What is Commons Intelligence?

A reporting tool with an AI analyst built in. Connect your systems, ask questions in plain English, and let Ben send you a briefing every morning about what changed and what to do.

What happens after the 30 days?

Pick a plan to keep going. There is no card on file during the trial, so if you do nothing, nothing is charged.

Does Ben see our whole database?

No. The model gets the shape of your tables and a small sample of rows. Counts and calculations are done in code, and credentials are never shared.

Can we use our own Anthropic key or self-host?

Yes. On Enterprise you can bring your own Anthropic key, run your own warehouse engine, or self-host the whole platform.

Do you offer training and onboarding?

Yes; we stay through launch until your team runs it.

CSO and NGO Systems · Commons Intelligence

Talk to your programmes.

Impact intelligence for the people who run programmes. Ask any question about your programmes in plain English and watch it turn into live reports. Ben, your personal AI analyst, watches them around the clock and tells you what deserves your attention.

Start 30 days free See pricing

30 days free · No credit card · Free training session

Scroll

Say it the way you'd say it to a colleague.

No formula bar, no aggregation dropdown, no join diagram. You type a sentence; Ben picks the chart, writes the query, groups the dates and puts it on your report.

01“Reading levels by district for the last term, monthly”A sentence becomes a chart, grouped the way you asked for it.
02“Which districts fell behind last year at the same point?”Period comparison, without anybody building a second report.
03The assumptions, in plain words“I took ‘at grade level’ to mean reading a Class 2 story.” Say the word if you use a different benchmark.
04Every figure opens onto its rowsDrill from the chart to the records it was drawn from.

A question asked in plain English, answered with a chart and its assumptions

Most tools stop at the report. This one keeps reading it.

The reporting layer answers the questions you ask. The intelligence layer tells you what you did not think to ask, and it runs whether or not anybody opens it.

01It runs every morning at 9Reporting waits for you to open it. Intelligence has already looked.
02It explains, not just showsWhat happened, why, what it affects and what to do next.
03It connects measures across reportsNot one chart at a time; everything you own, read together.
04It ranks changes by what they are worthAnd it flags the survivorship traps you would otherwise have to spot yourself.

The intelligence layer, ranking what changed across every report

One trusted view of your whole impact.

A multi-tab report built from connected systems, with Ben one click away

The report on the desk, and the morning briefing wherever you open it.

Reporting tells you what happened. Ben tells you what to do.

Ben is your personal AI analyst. He runs on a schedule, not on a button: a briefing every morning, a story when something specific happens, and a recommendation you can argue with.

01A briefing every morningWhat moved overnight across every report you own, and what he would look at first.
02A story when something happensWhat happened, why, what it affects, what is next and what to do, with a confidence score.
03What changed, ranked by worthNot everything that moved. The moves that are worth the meeting.
04A thread you can argue inTell him what the data cannot show. It shapes the next story about that area.

The morning briefing: what moved, why it matters and what to do

And the four things Ben won't do.

An analyst you cannot check is an analyst you cannot use. Each of these is a property of how it is built, not a promise about how careful we are.

He never changes your data quietly

Cleaning is proposed, not applied.

Every step comes with the row count it affects
Nothing runs until you accept it
Cleaning happens on a copy; your sources are read-only

He never hides his confidence

A score, and the sentence it rests on.

Two possible readings? You get both
The figures sit under every claim
Survivorship and mix warnings come first, not last

He never invents a finding

A quiet feed means a quiet programme.

A scheduled run that finds nothing posts nothing
Anomalies are measured against each metric's own normal range
Every claim is grounded in rows you can open

He never sees your whole database

The model gets the shape, not the contents.

A sample of rows, never the full table
Counts and calculations are done in code
One database per organisation, separated physically

How it works

Four systems in. One answer out.

The part that normally needs a data engineer happens in four steps, with no SQL, and your raw data is never modified.

01
Connect
Analytics, spreadsheets, databases, field tools and uploads. Tick the tables; pick a refresh.
02
Transform
Ben lines up columns across systems and proposes the fixes. You approve. It stays in sync by itself.
03
Report
Describe what you want, or take the controls. Thirty-plus chart types, no SQL either way.
04
Be told
Briefings, stories, changes and recommendations arrive without anyone pressing anything.

