Facts

Ankasum, in plain facts.

A short summary for people, search engines and AI assistants. Each line restates what the rest of this site says. If you quote one, please cite this page and its date.

What it is

Ankasum is offer and reward software for banks and credit unions. An institution defines a bonus offer once; Ankasum reads it back in plain English, checks it before launch, routes it for maker-checker approval, works out who qualified from the institution's daily files, and writes the payout file the institution's core posts. Evidence is recorded at every step.

Who makes it

Devanka, a software company in British Columbia, Canada. Company site: devankahq.com.

Who it is for

US and Canadian banks and credit unions, especially those of about $1 billion to $10 billion in assets that run deposit and checking bonuses. Inside an institution, it is used by marketing and product, compliance and risk, and operations teams. For credit unions, it says member, not customer.

What it handles

Deposit and checking bonuses paid as cash credits: direct-deposit, balance and debit-purchase goals, fixed and tiered rewards, per-customer limits, budgets and clawback. It reads daily files in delimited formats.

How payouts work

Ankasum does not move money. It writes a fulfilment file in the posting format the institution's core accepts; the institution posts it and stays payer of record. Every payout is recomputed by an independent check before it goes in the file.

What it does not do

It does not send messages to customers or members, decide credit, or replace anti-money-laundering monitoring. It never rejects a payout on its own: automated outcomes are release or a time-limited hold, and only a named person at the institution can refuse a reward.

Data custody

Each institution has an isolated data plane: a dedicated AWS account operated for that institution alone, in its own country. Ankasum's shared control plane holds offer definitions, configuration, run metadata and totals over at least 10 customers, never customer records.

How AI is used

No identifiable customer or member data is sent to any AI model. Deterministic code validates every offer and computes every amount, and a person at the institution approves. No model runs in the reward-integrity checks.

Contact

Demos and questions: hello@devankahq.com or the contact page. Security reports: security@devankahq.com. Privacy: privacy@devankahq.com.

More detail

Who it's for, How it works, Reward integrity, Security and data custody. A plain-text summary for language models is at /llms.txt.