How Fiscari compares
An honest look at Fiscari against the tools UK accountants actually weigh it against — Tolley, Croner-i, Blue J, and general-purpose AI. Where we win, where we don't, and how to tell the difference.
Prefer to talk it through? Book a demo
What to ask any vendor
Most differences sit underneath the marketing. These four questions surface them.
What corpus is it grounded in?
The best tools restrict the model to a defined, auditable source set — either a paid commentary library or the public UK primary-source corpus. The worst are thin wrappers with no corpus control at all.
Can you click through to the source?
Citations should not be cosmetic. You should land on the specific section or paragraph the answer rests on. If you can't, the grounding is marketing rather than engineering.
Does it refuse when it cannot ground?
The single biggest quality signal. A system that always produces an answer is a system that will hallucinate. One that refuses — and tells you what it could not find — is one you can sign off on.
Who owns the risk if it's wrong?
Every serious vendor preserves your professional judgement, and that's correct. But the tool sets the floor. Citing primary law and refusing when ungrounded gives you a far higher floor than paraphrase without attribution.
At a glance
| Fiscari | Tolley+ AI | Croner-i | Blue J | ChatGPT | |
|---|---|---|---|---|---|
| Grounding corpus | UK legislation, HMRC manuals, VAT notices, tribunal case law | Tolley paid commentary, plus HMRC and legislation | Croner-i paid commentary, plus HMRC | Curated primary sources, plus licensed commentary | Open web training data |
| UK coverage | UK only, by design | Deep UK | Deep UK, UK only | Multi-jurisdiction | No jurisdiction guarantee |
| Citations link to source | Yes — to the manual paragraph or judgment | Yes — within Tolley's own content | Yes — within Croner-i's own content | Yes | Often vague or fabricated |
| Refuses when ungrounded | Yes — published policy | Not marketed as a policy | Not marketed as a policy | Not marketed as a policy; Blue J reports a disagree rate of around 1 in 700 | No — always answers |
| Published pricing | Yes — from £19.50 per user / month | Quote on request | Quote on request | Individual £1,500 / year per user; Team on quote | Consumer subscription |
| Available to any UK firm | Yes | Yes | Yes — content alliance with KPMG UK | Exclusive to KPMG among the UK Big Four for three years (announced February 2024) | Yes |
| Trial without a sales call | Yes | No | No | No | Yes |
| Best suited to | Solos and practices wanting defensible UK answers fast | Large firms already on the Tolley stack | UK-only mid-tier firms wanting a helpline | Cross-border and enterprise tax teams | Not professional tax work |
| vs Tolley | vs Croner-i | vs Blue J | vs ChatGPT |
Compiled from each vendor's own published material, August 2026. Third-party prices and accuracy figures are those vendors' own published figures. Confirm current terms with the vendor.
Where each tool genuinely wins
Each of these can be the right answer. The useful question is which buyer you are.
Fiscari wins on defensibility and speed
Every answer is grounded in the same public corpus you would cite in correspondence with HMRC, and the refusal policy is the product rather than a disclaimer. Pricing is published, and you can test it on real client questions before committing.
Tolley+ AI wins on editorial depth
If you rely day to day on Simon's Taxes, De Voil, or the Yellow and Orange books, an AI layer over that corpus is a real upgrade. For complex structuring you are paying for decades of editorial judgement baked into the source material — and that is worth paying for.
Croner-i wins on UK focus plus a human backstop
Entirely UK content, with a telephone advisory helpline so a human adviser can sanity-check a borderline position. If you want a person at the end of a phone as part of the subscription, that is a legitimate reason to choose it.
Blue J wins on AI maturity and reach
Mature citations, multiple jurisdictions, and a reported disagree rate of around 1 in 700 — the lowest figure published in this market. If your work is cross-border, it belongs on your shortlist and Fiscari does not.
Two caveats: £1,500 per user per year against £468 for a Fiscari individual licence, and exclusivity to KPMG among the UK Big Four until 2027, so PwC, Deloitte and EY cannot use it directly.
Fiscari is probably right for you if…
-
You are a solo practitioner, bookkeeper, or small practice
The incumbents will quote you a figure that doesn't match your revenue per client. Our published pricing is sized for you, and you can start without a procurement conversation.
-
You don't want to commit before you've evaluated
Put real client questions in and judge us on what we refuse as much as what we answer. No annual contract to sign first.
-
You need primary-source grounding more than proprietary commentary
For checking a VAT treatment, clarifying a Corporation Tax section, or interpreting an HMRC manual paragraph, legislation and HMRC's own materials are the sources that matter.
-
You have been using ChatGPT and it worried you
If you have seen a confident answer citing a section that turned out not to exist, you already understand the problem this solves.
Try it on your hardest question
Put a real client question in and check the citations yourself. Free to start, no sales call.
