Data Semantics builds AI agents around the decisions that move your business. They spot what needs attention, work out why, decide within your rules, act in your ERP and CRM, and measure whether it worked. You choose which decisions run on their own.
Signals from
ERPCRMCore bankingData platformTeams & emailDecide and act
You set each automation level
Runs on
Fabric IQ & CopilotDatabricks GenieSnowflake CortexInsideOpsData Semantics improves specific business decisions with data, AI and clear rules. We deliver decision intelligence with our own method and industry decision catalogs, natively on Microsoft Fabric IQ and Copilot, Databricks Genie or Snowflake Cortex, or with InsideOps, our own engine. AI agents run on your ontology and context graph and write actions back to your ERP and CRM. You decide how far each decision is automated.
Organizations have hundreds of reports, and AI now builds more in minutes. Yet blocked payments, overdue receivables and stock problems still surface at month-end, someone still reads the report, decides alone and rekeys the fix into the ERP, and nobody checks whether it worked.
Dashboards stop at the first step and leave the rest to people. Our agents carry the decision through to action in your systems, and measure the result.
Each outcome improves the next recommendation. That's what turns analytics into intelligence.
Need trusted reports and dashboards first?Governed KPIs and performance reporting have their own service.Business Intelligence & Analytics →Every decision is set to the level it's ready for, and can move up as trust grows.
AI raises the signal and explains the cause. Your people decide and act. Right for new, high-stakes or judgement-heavy decisions.
Signals and root causeAI proposes options with reasons and checks them against policy. A person approves with one click. Right for frequent decisions with clear rules.
RecommendationsAn agent acts within guardrails you set and escalates exceptions to a person. Right for high-volume, low-risk, well-understood decisions.
Action stepThe Data Semantics decision intelligence method turns data into decisions that get made. Seven stages take each decision from discovery to measured outcome, using the same canvases, interviews and catalogs on every engagement, whichever platform it runs on.
Bring one decision your team makes every week. We’ll show how it runs on your data, within your rules.
Which outlets will miss target this month, and what to do about it.
Which customers to contact first, how and by whom.
The ontology tells AI what your data means. The context graph tells it how your business decides: the policies, approvals, precedents and judgement your best people apply. Together they let an agent recommend, or act, the way your organization would.
Credit limits, discount bands and approval rules are part of the model.
How similar cases were handled informs the next recommendation.
Each catalog maps 30 to 50 recurring decisions, with the data, owners and policies each needs. A few examples:
No large analytics program up front. Start with one function, prove one decision, then add more on the same foundation.
We map the decisions in one function using our catalog and human decision interviews, then prove the best candidate on real data.
We add decisions one by one, move each up the automation levels as trust grows, and run and improve the whole loop for you.
Instalment and service-charge receivables scored, prioritized and worked by AI agents, with disputes and key accounts routed to the right people.
10+ leasing and portfolio analytics programs on Yardi and ERP data, giving leasing, finance and leadership the same view of occupancy, renewals and tenant performance.
Whichever route you choose, we use the same method: find the decisions that matter, connect them to trusted data and rules, act within policy and measure the outcome.
Native on Fabric IQ, Copilot, Power BI and Teams.
Native on Genie and Unity Catalog.
Native on Snowflake Cortex and Horizon.
Our own engine, for teams without these platforms or who don’t want extra AI assistant licences.
About decision intelligence
Business intelligence produces reports and dashboards that people read and interpret. Decision intelligence starts from the decision itself, such as which customers to call or which stock to move, and builds the signals, recommendations, actions and outcome tracking around it.
Decision intelligence decides what should happen: it detects what needs attention, explains why and recommends or takes the decision within your rules. Agentic AI carries out work across business systems, such as resolving an invoice or following up a payment. Data Semantics often combines them: decision intelligence picks the decision, and agents like Serina and Zebo act on it.
Only as far as you choose. Each decision is set to one of three levels: support, where AI raises signals; recommend, where AI proposes options with reasons; or automate, where an agent acts within guardrails. Business-critical decisions stay with accountable people, and every action is logged.
Yes, because you set how far each decision is automated. High-value or regulated decisions stay at support or recommend, so a person decides with the evidence in front of them. Every signal, recommendation, approval and action is logged with the records and rules behind it, so audit can trace any decision.
How it works
To recommend or act safely, AI has to know what the data means and how your business decides. The ontology gives it the meaning; the context graph gives it the policies, approvals and precedents your best people apply.
Frequent, rule-based decisions with clear data and measurable outcomes, such as which overdue accounts to call first, where to move stock or which invoices to release. Data Semantics ranks candidate decisions by value and readiness in a Decision Discovery Workshop and starts with the one that pays back fastest.
Platforms
No. Data Semantics builds decision intelligence natively on Microsoft Fabric IQ and Copilot, Databricks Genie or Snowflake Cortex if you run them, or with InsideOps, its own engine, on a single database or directly on your source systems, with no extra AI assistant licences.
InsideOps is Data Semantics’ decision intelligence product. It runs decision catalogs, alerts and AI agents within your rules: built into Serina and Zebo, or on a single database or directly on your source systems for teams that don’t use Fabric IQ, Copilot, Databricks Genie or Snowflake Cortex.
Getting started
A bounded pilot typically takes 6 to 8 weeks: one function, a few prioritized decisions and agreed measures of success, run on your own data and systems.
Data Semantics starts with a fixed-price Decision Discovery Workshop and pilot. After that, decisions are added and improved through Decision Intelligence as a Service, an ongoing engagement that runs and refines them.
Bring us one function. We'll map its decisions, find the one worth the most and show you what it looks like automated, on your own data.
