A weather system over the Pacific, photographed from orbit

Credence · Continuous actuarial intelligence

Risk moves continuously.
Now so do you.

The AI-native platform for reserving, pricing, and risk control — actuarial answers on demand, with every step of the mathematics shown.

Built for the profession where “show your work” is the law.

  • Glass-box outputsevery figure traceable to method, assumption, and data lineage
  • Reproducible runssame inputs, same answer — years later, in front of a regulator
  • Actuary holds the penAI drafts and evidences; a credentialed human signs

The story

Insurance was the first data science.

Three hundred years before anyone said “machine learning,” actuaries were building predictive models — pricing the unknowable, putting numbers on the future, and standing behind those numbers with their names. It is the most mathematically serious profession in business.

And it runs on the least serious workflow in business. Ask any actuary how a reserving cycle is actually spent: weeks assembling and reconciling data, days coaxing it through software designed when “the cloud” meant weather. The judgment — the part only they can do — gets whatever hours are left before the deadline.

Then the answer ships as a PDF, and the world it described has already moved.

The constraint was never actuarial talent.
It was the interface between the actuary and the answer.
Two analysts working through model documentation by hand
Fig. 01 — The ratio is backwards: most of the cycle goes to moving numbers, a fraction to understanding them.

The world outside the model

Risk stopped keeping a calendar.

Cyber events don’t wait for the quarter to close. Climate volatility doesn’t respect the annual pricing study. Social inflation doesn’t pause while the triangles refresh. The industry that invented modeling the future is reporting on it in arrears.

Lightning splitting a night sky
01Peril moved to the tail

Cat, cyber, and casualty shocks arrive between reviews — exactly where cycle-based reserving is blind.

Aerial view of a container terminal
02Exposure compounds daily

Books grow, mix shifts, treaties attach. The portfolio you priced is never the portfolio you hold.

Wind turbines on a ridge at dusk
03New risks, thin history

Renewables, AI liability, parametrics — the future keeps writing lines your triangles have never seen.

The platform

A colleague, not a console.

You ask the question the way you’d ask the sharpest actuary you know. Credence agents assemble the data, run the methods, stress the assumptions — and return the answer with every step shown.

  • 01
    Ask in plain language

    Reserving, pricing, and monitoring queries — no batch jobs, no export-import ballet.

  • 02
    Agents do the janitorial work

    Data assembly, reconciliation, method runs, and sensitivity sweeps happen in minutes, not weeks.

  • 03
    Every answer shows its work

    Methods, factors, assumptions, and lineage attached to every figure. A glass box, never a black one.

credence · reserving workspacelive
Chief Actuary · 06:12

agent plan · run #8841

  • ▸ assemble paid + incurred triangles, AY 2016–20252.1 s
  • ▸ chain-ladder & Bornhuetter-Ferguson, tail fitted 84-ult4.7 s
  • ▸ severity trend +1 pt stress · 2 340 bootstrap sims11.9 s
  • ▸ credibility-weight vs. industry development, Z = 0.631.4 s
Credence · 06:13 · every figure traceable

2023 liability is developing 2.8 points above the booked pick, concentrated in latest two diagonals.

held41.2 M
indicated44.1 M
range 25–7542.9 – 45.6
show the work — methods · assumptions · lineage
  • data — claims warehouse snapshot 2026-07-04T23:59Z · 148 112 records · checksum 9f2c
  • methods — volume-weighted CL (5-yr) · B-F with a-priori 64.1% · GLM severity trend
  • assumptions — trend 4.2% · tail factor 1.018 (curve-fit, Bondy alt. shown)
  • review — flagged for sign-off · S. Okafor, FCAS

The method

One promise, three disciplines.

I

Continuous, not quarterly

Reserving, pricing, and monitoring that refresh as data lands — not when the calendar says so. The quarter becomes a checkpoint, not a discovery.

  • always-on development monitoring
  • assumption-drift alerts
  • on-demand reserve ranges
II

A colleague, not a console

Plain-language questions in, actuarial-grade answers out. Agents do the data janitorial work and draft the memo; your team does the thinking.

  • conversational workspace
  • agentic data preparation
  • first-draft regulatory exhibits
III

Glass box by design

Every number traceable to a method, an assumption, and a data lineage — built for the profession where defensibility is not optional.

  • full audit trail on every run
  • reproducible to the checksum
  • regulator-ready documentation
City lights across a continent, seen from orbit at night

02:47 local time

Your book doesn’t sleep.
Neither does your actuarial platform.

Why “Credence”

Named for the oldest idea
in modern actuarial science.

Pc=Z·X+(1−Z)·μ

Z — credibility factor · X — your own experience · μ — the collective mean

Credibility theory is the century-old actuarial formula for blending your own experience with the weight of the world’s evidence. That is not just our name — it is our operating principle.

The machine brings scale, memory, and speed. The actuary brings judgment, context, and accountability. Credence is the weighting between them — and the actuary always holds the pen.

A coastal city at dusk, seen from above

Request a demo

See your book the way risk sees it.

A 30-minute working session on your lines of business — reserving, pricing, or risk monitoring. Bring a hard question.

hello@credence.example early-access cohort · Q3 2026 · eight carriers