Retirement income

From a client's paperwork to an annuity quote you can stand behind

Pension statements, health questionnaires and GP letters arrive as scans and PDFs. An AI client reads them, Ptally prices the case on your basis, and every field shows the page it came from. Every quote is stored, explained and replayable.

Talk to our team
Claude over MCP
Documents in. A quote you can trace out.
The adviser or intermediary drops the client pack into the chat. Claude fills in the model's inputs and shows where each came from. Ptally prices the case on your basis, flags what needs an underwriter and saves the quote.

Synthetic case. Illustrative figures from a demo basis.

What it takes

Accurate, defensible, auditable

Your actuaries build the basis. Ptally takes on the reading and rekeying around it, so every input is one you can trust.

Accurate
The evidence isn't fields
Health answers, prescriptions and pension statements arrive as scans and photos. Each value is read with a link to the page it came from.
Defensible
The price is only as good as what was declared
A medicine with no matching condition, or dates that disagree, should be caught before the price. Not after completion.
Auditable
Every quote has to be provable
A quote is held for a set period, then rechecked at completion. You need to show which data and which basis each price used.
At volume
Thousands of cases, each read properly
Every case is worth reading carefully. None is worth an afternoon of typing.
How it works

Read, check, price, bind

Your actuaries own the basis. We build it in Ptally from your rate tables and rules, and your team uses it from the tools they already have.

01 · AI client
Read
Claude or ChatGPT reads the documents and maps what it finds to your basis inputs. Each value links to its source page, and gaps are flagged.
02 · Rules in your model
Check
Medicines are checked against conditions, ages against dates, and the purchase price against your limits. Conflicts go to an underwriter instead of pricing silently.
03 · Your basis
Price
Standard and enhanced income side by side, level or escalating, with or without a guarantee. The same deterministic engine as every Ptally product.
04 · Audit trail
Bind and record
The priced data is hashed and travels with the application pack for signature. Any change after quoting means a new quote.
Every step recorded
A full history for every case
Every field shows the document and page it came from. Every quote shows the basis version it used. A change to the data means a requote, and the hash at signature proves what was priced.

Synthetic case. Illustrative figures from a demo basis.

Checks and balances

Every check, on every quote

Before a price is issued, each case runs the same checks. Every result is stored with the quote, readable by your team and by the AI client.

PTQ-DEMO-2411724 checks✓23 passed!1 warning0 declines
Case setup3 of 3
  • ✓Consent to use health dataRecorded before upload
  • ✓Adviser identifiedDemo Financial Planning Ltd
  • ✓Product options chosenJoint, two thirds, level, 5 years
Documents4 of 4
  • ✓Documents classified5 of 5
  • ✓Fields linked to a source page41 of 41
  • ✓Required answers present0 missing
  • ✓Dates of birth agree across documents2 lives, 0 conflicts
Health3 of 4 · 1 warning
  • !Medicines match declared conditions1 of 3 unmatched: atorvastatin (spouse)
  • ✓Prescription in dateIssued 21 Oct 2026
  • ✓BMI derived from height and weight30.9
  • ✓Smoking answers consistentStopped 2015
Eligibility4 of 4
  • ✓Purchase price at least £10,000£184,912.50
  • ✓Applicant age within limits67
  • ✓UK residentUK postcode and address
  • ✓Spouse details complete for joint lifeName, date of birth, NI number
Funds5 of 5
  • ✓Fund values match statements£246,550 across 2
  • ✓Tax free cash within limit£61,637.50, 25%
  • ✓Safeguarded benefits over £30,000 have adviceNone held
  • ✓Protected tax free cashNone
  • ✓Pension sharing orderNone
Price and bind4 of 4
  • ✓Priced on the basis in forcedemo-basis-2026-11-03
  • ✓Referrals routed to an underwriter1 routed
  • ✓Quote guarantee date setUntil 24 Nov 2026
  • ✓Data hash ready for signatureSHA-256 of this case
Replay to the penny
Rerun any saved quote on the basis version it used. The same inputs give the same answer, every time.
Stored with the quote
Every check, its result and the evidence behind it are saved against the quote reference, in plain terms.
Ask why in plain English
An underwriter or auditor can ask the AI client why a price is what it is. It answers from the stored checks and price.
Human in the loop

Pause for an underwriter where you choose

Refer rules mark the cases a person must see. A referred case is still priced and saved, but it waits for an underwriter before it can be bound.

