All products
04Finance & accountingAI assistant

LSync AI

An AI assistant for chat, documents, web search and image edits.

LSync AI is the assistant portal LSync runs at ai.lsync.io. One account gives you chat through LSync's own model gateway, with an open model by default, web search with sources, questions over the documents and images you attach, and image edits on LSync's own server. New accounts are approved by an administrator before first use.

At a glance

Status
Live
Sector
Finance & accounting
Best for
Chartered accountants and tax practitioners, lawyers and legal teams, finance teams at SMEs, developers, students, founders and privacy-sensitive teams
Built with
Next.js 15 · LiteLLM model gateway · SearXNG · FLUX.2 [klein] · SQLite
Please note
Not legal or tax advice.

The problem

An AI assistant for professional work usually means several: one to chat, another to read a PDF, another to search the web, another to touch up an image, each with its own account and its own idea of where your files go. And on a question of Indian tax or company law, a general model answers from memory, with no section to check it against. LSync AI puts chat, documents, web search and image edits behind one LSync account. The portal, your history, search and the image engine run on LSync's own server; chat answers come from cloud models through LSync's gateway, with a last fallback on LSync's server. Its next release adds a library of Indian Acts and Rules, so law answers can cite the section they rely on.

Highlights

  • Chat through LSync's model gateway: a Cloudflare-hosted open model by default, cloud fallbacks behind it, and a last fallback on LSync's own server
  • Web search through SearXNG, with the sources listed alongside the answer
  • Attach PDF, DOCX, TXT, MD or CSV files, or images for a vision model to read
  • Image edits on LSync's own server with FLUX.2 [klein], up to 768 px, 15 jobs per user per day

What LSync AI does today

01

Chat that does not hang on one provider

Every message goes through LiteLLM, the model gateway LSync runs. The default model is an open model hosted on Cloudflare. If it errors or hits a limit, the request falls through a chain of other cloud models, and the last link is a model on LSync's own server. Conversations are kept in your history until you delete them.

02

Answers from your files and the web

Attach a PDF, Word document, text, Markdown or CSV file and ask about it, or attach an image for a vision model to read. Switch on web search and the question goes to a SearXNG search instance; the pages it finds are read on the server and listed as sources with the answer.

03

Image edits on LSync's own hardware

The Private image engine runs FLUX.2 [klein] on LSync's server, at up to 768 px. An edit takes about 2½ minutes at 512 px and 4 to 5 minutes at 768 px, longer with several images or when others are in the queue. Each instruction and image is screened by automated safety checks, and every account gets 15 image jobs a day.

04

Your account, your settings

Set custom instructions once and every chat follows them, and pick a light or dark theme. From Settings you can export your data, or delete one chat, every chat or the whole account. LSync AI is free to use at present, with no daily message cap.

Who it is for

Not legal or tax advice

01Chartered accountants and tax practitioners
Summarise a client's PDF or CSV export, draft correspondence, and search the web for a circular with the sources listed. Law mode, coming soon, adds the Acts and Rules with section citations.
02Lawyers and legal teams
Read and summarise long documents, compare two drafts, and check web sources before you cite them. Law mode is coming soon.
03Finance and accounts teams at SMEs
Ask questions of the CSV exports and reports you attach, and draft emails, notes and process documents.
04Developers and IT teams
Paste code, logs or a stack trace and ask what is going on. Code mode, coming soon, sends coding questions to an open coding model on LSync's own server, with the default cloud model as its fallback.
05Students: CA, CS, CMA, law and engineering
Ask for an explanation in plain words, summarise study material you attach, and follow the web sources behind an answer.
06Founders and small businesses
One assistant for writing, research, documents and image edits, free to use at present.
07Privacy-sensitive teams
Image edits run on LSync's own server, images you edit (and the results) are not used to train models, and deleting your account removes your conversations. Images you attach to a chat for the model to read go to the cloud chat model as uploaded. Chats can be used to improve and train the service, so read the terms first.
Privacy

Where your data goes

01

Chat goes through our gateway to the model

Most answers come from a cloud-hosted open model, and a fallback can be another cloud provider, so your message is processed by that provider under its terms. The gateway, the routing and the last fallback model are LSync's own.

02

Image edits run on our server

The Private engine runs on LSync's hardware. The automated safety check may send a reduced copy of your image and your instruction to a cloud model provider. Camera and GPS metadata are stripped from photos uploaded for editing, results are labelled as AI-made in their metadata, and uploads are kept 30 days and results 180 days, or until you delete the chat or your account.

03

Images you attach to a chat

Images you attach to a chat for the model to read are sent to the cloud chat model as uploaded and kept with that chat until you delete it.

04

You hold the delete button

Conversations stay until you delete them. Export your data, or delete one chat, all chats or the account itself; deleting the account removes your conversations from the service.

05

Plain terms

The terms allow chats to be used to run and improve the service, including training; images you edit, and the results, are not used to train models, and personal data is not sold. Read them before you paste anything confidential.

Full terms and privacy policy: ai.lsync.io/privacy (opens in a new tab)
Coming soon

In the next release

Law mode

Coming soon

Answers drawn from a library of 914 Indian Acts and Rules, 42,447 sections in all, with the section cited: the Income-tax Act, 2025 and the Rules of 2026, the Income-tax Act, 1961 for earlier years, the CGST, IGST and UTGST Acts, the CGST and IGST Rules, the Companies Act, 2013 with Schedule III, the LLP Act, the IBC, FEMA, the Contract Act, the labour codes and about 880 other central Acts. Accounting standards are covered only in part: AS 1 to 29 through the Companies (Accounting Standards) Rules, 2021, and Ind AS 115 and 116.

Code mode

Coming soon

Code mode sends coding questions to an open coding model, Qwen3.6-35B-A3B, on LSync's own server. If that model is unavailable, or your message includes an image, the default cloud model answers instead and the reply says so.

Start here

Have something to build, or something to keep running?

Whether you need a product shipped, a system kept healthy, or a hand with a hard integration — tell us what you're dealing with.