Ben already knows your sector's questions.

Thirty years of reporting across these fields went into how he reads them: the metrics that matter, the ones that mislead, and the trap each sector falls into.

Education

Enrolment, attendance, learning outcomes and who is falling behind.

Agriculture

Plots, seasons, yields and prices, with the weather in the picture.

Climate

Emissions, energy and restoration, measured over years not quarters.

Healthcare

Coverage, outcomes, follow-up and where care is dropping off.

Volunteering

Recruitment, retention, programme health and the hours behind them.

Fellowships

Applications, selection, cohorts and what fellows do afterwards.

Crowdfunding

Campaigns, donors, repeat giving and what each channel really costs.

Enterprise data

Many systems, many teams, one version of the numbers.

Properties of how it's built, not promises about how careful we are.

Commons Intelligence sits on top of your systems rather than replacing them, and everything it produces exports as PDF, Excel, CSV or through a documented API.

Credentials never come back

Passwords, SSH keys and tokens are encrypted with AES-256-GCM before storage. No endpoint returns them.

Raw data is read-only

Cleaning happens on a copy in a separate layer. Your source systems are never written to.

One database per organisation

Separation is physical, not a column in a shared table.

Permissions per action

Roles per entity and per action: view, create, update, delete.

Every sync is logged

What ran, when, how long it took and what it changed, including failed runs.

Your data can stay on your servers

Self-host the whole thing, or run in our cloud with a dedicated database of your own.

Ask, be told, prepare.

Built for the person who owns the number.

30+ chart typesAll reachable by describing what you want to see

15+ connectorsAnalytics, spreadsheets, databases, commerce, CRM and field tools

30 days freeNo credit card, and a training session with a human

0 lines of SQLFrom connecting a source to the morning briefing

Every step from data to decision, handled.

Four things every team asks for first.

01 · Connect
All your systems in one place
Analytics, spreadsheets, databases, field tools and uploads.
02 · Transform
Clean data, no engineer
Columns lined up and fixes proposed. You approve.
03 · Report
Describe it, get it
Thirty-plus chart types, no SQL either way.
04 · Be told
Briefings and stories
What changed, why, and what to do, without pressing anything.

Who it is for.

Not the person who owns the database. The person who owns the number.

Founders and programme heads

The number now, and whether to believe it.

A team working at laptops in a meeting room

Operations and programme teams

Data from five tools joined without an engineer.

A group in formal dress at an awards presentation

Finance and impact reporting

The monthly pack, current and traceable.

Two women standing together in an office

Marketing and fundraising

Channels, cohorts and retention in minutes.

Field and delivery leads

Field data turned into something a funder will read.

The Global Goals it moves.

Quality Education

Decent Work and Economic Growth

Industry, Innovation and Infrastructure

Reduced Inequalities

Partnerships for the Goals

Every feature

Everything you can do here, without writing a query.

Every plan has every feature, including the free month.

Ask

Build a chart by describing it
Ben picks the chart type, writes the query and groups the dates.
Ask a question of a report
Answers grounded in the same rows the charts are drawing.
Turn an answer into a widget
Save it to the report in one click.
Get a whole report generated
A starter report of twenty widgets, organised into tabs, from your connected data.
Ask what is worth reporting on
Ideas based on the columns you actually have.
Understand a source you inherited
A plain-language brief on an undocumented database.

Intelligence

Anomaly detection that stays quiet
Measured against each metric’s own normal range.
Segment attribution
When a total moves, which segment carried it, named and quantified.
Survivorship and mix warnings
Said first when an average improves because the population changed.
Recommendations with reasoning
What to do and why, with the argument shown.
Confidence on every finding
A score and the sentence it rests on.
The archive
Briefings and stories kept by date, so you can read back through a quarter.

