Ontelis helps analytics and insights teams gather and check the data behind business reviews, reports and forecasts, with each figure sourced and dated, so analysts can focus on what it means.
Trusted by 2 out of the 5 largest life sciences companies
Agents do the assembly. Analysts own the definitions.
Every figure sourced and dated. Each figure in an answer carries its source and as-of date, and is checked against the agent’s sources before the answer is shown.
Start from checked inputs. Agents read sales, claims, access, call and promotion data in place, check it, and match each HCP across sources. Doubtful matches go to your data steward.
Your definitions, your forecast. Analysts own the definitions, the forecast and its assumptions. When a definition a rule depends on is superseded, the rule goes back to suggestion only.
From the question to the decision
Agents gather and check the data for each question and draft findings with the method written down. Analysts review the work and present it.
Agents gather inputs and commentary, build the review pack and pre-read, then draft the decision log and follow up each action with its owner, escalating when one stalls.
Agents check data and test reports before release, check each refresh, and find which reports still show old numbers. Ontelis does not build BI dashboards beyond its own process reports.
Agents gather and refresh forecast inputs, document each assumption, and track actuals against the forecast and its assumptions. The forecast owner reviews and approves every cycle.
Agents gather coverage, formulary, denial, time-to-therapy and abandonment data; patient-support figures arrive only as approved aggregates. Each alert opens a follow-up run with a named owner.
Agents match spend, delivery and outcome data and run the test and control reads approved in Govern. When an effect isn’t measurable at your volume, the result says so.
Agents gather competitor signals and congress readouts with sources, prepare study records for each research vendor, and draft stakeholder briefs. Any counter-material goes to promotional review.
Watch Ontelis at work
Short walkthroughs of agents doing commercial work, from the first draft to the person who signs off.
Microsoft · Dynamics 365, Azure, Microsoft 365 (Outlook, Excel, Teams, SharePoint)
Tableau
Power BI
Workday
Salesforce · Life Sciences Cloud, Marketing Cloud
Komodo Health
Databricks
Definitive Healthcare
HealthVerity · Symphony Health
Adobe · Experience Cloud
Model N
AWS
SAP · S/4HANA, Ariba
Pitcher
H1
Informatica
CMS · NPPES, Open Payments
Merative · MarketScan
MMIT
Clarivate
Evaluate
Citeline
AlphaSense
Aprimo
Marketo
Doximity
Medscape
DeepIntent
Cvent
Medispend
IntegriChain
CoverMyMeds
Surescripts
AssistRx
CareMetx
TraceLink
Qlik
SAS
dbt
MuleSoft
Fivetran
Coupa
Oracle · ERP Cloud
Oracle NetSuite
Varicent
Xactly
Okta
Slack
Zoom
Box
Jira
Definitions, the forecast and each action stay human
People decide
The head of commercial analytics decides definitions and split rules, in IntuitionDB.
The forecast owner approves the forecast and its assumptions.
The data steward rules on low-confidence doctor merges, in Veeva Network.
The decider named on each card decides that action and any write, in Workspace.
MLR reviewers approve any counter-material from a competitive brief, in Vault.
Agents never
Present a simulation as a measured result.
Copy claim lines. Copying them would build a second warehouse of patient health information.
Bring patient-level data into analytics. Flows between partitions are one-way and aggregate-only.
Use prescribing-derived inputs or outcomes, including proxies such as decile, in a decision that gives a healthcare professional (HCP) something of value.
“We had more data about our physicians than ever, and every question about it still started with a ticket. By the time the answer came back, the plan had moved on.”
Director, Omnichannel EngagementOne of the top five US biopharma companies
Enterprise-Grade Security is the foundation
Ontelis meets the security and compliance standards life sciences companies hold their vendors to. Every workspace includes the controls enterprise teams expect: SAML SSO, role-based access, audit logs for every agent action, approval gates, data retention policies and more.
Your data never trains modelsCustomer data, prompts and outputs are never used to train Ontelis or third-party models.
SOC 2 Type IIIndependently audited controls for security, availability and confidentiality.
ISO/IEC 27701Certified privacy information management across the platform.
HIPAASafeguards for protected health information in every workspace.
GDPRPersonal data processed in line with GDPR, wherever your teams work.
Questions analytics teams ask
Ontelis agents do the assembly work behind your analytics. They gather and check the data for a business question, build the review pack, check each report refresh and keep forecast inputs current. They read prescription and claims totals in place in Snowflake, along with your coverage feed and Veeva CRM, and every figure carries its source and as-of date. Your analysts review the findings and present them. Your head of commercial analytics sets the definitions, and the forecast owner approves every cycle.
No. Ontelis builds no BI dashboards beyond reports on its own processes, and Tableau or Power BI keeps your scorecards. Ontelis agents work on what feeds them. They check the data and test a report before release, then check each refresh. When a vendor restates a feed, they find which reports still show the old numbers and tell the owners.
Keep Copilot as the place your leaders ask, and put Ontelis behind it. Copilot passes the question to Ontelis through the Agent Gateway. Ontelis answers on the definitions your team agreed, with each figure's source and as-of date, and flags any data past its freshness target. It also checks the asker's access in the source system, so a district manager asking the same question as a VP gets a narrower answer. Ontelis governs only what passes through it.
Your head of commercial analytics. Ontelis uses one definition of a metric such as NBRx (new-to-brand prescriptions) in every answer, the one that person chose, and keeps older versions with their dates. A split that rests on a rule your team agreed, not on a measurement, is labelled "ruled, not measured" and carries a review date. If a definition the rule depends on is replaced, or territories are realigned, the rule goes back to being a suggestion only.
Each figure in an Ontelis answer carries its source and as-of date, and is checked against the data the agent read before anyone sees it. A figure that comes from a rule and not from a measurement says so, and agents never present a simulation as a measured result. Ontelis keeps every answer with a snapshot of the data behind it. After a vendor restates a feed, you can still see what last week's answer was built on.
No. Ontelis reads NBRx and claims totals in Snowflake, in place, and claim lines and other patient-level rows stay in the source. Agents never copy them, because that would build a second warehouse of patient health information. What Ontelis does keep, such as identifiers, relationships and permitted aggregates, sits in a single-tenant data plane that you place. Patient-support figures reach analytics only as approved aggregates, with counts of 1 to 10 shown as "<11".
Usually not, and Ontelis says so. A call plan's effect on new prescriptions takes months to show, and other things move in the same months. Reading it takes a territory-level test, and the result is often "not measurable at this volume". Ontelis reports that like any other result. Faster effects are different: what an email cadence does to unsubscribes can be tested practice by practice. Agents run only the test and control reads approved in Govern.