ActionLens

ActionLens is a self-reported document-processing tool that claims to convert official letters, emails, and instructions into verified actions with source-backed evidence. The author states it uses GPT-5.6 for understanding documents and extracting structured data, and implements deterministic code verification to ensure recommendations are traceable to the original text. It is built as a local-first browser application using IndexedDB, with no retention of uploaded files. The system requires explicit human confirmation before saving any AI-generated action.

The project appears to be a solo effort by Jhaman Hirani, who describes it as an attempt to solve the problem of missed deadlines due to unclear or complex official language. It is positioned as a privacy-first tool that avoids AI hallucination by enforcing verification through code rather than relying on model confidence alone.

The single most important open question is: What actual user base or traction exists beyond the author's own testing and evaluation suite?

The description states that ActionLens turns letters, emails, and instructions into proof-linked actions. It claims to propose deadlines, tasks, reminders, and payments from uploaded documents.

It uses GPT-5.6 for document understanding, structured extraction, photo transcription, and scam-signal detection.

The system is built as a local-first application using IndexedDB, where saved actions and history remain in the browser. Original uploaded files are not retained.

Every AI-generated action requires explicit user confirmation before being saved.

The author claims that verification is enforced by deterministic code rather than model behavior alone.

Inference: The product appears to be an AI-powered document parser with a focus on trustworthiness through evidence linking and human confirmation, rather than a general-purpose summarizer or assistant.

The description states that ActionLens was inspired by the difficulty people have in converting official language into actionable tasks—especially when tired, scared, or in a second language.

It positions itself as an alternative to traditional AI summarizers, emphasizing proof-linked actions and deterministic verification over model confidence.

The author claims to have built a system where “AI proposes. Code verifies. You decide.”

It is described as a privacy-first workflow that avoids storing uploaded files and requires explicit human confirmation before saving any action.

Inference: The positioning evolved from solving a personal problem (missed deadlines in public administration) into a broader tool for trust in AI-generated actions, with an emphasis on evidence-based outputs and user control.

The description states that real obligations arrive buried inside official letters—councils, hospitals, landlords, tax offices, universities.

It targets people who are "scared, tired, or in a second language" when dealing with official documents.

It is described as being built for individuals who struggle to convert complex or unclear official language into actionable tasks.

Inference: The ICP appears to be individuals or small businesses who regularly encounter complex official communications and need help extracting clear actions from them. It may also appeal to those concerned about AI hallucination in sensitive contexts.

Not evidenced.

The description does not mention any pricing model, monetization strategy, or business model. There is no indication of whether the tool will be offered as a freemium service, subscription, or one-time purchase.

Built with: api, codex, css, gpt-5.6, indexeddb, next.js, openai, playwright, react, tailwind, typescript, vercel, vitest

The system is local-first using IndexedDB.

Original uploaded files are not retained.

Every AI-generated action requires explicit user confirmation before being saved.

Development was done conversationally with Codex and GPT-5.6.

An adversarial evaluation suite was built to test for hallucination traps and misleading letters.

Inference: The technical stack suggests a modern web application with strong emphasis on privacy and deterministic behavior. The use of Codex and GPT-5.6 implies reliance on large language models, but the architecture is designed to prevent over-reliance on model output through code-based verification.

Not evidenced.

There is no mention of users, customers, revenue, or adoption beyond the author’s own testing and evaluation suite.

The project was submitted to a hackathon (OpenAI 2026), but there is no indication of follow-up traction or product development post-hackathon.

Not evidenced.

There is no mention of competitors or existing solutions in this space. The description does not compare ActionLens to other tools or platforms that might perform similar functions.

• Solo team: Only one member listed (Jhaman Hirani), which raises concerns about scalability, maintenance, and long-term development.

• No traction or revenue evidence: The project is described as a hackathon submission with no indication of real-world usage or monetization.

• Unverified claims: All claims are self-reported and unverified. No third-party validation or data on performance exists.

• Limited scope: The system only handles document parsing and action extraction, without integration into workflows or external systems.

• Privacy vs. utility trade-off: While local-first and privacy-focused, the lack of cloud storage or sharing features may limit usefulness for collaborative environments.

1. What specific types of documents have you tested with? How many different formats?
2. Have you conducted any user testing beyond your own adversarial evaluation suite?
3. Are there plans to integrate with calendar, task management, or other productivity tools?
4. What is the expected path to monetization or scaling beyond a personal tool?
5. How do you plan to handle edge cases like multi-language documents or ambiguous instructions?
6. Is there any intention to open-source parts of the system or make it available as an API?