Reporting

Multi-tab, cross-source reports
Several systems in one report, joined and kept consistent.
Power Filters
Describe how to slice the report and every widget follows.
Drill to the rows
Every figure opens onto the rows behind it.
Annotations
Mark the campaign, outage or launch on the timeline.
Conditional formatting
The exception is visible without reading every row.
Scheduled and shared
Password-protected links, PDF, Excel and image export.

Data transformation

Combine tables across systems
Ben lines up columns that mean the same thing.
Rename to business language
dim_ord_dt becomes order_date, and you can overrule it.
Standardise messy values
Spellings, dates and currencies fixed with row counts shown.
Remove duplicates
“I found 342 duplicate students”, with the rule it used.
Refuse to guess on ambiguity
A join that would double-count children stops and asks you.
Automatic rebuilds and drift warnings
Everything rebuilds on schedule; renamed columns are flagged.

Connectors and charts

Live now
Google Analytics, Google Sheets, MySQL and CSV upload.
In the library
Shopify, Stripe, Salesforce, HubSpot, Zoho, PostgreSQL, Snowflake, BigQuery, CommCare, SurveyCTO.
Commons programmes
Better Together and Campus Cloud: programmes, partners, cohorts and placements.
Thirty-plus chart types
Bars, lines, maps down to Indian districts, funnels, Sankey, pivots and KPIs.
Time handled properly
Grains, rolling windows, period compare and forecasts with a visible band.
Formatting that matches the country
Lakhs and crores or millions, with currency and date formats per field.

How the AI is used

Grounded in your actual rows
Every prompt carries a sample and the real values being discussed.
Arithmetic is not the model’s job
Counts and deviations are computed in code; the model explains.
It only ever proposes
Matches, fixes and recommendations wait for your Accept.
Labelled when it answers
Findings say whether the model or the built-in rules wrote them.
A fallback that still works
If the AI service is down, the rules-based path answers.
Metered per feature
See exactly where every token went.

What a programme team sees at a glance.

Pieces of the product teams open every day.

On any screen

CMCharu MenonDistricts behind · today×
RMRohan MehtaDonor retention · yesterday×
AIAnanya IyerFellows after two years · 2 days×

Questions asked

312

Asked today

Connected
Cleaned
Reported
Acted on

Report health

CSO and NGO SystemsCommons Intelligence

Field appOffline
Auto syncOn

Grounded in your own rows

Active in 18 states

Chapters

48,620

BaselineMidlineEndline

Children assessed

अ ಅ অ

In 11 languages

% at story41

Reading levels

Everything a reporting team runs on.

Open any one to see it in the product.

Reports, briefings, stories and clean datasets in one place.
Ask Ben
QuestionReading levels by district, last term, monthly
Assumption“At grade level” = reads a Class 2 story
Grouped byMonth of assessment
ConfidenceHigh · 48,620 rows
Connected sources
Google Sheets4
PostgreSQL2
CommCare6
CSV uploads2
Describe a chart and Ben adds it, with his assumptions shown.
This morning's briefing
!Reading levels fell in 3 districts
Attendance back above 90%
!Survivorship warning · Class 6 cohort
Who asked today
AIAnanya Iyer
RMRohan Mehta
FAFarida Ansari
DKDev Kapoor
Questions answered from your own rows, with a confidence score.
Reading levels
% at story41
At story levelBelow
Children assessed
48,620
BaselineMidlineEndline
What moved, why it matters and what to do, every morning.
Questions asked
CMCharu MenonDistricts behind · today×
RMRohan MehtaDonor retention · yesterday×
AIAnanya IyerFellows after two years · 2 days×
Report
Connected
Cleaned
Reported
Acted on
Join systems, fix messy values and rebuild automatically.
Devices
Field tablets · 42Offline
Auto syncOn
Conflict checkOn
Records queued
318
will sync on Wi-Fi
Briefings and shared reports read well on a phone.
Roles
AdminFull access
Source leadOwn data
Field staffCapture only
Sources
Active128
Pending6

Ask Ben

QuestionReading levels by district, last term, monthly
Assumption“At grade level” = reads a Class 2 story
Grouped byMonth of assessment
ConfidenceHigh · 48,620 rows