Your rules
You decide what needs a person
Your pricing team writes the refer rules into the basis, with their thresholds. In the demo basis, a medicine with no declared condition, a rating over 6 years or a very large purchase price.
Priced, then paused
The underwriter starts from a number
A referred case is priced on your basis and saved. The response lists each rule that fired and the values it used, so the underwriter sees why at once.
Recorded
Every decision has a name on it
The underwriter accepts the quote, changes the price or marks it not valid. Each change is logged with who made it, when and why. The engine's price is kept.
Optional
Tighten or relax as you learn
Refer rules are part of the versioned basis. Add, tighten or remove one in a draft, run the tests, then approve and publish it like any other change.
Refer step in the demo basis
{
  "id": "annuity_referrals",
  "key": "annuity_referrals",
  "step": "refer",
  "name": "Refer rules",
  "def": false,
  "description": "Cases priced but sent to an underwriter. A medicine with no declared condition, a rating over 6 years, or a purchase price over 1,000,000.",
  "items": [
    {
      "id": "unmatched_medicine",
      "name": "Medicine with no declared condition",
      "exp": "{{medicines_unmatched}} === true",
      "def": false
    },
    {
      "id": "high_rating",
      "name": "Health rating over 6 years",
      "exp": "{{rating_years}} > 6",
      "def": false
    },
    {
      "id": "large_purchase",
      "name": "Purchase price over 1,000,000",
      "exp": "{{purchase_price}} > 1000000",
      "def": false
    }
  ]
}
Response for the demo case, trimmed
{
  "result": 14876.14,
  "valid": true,
  "refer": true,
  "refers": [
    {
      "step": "annuity_referrals",
      "name": "Medicine with no declared condition",
      "medicines_unmatched": true,
      "exp": "{{medicines_unmatched}} === true",
      "result": true
    }
  ]
}

Only the unmatched medicine fired. A rating of 4 years and a purchase price of £184,912.50 stay inside the other two rules.

Deployment and data

Your cloud, your basis, your model

Nothing runs on our servers. Every component sits inside your own private cloud, and health data never leaves it.

Your private cloudAny cloud provider, or your own data centre
Pricing engine

Your basis, rules, quotes, audit trail and data hash.

ClaudeZDR
AI client

Zero data retention. Reads the case, asks Ptally for the price and explains it.

Document / OCR LLM
Document reader

Turns scans, photos and spreadsheets into fields. The model of your choice.

Health data stays inside. Ptally ships releases in; no data comes out.
Self host
Ptally runs in your own cloud account or data centre. We ship releases. We hold no customer data and need no standing access.
Your model, your approval
Your actuaries own the basis. New versions are drafted, tested against your test pack and published only by your pricing team.
AI with zero data retention
The AI client can use Claude with zero data retention, processed in your region. Or bring your own model.
Documents where you choose
Use the document or OCR model that suits you, on your own GPUs or your cloud's service. Uploads follow your retention rules.
Least data to the model
Documents and extracted values pass between tools by reference where possible. The language model sees what it needs to reason, not every field.
Separate roles
Advisers can quote. Only the pricing team can change the model.
Two ways in

Advised or direct, one price

An intermediary's client and a customer buying direct reach the same engine and the same basis version. What changes is where the data comes from.

Adviser or intermediaryQuotes for a clientCustomer directAn annuity customer buying from you
InputsUploads the client pack: pension statements, health questionnaire and prescriptionsPension details come from your own records. The customer uploads health evidence only
Who confirms the dataThe adviser checks each extracted field against its source pageThe customer confirms their answers, with guidance prompts in the journey
What they seeOptions side by side, a downloadable quote and the application for signatureTheir income options and a quote, with guidance prompts before they commit
What's recordedEvery field and its source, the basis version, every requote and the hash at signatureThe same
Priced byYour basis, the version in forceYour basis, the version in force

Price your next enhanced case in minutes

Send a rate table or a sample basis. We'll build it in Ptally and show a synthetic case priced end to end in your AI client.

Talk to our team