Every year, lists of twenty BI trends make the rounds and repeat the same thing: AI, AI, and more AI. If you run an SMB, the useful question is a different one: which ones can you take advantage of this year with the data you already have? Several of them already work, but all of them rely on well-defined KPIs and a well-organized ERP.
In this guide you'll see what has changed in BI, the five business intelligence trends that matter in 2026, their maturity and requirements in a table, and a plan to get started without a data team. For a from-scratch definition of BI, see the guide to Business Intelligence for SMBs.
What business intelligence trends are and what has changed in two years
Business intelligence trends are the shifts in how companies query, monitor, and use their data to make decisions. In 2026 they all point in the same direction: less time building reports and more time taking action, with AI that answers questions, alerts that notify you on their own, and analytics inside the system where the work gets done.
Three facts explain it:
- Generative AI is replacing classic queries. In February 2027, Microsoft will retire Q&A, Power BI's natural language question feature, and recommends Copilot instead (Microsoft Learn, checked in October 2026).
- ERPs are opening up to agents. According to its release notes, Odoo 20 lets you connect the database via MCP and call AI agents from automated and scheduled actions.
- Semantics has moved to the center. Gartner grouped its data and analytics trends for 2026 into agentic data and analytics (D&A), semantics at the center, and platformization.
1. Natural language queries: asking your data questions with AI
Natural language queries let you type "How much did we invoice this quarter compared with last quarter?" and get the figure or the chart without building filters or formulas. They're the most visible face of BI with generative AI.
- Copilot in Power BI answers questions, summarizes reports, helps create them, and writes DAX queries. According to Microsoft, it requires a paid Fabric capacity (F2 or higher) or Power BI Premium (P1 or higher), since a Pro license isn't enough, and in its standalone experience, use in languages other than English is not yet officially supported (checked in October 2026).
- Odoo 20 includes an AI assistant that answers questions about your finances using only the accounting data in your database; all of its AI consumes IAP credits that are purchased separately.
Microsoft warns that, without preparing the semantic model, Copilot can misinterpret the data and give generic or inaccurate answers. AI doesn't fix a confusing model; it exposes it.
2. From dashboard to alert: data that notifies you on its own
Proactive alerts flip the logic of the dashboard: instead of waiting for someone to open it, the system checks the metric and notifies the owner when it crosses a threshold, with the data and the link to take action.
It's the most mature trend and the one that depends least on AI; the hard part is defining the threshold, who to notify, and what to do. Examples: "Sale with a margin below X%" or "Products projected to run out of stock." The step-by-step is in the guide to automatic threshold alerts.
3. BI embedded in the ERP instead of a separate project
BI embedded in the ERP is analytics that lives inside the system where sales, stock, and accounting are recorded. The dashboard reads the data at the source, and from each chart you can drill down to the transaction.
Compared with a separate project (extraction, an intermediate data warehouse, and its own licenses), there are fewer pieces to maintain, the numbers match the ERP, and role-based permissions already exist. That way, a restaurant group can compare food cost and waste by location without exporting data, as long as it sets up every location the same way.
Its limit: if much of your information lives outside the ERP (online store, production, spreadsheets), you need a layer that consolidates multiple sources. And in Odoo Enterprise, connecting an external tool through the API requires the Custom plan (price list checked in October 2026).
4. AI agents that analyze, explain changes, and propose actions
An AI agent for analytics doesn't stop at answering: it queries several data sources, explains why a metric changed, and prepares the next action, such as a task or a draft email, for a person to approve.
Gartner calls this agentic D&A. One piece that makes it easier is MCP, the open protocol that connects an agent to your systems: in Odoo 20, the MCP server exposes query tools by default, such as Search and Read group (grouped data), and hides the rest until someone enables them. The precautions it requires are covered in the guide to MCP.
Maturity is early: these are recent features, part of this doesn't exist in earlier versions (see how to upgrade your Odoo version), and on Odoo's price list agentic AI appears in the Custom plan. For a pilot: minimum permissions and read-only access.
5. Data quality and governed KPIs: the requirement nobody sees
Governed KPIs are metrics with a single agreed-upon definition, an owner, and a source: "net sales" means the same thing on the executive dashboard, on the sales dashboard, and in the AI's answer.
The semantic layer stores those definitions (what the margin includes, what counts as an active customer); in Power BI it's called the semantic model. With AI, it has gone from best practice to essential, because an assistant that doesn't know what "margin" means can give an answer that sounds plausible but is wrong.
The other half is data quality: duplicate contacts, opportunities with no expected revenue, or products with no cost distort any dashboard, and an AI answer even more so. How to avoid them is covered in the guide to customer history.
Which trends apply to an SMB today
| Trend | Maturity (October 2026) | What you need first | Does it apply to an SMB today? |
|---|---|---|---|
| Natural language queries | Available; cost and Spanish support vary | Defined KPIs and clear names | Yes, for narrowly scoped questions |
| Proactive alerts | Mature, with no dependence on AI | Threshold, owner, and action | Yes, it's the easiest |
| BI embedded in the ERP | Mature | Operations recorded in the ERP | Yes, if you run on the ERP |
| AI agents that analyze | Early | Minimum permissions and human review | As a read-only pilot |
| Governed KPIs | Mature practice, now essential | A definition and an owner per KPI | Yes, before anything else |
| Predictive analytics (add-on) | Mature in specific use cases | History with no gaps or duplicates | Yes, for purchasing or cash flow |
Predictive analytics for SMBs: where it really pays off
Predictive analytics uses historical data to estimate demand, stockouts, cash flow, or customers at risk of churning. It's already reaching the ERP: Odoo 20 suggests the minimums and maximums for replenishment rules based on demand history. It pays off in repetitive decisions, such as how much to buy, as long as the history is clean.
How to get started without a data team
- Pick three decisions, not three charts: how much to buy, who to collect from, which customers are going cold.
- Write down the definition of each KPI: formula, source, and owner.
- Clean up the data you use most: duplicates, contacts without a RUT (Chilean tax ID), and products without a cost.
- Put the dashboard where the work already happens: if your operations live in the ERP, so should your analytics.
- Add alerts for the exceptions and save the dashboard for the weekly review.
- Test AI with questions you already know how to answer and compare the results with your report.
- Then, run a pilot with an agent in read-only mode and with human review.
Why do it with KPB Solutions
We have a piece for every step:
- Keep BI, BI embedded in Odoo: a catalog of 31 prebuilt dashboards with direct access to the transaction. Plans with 3, 5, or 10 dashboards (10, 15, and 25 UF + VAT per month; the UF is Chile's inflation-indexed unit of account), setup in 7 days, and a Predictive Analytics add-on for 8 UF + VAT per month, with differentiated delivery times.
- Keep Checking, alerts in Odoo: plans with up to 10, 30, 50, or 100 alerts (from 3 to 10 UF + VAT per month), with setup in 7 days. You choose them in the quote builder and request your quote.
- Podiux, BI across multiple sources: dashboards, KPIs, automated reporting, alerts, and AI assistants, built on Power BI and connected to Odoo, other ERPs, online stores, or spreadsheets.
- Kersify, AI agents: custom assistants and agents for the pilot in step 7, read-only on top of your ERP or CRM (Odoo or another) and with a person who approves every action.
Behind it all is a team of more than 40 KPBeers and the experience of an Odoo Gold Partner with more than 70 active clients.
Want to know which trends will work for your company this year? Book a meeting and we'll review your data and where to start.
