
1. Check the plan type and timingSlack
  annual plan · booked 41 days ago#support-ops · Apr 12
2. Fork: refunded within the window?Service desk
  annual → up to 45 daysmonthly → 30 daysrefunded up to 60 days, 12× since Jan12 resolved tickets
3. Never refund after a chargebackNotion
  hard rule, no exceptionsFinance policy
/onboarda whole onboarding flow, end to end
1. Provision accounts and accessWorkspace
2. Send the welcome and first-week planSlack
3. Book the kickoff callNotion
/copythe small things, in Fernhill’s voice
1. Match the customer’s toneSlack
2. Reuse approved phrasingGmail
3. Keep the SLA promises honestNotion

Doc vs practice

Your policy says 30 days. Your team has refunded annual plans up to 60 days, 12 times since January. Which is right?

Recorded as a rule

annual plans → refunds up to 45 days

A person signs off before any refund goes out

You decide where the doc is wrong, on purpose.

No conflict? Praxeum still builds the skill from your policy, ready for your AI to run.

One human approval

This is a skill: the actual file an agent reads and follows. Review it, edit if you need to, then approve. Nothing reaches an agent until you do.

refund-exceptions / SKILL.mdproposedapproved

# refund-exceptions

Decide a customer refund from the plan and timing.

## steps

Check the plan type and timing.
Within the window? annual 45 days · monthly 30.
Never after a chargeback. Ask a person to sign off.

## tools

stripe · refunds.create

v3 · approved by Sarah Chen · Jun 30evidence current as of Jun 30

Added to your catalog

refund-exceptions@v3
onboarding-runbook@v5
tone-of-voice@v2
billing-exceptions@v1

Served over MCP

No new AI to buy. Praxeum serves the skill over MCP to the licensed AI your team already runs, so it acts the way your company does.

refund-exceptions@v3over MCP

Claude
ChatGPT
Cursor
Copilot
Gemini
your own agents
or hosted by Praxeum

readsuggestexecute with sign-offsexecute

The skill interpreter

Skills are Markdown, the format the big AI platforms are standardising on. But a model can read Markdown and still improvise. To make sure it runs the skill as intended, Praxeum's skill interpreter binds the parts that must never vary to reviewed, tested code.

refund-exceptions / SKILL.mdMarkdown

# refund-exceptions

Decide a customer refund from the plan and timing.

## steps

Check the plan type and timing.
Within the window? annual 45 days · monthly 30.
Never after a chargeback. Ask a person to sign off.

## tools

stripe · refunds.create

refund-exceptions / index.tsenforced

// the parts that must never vary
const WINDOW = { annual: 45, monthly: 30 }

function inWindow(days, plan) {
  return days <= WINDOW[plan]
}

annual at 41 days → in window, sign-off as always
annual at 60 days → outside the window, escalate
after a chargeback → never

Skills become a co-worker

Approved skills group into a co-worker: a named agent with a job, reachable where your team already works.

CCarmenFinance

refund-exceptionssign-off
billing-exceptionssign-off
invoicingexecute

Runs on Claude · your connection@carmen · carmen@fernhill

JJeffHR

onboarding-runbookexecute
leave-requestssign-off

Runs on ChatGPT · your connection@jeff · jeff@fernhill

Reach them in Slack, Teams, or email, or from Claude and ChatGPT with a command.

Fernhill · Slack

Today

Priya Nair9:41 AM

@Carmen a customer wants a refund, annual plan, 41 days in. Can we do it?

CarmenAPP

Message #support-ops

Mail

New message

ToJeff jeff@fernhill

SubjectNew hire: Marco Reyes starts Monday

Hi Jeff, Marco Reyes joins the support team on Monday. Can you get him set up before then?

Your company brain

The same reading that builds your skills also builds a brain your team can just ask, in Claude, ChatGPT, Slack, or Teams. Every answer is scoped to what that person can see. A real time saver, from day one.

Claude

ClaudeCompany brain · Praxeum

What’s our refund window for annual plans?

Reply to Claude…

Drift detected

Praxeum keeps reading. When the way of working moves, it notices, and it does not quietly average the change in.

“We’ve been holding annual refunds to 30 days since September. 45 was too generous.”

seen 8× · 3 people · since Sep#support-ops

refund-exceptions / SKILL.mdquestioned

annual plans → refunds up to 45 days ask a human

When reality moves, the skill goes back for review. Never silently averaged.

Re-review

You decide, not an average. Here is what changed, and why, before anything ships.

refund-exceptions / SKILL.mdquestionedapproved · v4

annual plans → refunds up to 45 days

annual plans → refunds up to 30 days

Change reasonTightened after Q3. 45 was too generous; confirmed with Finance.

Approved as v4. Back in service, and your co-workers pick it up.

Until a human rules, the agent holds off on the part in doubt.

Your company already knows how it works.

Now your AI does, too. Verified, audited, and yours.

Private beta, a few teams at a time. Bring your most painful repeated workflow. We’ll turn it into a skill your AI actually follows.

Join the waitlist praxeum.io