The analyst agent for every team.

Agents that answer questions from your ERP, CRM and spreadsheets, explain what changed against budget and send the brief before the meeting. Each person sees only the data they may see.

Agents already doing real work.

70%+

less manual reporting effort

Kolektif House

Agents at work in enterprise teams

  • IKEA
  • Kolektif House
  • Les Benjamins

Ask. See the answer and where it came from.

A question in plain words returns the number, the chart and the query behind it, from the systems you already run. The same agent prepares the weekly report and tells you what moved. Live in four weeks.

A question about margin answered with a chart, a table and a proposed follow-up

From the question to the decision

One agent that answers, reports and follows up.

  1. Ask

    The number, and where it came from.

    Anyone asks in plain words. The agent queries your systems, shows the result with the query and the source, and says when the data cannot answer.

  2. Report

    The report arrives before you ask.

    Daily and weekly briefs against budget and plan, checked for anomalies before they go out, sent where your team already reads.

  3. Act

    From the finding to the follow-up.

    A deviation becomes a task with an owner. The agent drafts the message, opens the record and checks next week whether it moved.

Everything reporting runs on

The parts of the analyst agent, each under your access rules.

Questions on your data

Plain-language questions answered from your databases and files, with the query shown.

SQL · Charts · Tables · Sources

One definition for each number

Revenue, margin and overdue mean one thing: your definitions, written once and used in every answer.

Definitions · Company brain · Corrections remembered

Scheduled briefs

The daily and weekly reports built and sent on schedule, with what changed since the last one.

Daily brief · Weekly review · Budget against actual

Anomaly checks and alerts

Each report checked before it goes out; an alert when a number leaves its range, with the likely cause.

Data checks · Thresholds · Alerts · Owners

Updated 07:30 from the ERP.
Budget against actual, October

Built in a conversation, refreshed every morning

CategoryRevenueMarginBudgetGap
Power tools
€1,400k22.0%28.5%−€91k
Workwear
€900k30.0%31.0%−€9k
Fasteners
€1,800k33.0%33.0%€0
Total
€4,100k28.6%31.0%−€100k

Reports and apps

Documents, spreadsheets and small apps on your data, built in a conversation and kept up to date.

Documents · Spreadsheets · Apps · Decks

Pre-built for the reports you build by hand.

Each is a named workflow with an owner and a target, not a general assistant. We fit it to your systems and your definitions.

1

Management

Daily management brief

Every morning the agent reads yesterday’s sales, orders and cash from your systems, sets them against budget and plan and sends one page to the people you name. What changed since yesterday comes first.

2

Finance

Budget against actual, explained

At month end the agent sets actuals beside the budget line by line, finds the lines that explain the gap and says what caused each one. The query behind every figure is one click away.

3

Sales

Sales funnel review

Every week the agent reads the CRM, counts what entered, moved and stalled at each stage and names the deals that changed the forecast.

4

Finance

Overdue receivables view

The agent lists what is overdue by customer and age from your ledger, shows what changed since last week and drafts the reminder for each account. A reminder goes out with your approval.

5

Data

Data check before the report goes out

Before a report is sent the agent checks it: missing days, duplicate rows, totals that do not match the ledger, a number outside its usual range. A report that fails is held and its owner is told why.

6

Any team

Ad-hoc question from a manager

A manager asks in their own words. The agent answers from the systems that person may see, shows the query and says so when the data cannot answer.

Works on the data you already have.

No warehouse project and nothing replaced. The agent reads through the access you grant, in our cloud, yours or on-premise.

Databases

  • SQL Server
  • PostgreSQL
  • MySQL

Business systems

  • Your ERP
  • Your CRM

Reporting

  • Power BI
  • Your management reports

Files

  • Excel
  • CSV
  • Google Sheets

Live in four weeks.

Our engineers work inside your environment until the first workflow is live, then hand it to your team.

01.
Week 1: Explore
One report chosen with its numbers: the hours it takes, the sources behind it, who reads it. The business case follows within ten days.
02.
Weeks 2–3: Deploy
The agent is connected and builds the report in shadow; its figures are compared with your team’s.
03.
Week 4: Go live
Live under your access rules: the report sent on schedule, questions open to the people you name. Further reports follow.

How it looks in your industry.

1Retail and e-commerce
Sales, stock and margin by store and channel, with the reason behind each week’s change.
2Manufacturing
Output, scrap and cost against plan, in the brief every morning.
3Holding groups
One report across every business unit, each company’s figures on the same definitions.

Each person sees only what they may see.

Access follows your sign-in and groups, personal data is masked before any model sees it, and every query is logged. Hosted in our cloud, yours or on-premise.

AES-256

Encryption at rest, for databases, files and stored credentials

TLS 1.3

Encryption in transit, HTTPS on every connection

Anywhere

Hosted where you choose: our cloud, yours, or fully on-premise

Agents that are already doing the work.

Kolektif House
10K+
customer interactions handled
IKEA
4 weeks
live from kickoff
Kolektif House
10,000+
hours saved
Kolektif House
70%+
less manual reporting effort

Questions about the analyst agent.

Keep them for dashboards. The agent answers the question nobody built a dashboard for, explains the deviation and starts the follow-up.

Next step

Get started today.

Bring one report you build by hand. See it built, checked and explained on your own data in four weeks.