Getting Started

Hissab vs Alternatives

Hissab is the calculation layer between casual calculators and full programming environments. It gives you the convenience of plain-language math with the reliability of a real parser and calculator underneath — and it runs the same engine across the web app, a CLI, an embeddable library, and as a tool that AI agents can call.

25% of 200
15 kilometers to miles
emi(500000, 6.5%, 30 years)
network(192.168.1.130/24)

Most "natural-language calculators" stop at everyday arithmetic and units. Hissab is different along four axes at once: it is strict (it never guesses), unit- and type-aware (metadata flows through the whole computation), broad (around 25 calculation domains in one language), and portable (one engine, many surfaces — including AI agents).

How Hissab compares

Hissab vs Python tools (Claude Code, Codex, code interpreters)

The framing is right tool for the job. A coding agent can write Python for a calculation, but that turns every calculation into a small software task: choose libraries, write code, parse units, pick formulas, manage dependencies, run code, and hope the generated program matches your intent. Hissab removes that layer — the agent emits a short, documented expression like emi(500000, 6.5%, 30 years). The concrete advantages:

1. It collapses the error surface from "an arbitrary program" to "a one-line expression." When an LLM solves math by writing Python, the code itself is generated, so the model is both author and validator. It can pick the wrong formula, botch rounding, mix up percent vs decimal, or fumble a date edge case — and the program runs successfully and returns a confidently wrong number. With Hissab the computation lives in a fixed, tested engine; the only thing the model generates is a short expression you can eyeball.

2. Domain correctness is baked in — Python's isn't. Ask an agent to add 50 grams + 12 km in raw Python and it cheerfully returns 62. Hissab rejects incompatible dimensions by default and carries units natively. Same for timezones, number bases, IP math, and financial conventions — the domain rules can't be skipped.

3. Cheaper and faster for common calculations. For "15% off $240" or "175 lbs to kg", a Hissab expression is one small tool call or shell-out.

4. No code-execution environment required — and no arbitrary-code risk. Running LLM-generated Python is, by definition, arbitrary code execution (it can touch the filesystem, network, and env vars), and many agent contexts have no sandbox at all. Hissab evaluates a constrained expression grammar with no file/network/system access — the worst case is a wrong number, never a compromised host.

5. Reproducible, auditable artifacts. cagr(1000, 2000, 10 years) returns the identical answer in the app, the CLI, and the library. Ad-hoc Python is different every run and depends on library versions and float handling. Hissab leaves a diffable, re-runnable record of exactly how a number was produced.

The honest close: Hissab does not replace Python for heavy data science, ML, large datasets, or custom algorithms. It replaces the reflex of either letting the LLM do unreliable mental math or firing a code interpreter for what is really one expression.

Messaging in one screen

• Exactness — Hissab parses and evaluates deterministically. It does not guess, silently drop unknown words, or ask a model to do arithmetic from memory.
• Breadth — everyday, business, technical, educational, and agentic calculations in one expression language.
• Auditability — every answer traces back to a short, visible, editable expression.
• Portability — the same engine works in the web app, CLI, TypeScript library, and agent skills.
• AI leverage — Hissab turns AI from a risky calculator into a translator and explainer backed by a real math engine.