Connected sources

Google Sheets4

PostgreSQL2

CommCare6

CSV uploads2

This morning's briefing

!Reading levels fell in 3 districts
Attendance back above 90%
!Survivorship warning · Class 6 cohort

Who asked today

AIAnanya Iyer
RMRohan Mehta
FAFarida Ansari
DKDev Kapoor

Reading levels

% at story41

At story levelBelow

Children assessed

48,620

BaselineMidlineEndline

Questions asked

CMCharu MenonDistricts behind · today×
RMRohan MehtaDonor retention · yesterday×
AIAnanya IyerFellows after two years · 2 days×

Report

Connected
Cleaned
Reported
Acted on

Devices

Field tablets · 42Offline
Auto syncOn
Conflict checkOn

Records queued

318

will sync on Wi-Fi

Roles

AdminFull access
Source leadOwn data
Field staffCapture only

Sources

Active128

Pending6

From the people using it.

I used to walk into the board meeting with a number I could not defend. Now the chart opens onto the rows it was drawn from.

Ananya IyerDirector of Programmes, Saahas Foundation

My queue of ad-hoc requests emptied in a fortnight. I spend the week on the analysis that actually needs a person.

Rohan MehtaData Analyst

The survivorship warning stopped us claiming a gain that was really a change in who we were measuring.

Farida AnsariM&E Lead

Works with what you have. Leaves with you if you go.

Fifteen connectors and counting. Point it at your own warehouse, or use ours and move later. Everything exports as PDF, Excel, CSV or through a documented API.

Razorpay
WhatsApp
Mailchimp
Zoom
Salesforce

Pricing that fits before it scales.

Start free for a month. No card, no sales call, and a free training session. Pay annually and save 20%.

Individual

One person getting a proper handle on their own numbers

₹1,999/ month

1 user, 10 connected accounts
100 data sources, unlimited reports
10,000 AI tokens a month
Ben: briefings, stories and conversational reporting

[object Object]

Startup

Enterprise

Many teams, many sources, and data that has to stay where you put it

Custom

100+ users, 10,000 data sources
Self-hosted, or a dedicated database
Your own warehouse engine and Anthropic key
Named support and a migration plan

[object Object]

Still waiting on reports this year?

“Tell us how your programme runs today. We will show you the product on your own workflow, and be honest about whether it fits.”

Satyam GambhirCo-founder, commons.inc

Request a demo

Quick answers.

A reporting tool with an AI analyst built in. Connect your systems, ask questions in plain English, and let Ben send you a briefing every morning about what changed and what to do.
Pick a plan to keep going. There is no card on file during the trial, so if you do nothing, nothing is charged.
No. The model gets the shape of your tables and a small sample of rows. Counts and calculations are done in code, and credentials are never shared.
Yes. On Enterprise you can bring your own Anthropic key, run your own warehouse engine, or self-host the whole platform.
Yes; we stay through launch until your team runs it.

Built for the person who owns the number.

Founders and programme heads

The number now, and whether to believe it, before the next funder meeting.

Operations and programme teams

Five tools and a spreadsheet, joined without a data engineer.

Finance and impact reporting

The monthly pack, assembled and current, every figure traceable.

Marketing and fundraising

Channel, cohort and retention analysis without the fortnight wait.

You only spend tokens when Ben thinks.

Building charts, reading reports, cleaning data and running queries cost nothing. Tokens go only when the model is actually called, and your plan's monthly allowance comes first.

₹2,000 per million

Top up any amount; what you do not use stays on your balance.

What a token buys

A morning briefing or a full story, a few thousand. Matching the columns of two tables, a couple of thousand. A one-line question, a few hundred.

You cannot overspend

Ben stops before your allowance runs out and says so. Topping up is your decision, not a surprise on the invoice.

Or bring your own key

Put in your own Anthropic key and the intelligence runs on your contract, at your rates. Every call is still itemised.

“Reporting tells you what happened. Ben tells you what to do, and shows you why.*”

Prashant Mehra
CEO, commons.inc

“commons stayed through the messy months after launch, training our team and fixing what the field broke, until the system genuinely ran without them.*”