Dashboard + review
Local checklist board and draft review. Real screenshots are awaiting capture.
We build document-chasing workflows that turn missing bank statements, invoices and receipts into specific email drafts. Your team stays in control of every message.
Missing-Docs Agent is a synthetic-data prototype, not a live client service.
Local checklist board and draft review. Real screenshots are awaiting capture.
Post-fix results are published with the method and failure cases. The OCR-noise text category scored 4/7; the limits stay visible.
The demo records approval locally. It never sends an email.
Missing-Docs Agent turns a folder of client documents into a checklist of what is still missing, then drafts the chase.
LLM extraction reads the documents. Plain Python rules compare them with the client's checklist. The dashboard shows complete, missing and unclear items alongside the email draft.
Human approval, day 3/7/14 reminder logic and a local approval audit trail are built into the prototype. Live sending and the database-backed workflow still need pilot testing.
See the workflow and evidence plan ↗Reserved for a capture of the actual Streamlit dashboard using labelled synthetic client documents.
The recording will show extraction, missing-item flags, the chase draft and local human approval. No video is available yet.
Two small interactive demos are planned. They are not available on this site yet.
A company-number lookup for public filing dates and overdue records, with a plain-English summary of what needs attention.
No lookup form or API connection is active here. API credentials will stay server-side.
A missing-items list turned into polite, specific reminder drafts in a UK practice tone. Nothing sent, no real documents required.
The interactive version needs a separate build. The flagship already uses deterministic chase templates.
The prototype creates drafts and requires a named human to approve them. Approval writes a local outbox record; it does not send.
Checklist results show the missing item, reason and rule source. The prototype also keeps a local record of draft creation and approval.
Every demo uses labelled synthetic data. Before a real-data pilot, we need to agree the purpose, access, retention and deletion rules, and the data-processing terms.
UK GDPR and ICO considerations are part of pilot preparation. This is not a claim of GDPR certification, ICO registration or readiness for real client documents. We will review hosting, AI providers and any international data transfers with your practice before a real-data pilot. UK-hours contact times will be agreed before it starts.
An AI automation practice based in Hyderabad, India, founded by Mujtaba Javed. We build for UK accountancy practices, with a named human responsible for design, testing and review.
Our team uses AI agents for the build work. Accountants remain responsible for deciding what their clients need. The prototype gives no tax advice and files nothing with HMRC.
Gemini and Groq provider chain, checklist rules and labelled synthetic evaluation.
DIEMS / BATU · Graduated 2025
The post-fix results are published with their limits. Actual dashboard captures and a demo recording will follow when they have been checked.
Tell us which documents your team spends time chasing. We can discuss a small pilot, what it needs to prove, and where human review stays in the loop.
Email to arrange a time. A booking calendar is not connected yet.
Mojtaba Javeed Siddiqui
mojtaba.javeed@gmail.com
WhatsApp · +91 9699651506
Hyderabad, India