Skip to main content
Ontelis home
See a demo
Ontelis home

How your team decides becomes how your agents work.

IntuitionDB records the decisions people make at process steps, tests what other choices would have produced, and turns options that beat the benchmark into top rules.

Decided by Brand manager (role)Override of the top suggestion
Step 3, select speaker: Dr. A
  • Speaker program, US · run SPK-0217
  • Tavrenta, Q1, 12 Northeast dinner programs
Options shown, in order
Dr. BTop suggestion, R-044 v3
Dr. AChoiceRanked second
Dr. CRanked third
Reasonaudience fitNote: “community neurologists turn out for Dr. A”
Outcomes linked later
AttendanceBy HCPs who treat the condition, against plan
Evaluation scoresAfter the program night
Context package: signed snapshot and graph version, kept with the recordRule in force: R-044 v3
EvidenceOutcomesLedger
Evidence summary · rule R-044 v3Step 3, select speaker. Ranks eligible speakers by attendance against plan, evaluation scores and cost per attendee.
FormReadable ranking function
Evidence levelLevel 2Logged comparison
Comparable cases31 comparable programs
RankingDr. B, then Dr. A, then Dr. C
AdjustmentsKnown factors and display order
CorrectionApplied for the comparisons tried
Outcome windowAttendance on the program night; evaluation scores after it
InputsDeclared inputs passed Govern’s lineage check
Not measuredNot evaluatedPrescription effects (metric firewall)

Trusted by 2 out of the 5 largest life sciences companies

How IntuitionDB learns

  1. 01Capture every judgment, and its reason, at each step of a process.
  2. 02Evaluate what a different choice would have produced, and report it.
  3. 03When similar work comes up, rank suggestions.
  4. 04A human from your team chooses.
  5. 05An option that beats the benchmark is recorded as the top rule.
  6. 06Repeat.

Captured where the decision is made

  • One decision point per step; each choice saves the signed context package shown.

What the other choice would have produced

The result of the option not taken can’t be observed, so every estimate states which of three evidence levels it rests on.

  1. Re-runs the automatic downstream steps on the graph as it was at decision time. Exact for system effects; says nothing about how HCPs (healthcare professionals) or staff would respond.

What one company can measure

Small improvements take more recorded work to prove than big ones, so IntuitionDB is strongest on fast, high-volume operational outcomes and weakest on slow market response. Records pool across everyone doing the same step within your company, never across customers. When there isn’t enough yet, the answer is “not measurable at this volume”.

By decision type, from most to least measurable

Slower outcomes and smaller effects need more recorded work.
Decision typeMEASURABLE?Outcome arrivesBest evidence
Mechanical system effectSegment rule A vs BYes, exactlyAt onceReplay
Operational routingMLR review track, invoice toleranceYes, for medium and large effectsDays to weeksLogged comparison
Transfer of valueSpeaker, venue, formatYes, on allowed outcomes onlyDaysLogged comparison
Content vs reviewClaim wording vs first-pass approval10-point effects yes; 2-point effects noWeeksLogged comparison or test
HCP engagementEmail cadence vs unsubscribesEngagement yes; prescriptions usually noAbout 2 weeksTest by practice
HCP targetingCall plan vs new prescriptionsOften “not measurable at this volume”Months, confoundedTerritory-level test only
Expert judgment, no outcome“Needs medical review?”Only as agreementNeverExpert agreement
Decision typeMeasurable?Outcome arrivesBest evidence
Mechanical system effectSegment rule A vs BYes, exactlyAt onceReplay
Operational routingMLR review track, invoice toleranceYes, for medium and large effectsDays to weeksLogged comparison
Transfer of valueSpeaker, venue, formatYes, on allowed outcomes onlyDaysLogged comparison
Content vs reviewClaim wording vs first-pass approval10-point effects yes; 2-point effects noWeeksLogged comparison or test
HCP engagementEmail cadence vs unsubscribesEngagement yes; prescriptions usually noAbout 2 weeksTest by practice
HCP targetingCall plan vs new prescriptionsOften “not measurable at this volume”Months, confoundedTerritory-level test only
Expert judgment, no outcome“Needs medical review?”Only as agreementNeverExpert agreement

From suggestion to top rule

A candidate becomes the top rule only if it reliably beats the benchmark, doesn’t worsen guardrail metrics, holds up in a later period and clears a bar that rises with each candidate tried, since more tries mean more false positives. Ontelis sets a minimum bar no owner can lower, versioned in Evaluations.

Privacy for the people deciding

  • By default, records show the role, not the person’s name. No pattern is reported for a group smaller than five people.

“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

Who decides

IntuitionDB does

  • Records each decision and each undeclared judgment.
  • Reports other choices with their evidence level.
  • Ranks suggestions by the low end of each range.
  • Marks a winner as the top rule; the owner can revert.

People decide

  • The person at the step chooses, every time, unless the rule owner has approved auto-apply.
  • The process owner sets the metric and, with compliance, the guardrails.
  • The rule owner approves auto-apply, with compliance on controlled steps.
  • MLR approval, attestation and payment release stay Human only.

What IntuitionDB does not do

  • Train model weights.
  • Pool records across customers.
  • Rank or score employees.
  • Randomize regulated steps.
  • Present a simulation as a measured result.
  • Record screens or keystrokes.

Where it fits

  1. 01TriggerAgents
  2. 02 · This pageContext packageContext Graph, with IntuitionDB and Mask
  3. 03Agent draftAgents
  4. 04Govern checksGovern
  5. 05Decision cardWorkspace
  6. 06Write through ConnectConnect
  7. 07LedgerGovern
  8. 08 · This pageOutcome to IntuitionDBIntuitionDB

Watch Ontelis at work

Short walkthroughs of agents doing commercial work, from the first draft to the person who signs off.

Trace a share loss to access or field

Analytics & Insights · Commercial analytics agent · 4:06

0:00 / 4:06

Customer stories

One of the top five US biopharma companies

Brand teams get HCP answers in under 30 seconds

Read the story
One of the top 10 global pharma companies

Email campaigns cut from 131 effort-days to 46

Read the story

Questions about decision intelligence

Ontelis IntuitionDB is the decision intelligence product in Ontelis: it records each decision your people make at a step in a process, such as which speaker a brand manager picks for a dinner program, along with the options they saw, the reason they gave and what happened afterwards. It then estimates how the other options would have done and tells you how strong that evidence is. An option that keeps beating your benchmark ranks first in the suggestions the next person sees.

Security and data

Data in IntuitionDB
Decision records sit in an append-only store in the customer’s data plane, with its own access policy that the customer controls.
Vendor documents
Ontelis provides: a SOC 2 Type II report, an ISO 27001 certificate, a HIPAA business associate agreement wherever Ontelis processes PHI, a data processing agreement with transfer clauses, and a sub-processor list.
Models
Ontelis builds no foundation model. No model is trained on customer data, except an opt-in adapter that never leaves the customer’s data plane.
Your controls
Ontelis’s posture as a vendor is kept separate from Govern, which holds your own controls.

How your team decides becomes how your agents work.

See a demo