Ask your business dataa question.Business intelligence and AI for small and mid‑sized businesses. Each answer says why a number moved and what to do next.
Ask your dataYour data can explain your business, not just report on it.
Everything your business does leaves a record: every job, invoice, payment, call and booking. Most of it sits in separate systems and reaches you as a report, days later, showing what already happened. Then someone loses an afternoon working out why.
We join that data and put it to work. You ask a question in plain words. You get the answer, the reason behind it and the next step.
Follow one question through.
Example · A service business with three offices
A practice, an agency or a property portfolio would ask different questions. The answers arrive the same way.
1.You ask, the way you would ask a person
On screen
We did more jobs in March than February. Why did we make less money?
2.It tells you what moved
On screen
Jobs were up 9%. Gross margin fell from 41% to 35%. Nearly all of the fall is in one place: installations at your North office.
3.It tells you why
It follows the number back through the records behind it: the job tickets, the timesheets, the supplier invoices.
On screen
Two causes. Your equipment supplier raised prices 8% on February 1, and your quotes still use the old costs. And since two new installers started, North’s install jobs have run three hours over estimate on average.
Each cause opens to the jobs and invoices it came from.
4.It shows you where else to look
The same pattern is rarely in one place, and a problem in one line often sits beside an opportunity in another.
On screen
The problem: 14 open quotes worth $186,000 still carry the old costs.
The opportunity: maintenance plans are your highest-margin work, and 212 past install customers do not have one.
5.It tells you what to do next
On screen
Reprice the 14 open quotes before they are signed.
Update the cost table your quotes draw on.
Pair the two new installers with a lead for a month.
Offer the plan to the 212 customers, starting with installs more than a year old.
Not nowYes
Say yes, and the repriced quotes and the plan offer are drafted and waiting for your approval. Acting on an answer is a job for Workflow Automation.
6.You ask the next question
On screen
What happens if we raise install prices 5%?
It answers from your own history: which customers accepted which price, and which ones walked away.
AI reads your question and writes the explanation. The figures themselves are calculated by fixed rules from your own records, the same way every time, and every one opens to the records behind it.
Agents that watch your numbersAI agents watch your numbers and bring you what needs a decision.
A question only helps if you think to ask it. So the same system keeps asking for you.
An agent is software you give a goal and limits to. It uses AI to decide for itself what to check next, where to look and when to speak up. Agents work on your numbers every day, including the days you never open a report.
They compare
This week with last week. This month with the same month last year. Each office, crew, service and client with the others. Every number with the limits you set.
They find out why before they tell you
When a number moves, an agent traces it through the records, the way the example above did. On a harder question several agents work together: one pulls the records, one tests each possible cause, and one checks the finding against the source before it reaches you.
They tell the person who can act
The finding goes to whoever owns that number, as a short note where they already work: email, a text, your team’s chat. It says what moved, why, what it is worth and what to do.
They keep watching
After you act, they tell you whether it worked.
Example · Three notes from one week
Monday 6:02
Collections are down 8% on last week. Two insurer payments worth $3,680 have not cleared. Everything else is in range. Nothing to do yet. I will check again Wednesday.
Wednesday 7:15
One of your ten largest accounts has paid late three months running and has two open service issues. Its renewal is 60 days out. Suggested: a call from its account manager this week.
Friday 4:40
Tuesday’s repricing is holding. 11 of the 14 quotes have been accepted at the new price.
Have a number you want explained, or watched? Start there.
The decisions it guidesBetter decisions in every part of the business, from data you already have.
The same joined data answers different questions for sales, for operations and for finance. Here are six.
Example
Which customers to call before they leave
A customer rarely announces they are going. They pay a little slower, order or call a little less, and leave a service issue open. Each sign is small. Joined, they become a flag on the account, on the screen your team already uses, naming who should call and why. You make the call while there is still a decision to change.
Example
Which customers are ready for more
The same records show who is growing: who buys one service and fits a second, who has outgrown their plan, who has never been offered what your best customers buy. Your team gets a short list each week, with the reason beside every name.
Example
What to charge
You see the margin on each group of customers and how each group has responded to price before. A rise goes where it will hold, at a size the data supports, and you know who might leave before the letter goes out.
Example
Which work makes you money
Revenue tells you who is busy. Margin by job, service, crew, office and client tells you who is profitable. A full calendar can hide the work that loses money. This shows it, along with the work you want more of.
Example
Where the money goes
Every vendor, card and contract, grouped by what it pays for. Two tools doing one job. A contract that renewed unread. A cost climbing faster than the revenue it supports. You cut where the waste is and leave the rest alone.
Example
Where the operation slows down
Days from finished job to invoice. Days from invoice to cash. Days from inquiry to quote. Each one is broken down by step and by office, so you fix the step that is costing you and then see the number move.
In your industry
The numbers have different names in every business: net collection rate in a dental practice, the owner statement in property management, job margin in home services, one set of figures across a private-equity portfolio. See yours.
What we buildReports, dashboards, plain-English questions and AI agents, all on one set of numbers you trust.
