Source [1]
Industry consultant report
§2.1 · Market structure
“Lantea ranked first by store count among freshly-made tea drinks networks in China as at 31 December 2025.”
Built by former capital markets lawyers
A draft you can
put your name on.
Draftwell AI prepares your HK IPO prospectus using section-specific guidance and a database of 2,500+ IPO precedents. Every factual statement is grounded in your client materials and clearly linked to its source; missing information is automatically added to an information request list.
Comparable basis
- MIXUE Group 2025 · freshly-made drinks
- Guming Holdings 2025 · tea chain
- ChaPanda 2024 · tea chain
- Nayuki 2021 · premium tea
- Chagee 2025 · US listing
Top 10 ranked from 2,500+ precedents — pick 3.
Cited draft
OVERVIEW
OUR COMPETITIVE STRENGTHS
Source behind the fact
Source [2]
DD questionnaire responses
§4.2 · Store network
“As at 31 December 2025, the Group operated 9,120 stores in 318 cities across the PRC.”
Source [3]
Audited financials FY23–FY25
Note 4 · Revenue
“Cups sold in the year ended 31 December 2025: 1,204.6 million.”
A stronger factual foundation
Every factual statement has a source
Draftwell uses your matter materials as factual authority. Each factual sentence carries a citation to the supporting passage.
Nothing in your document set is skimmed
Every document you upload is read, parsed and indexed in full. A long accountants’ report is read through, not sampled—and no more of it than a task needs ever reaches a model.
Knows what’s missing—and who owns it
Unsupported facts stay bracketed and identify who must confirm them. Draftwell automatically collects those gaps into a Word information request list, grouped by section and responsible party.
Built for the whole matter
Precedent library
The matter first ranks comparables from an indexed library of 2,500+ HK and US IPO prospectuses, including 1,900+ A1 filings we archived before they lapse from HKEx.
Defined terms hold across chapters
Defined terms, party names and KPI definitions live in one place. A change flags sections drafted before it for review.
Earlier sections inform later ones
Relevant earlier sections are carried forward as context, so later drafting starts from the matter already established—not from a blank page.
Your data
Your choice of model and hosting
Drafting and research run on the frontier models your firm chooses, on the hosting it chooses. The default is Alibaba Cloud with DeepSeek and Qwen—no cross-border transfer, no training on your data.
Never used to train a model
We run no training pipeline of any kind. Drafting goes only to approved enterprise APIs bound by no-training terms, and an allowlist in our code refuses every provider outside that list.
Isolated matter by matter
Every request for content checks your membership of that specific matter. Your files never join the shared precedent library, and never enter any vector index.
Pricing
Evaluation
See it read your own materials
¥500 · one-time
- Your own deal materials, not a sample
- Overview in full, or your first 10 risk factors
- On-screen preview, watermarked
- Credited in full against your first purchase within 30 days
Matter Pass
For a live mandate
¥10,000 · one matter
- Early access price for all U.S. sections.
- Comparable and precedent research included
- Fair redrafting and re-runs within the matter
- Terminology, information request list and Word export
- Four members for your deal team
Credits
Pay only for what you draft
¥200 · per credit
- No minimum — buy any number of credits
- Drawn down per drafting task
- Same citations, terminology and information request list as Matter Pass
- Use across any matter you're working on
Representing a firm across multiple matters? Talk to us about unified contracting, security review and team onboarding.