People ask us for a dashboard. What makes a dashboard worth opening is the work underneath it, so that is where we start.
Example
ReportCollections down 8%
DashboardCollections down 8%
QuestionCollections down 8%
AgentCollections down 8%
Agent teamCollections down 8%
One set of numbers · one meaning each
CollectionsCleared in the weekBankdown 8%
BilledDated in the weekAccountingin range
Jobs doneCompleted in the weekSchedulingin range
Accounting
Scheduling
Customer records
Bank
Spreadsheets
The foundation: your systems, joined
We connect the software you run, your accounting, your scheduling or practice system, your customer records, the bank and the spreadsheets that fill the gaps, into one picture of the business. Each number gets one meaning, agreed with you and written down: what counts as collected, what counts as a finished job. From then on it is calculated the same way every time, and everyone in the business works from the same figure.
When the systems themselves need to stay in step, that is Systems Integration.
On that foundation we build what the job calls for.
Reports that arrive by themselves
The weekly numbers, the month-end pack, the statement you send your own clients. Built and delivered on schedule, with this period set beside the last and a line on what changed.
One of ours
For a manufacturer’s sales team, a summary posted to the team’s chat at the end of every business day: companies found, approved and set aside, contacts added. Nobody compiled it.
Dashboards and KPIs
The few numbers you run the business by, your key performance indicators, on one screen, with whatever needs you at the top.
Questions in plain words
Anyone you choose can ask and get an answer with its sources, without waiting for the one person who knows the spreadsheet.
One of ours
Our own company. One database is the record for our leads, pipelines, content and research, and we question it in plain words every working day.
Agents that watch
Agents that compare, investigate and report, as described above, on the numbers and limits you choose.
One of ours
Our monitoring system records what is rising and falling across the 35 sources it reads, every four hours, so we see what changed since last week and last month. It alerts us when a trend crosses a line we set.
Agent teams that find out why
For the questions that take real digging: several agents gather the evidence, weigh it and write up the finding with a recommendation.
One of ours
Our own market research. It tracks more than 1,400 investment firms and more than 3,000 of their deals, reads the news on them every day, and writes a brief on each firm worth approaching: what it sees, what that means and how to open the conversation. It has written 163 briefs so far.
We match each build to the decision it serves. A number you review every Monday gets a report that is waiting when you sit down. A number that can turn on any day gets an agent watching it. A question nobody has managed to answer gets a team of agents. Where the reporting software you already own can do the job, we set it up properly and connect it. Most businesses end up with a mix, and for each decision we tell you which build we would choose and why.
We run on it ourselvesThe data that runs our own sales also became a published market study.
Our own
Every dayWhich companies to approach, and why
Published503 problems · 2,477 contractors
One dataset · home-services companies, researched and scored
Business listings
Company websites
Contractor forums
HAIBRID runs on the kind of system this page describes. Our research on home-services companies sits in one database: it scores which companies are worth approaching, and agents report on it every day. When we wanted to understand that market, we asked the same research a bigger question and published the answer: 503 problems raised by 2,477 contractors in their own words, each matched to the automation that solves it.
The judgment behind it comes from enterprise analytics. Our founder was Vice President of Business Intelligence and Analytics at a Fortune 250 company, where he built the joined view of every client that its pricing and risk systems ran on.
Before you rely on an answerYour first build has to match the reports you trust, line by line.
We start from the decisions you and your managers make each week, not from an inventory of your data, so the first build in the Proof Sprint answers a question that matters to you.
It runs beside your current reports until every line agrees. Any difference is explained in writing, and you sign off what each number means.
Example
Current reportNew buildExplained: 2 deposits cleared Monday
14 lines · 13 match · 1 explained
You keep the record of every number: where it comes from, how it is calculated and who owns it. When the person who used to pull your numbers moves on, nothing leaves with them.
A dashboard shows you the numbers and leaves the why to you. This answers the why, suggests the next step and comes to you when something moves. We keep the dashboards your team uses and build on them.
Our data is messy. Can you still work with it?
Yes. Most business data is. Joining it and settling what each number means is the first part of the work and the source of much of the value. We start with the numbers behind one decision and build out from there.
How do I know an answer is right?
Every figure is calculated by fixed rules from your own records, the same way every time, and opens to the records behind it. AI reads the question and writes the explanation. Before you rely on it, the build runs beside your current reports until the two match.
Which systems can it read?
The ones you run: accounting, scheduling, practice and job software, customer records, bank feeds and spreadsheets. We read from them, and your team keeps working in them. Where a system has no data connection, an agent reads it through its screens.
Who can see what?
You decide. An office manager sees their office and an owner sees everything. Every answer and every note from an agent follows the same permissions.
Can we start with one question?
Yes. Pick the decision that matters most this quarter. Everything built for it, the joined data and the agreed meanings, is reused by the next one.
Bring the question your numbers can’t answer yet.
Start with the free Operations X-ray, or book a call and bring the question with you.
Eric Lopez30 min