CATALPA · Asset management digitalisation

The hard part of investment management is not the process. It is the definitions.

In most managers, funds, deals, holdings and valuations live across four spreadsheets and three departments. Over ten years and more than 500 institutions we have converged the breaks between pre-investment, portfolio, post-investment and disclosure into one ledger that survives a supervisory filing, an investment committee and an LP question in the same week.

500+ institutions served14,000+ funds on platformRMB 5.6 tn+ assets covered60%+ engineering headcountCMMI Level 3 · ISO 9001

Who we are

Advisory that stays on the hook after go-live

Catalpa is the brand under which Suzhou Zichuan Information Technology Co., Ltd. has worked since 2016 on financial-sector systems, extending into multi-industry digitalisation. The founding team came from Yuanhe Holdings, Founder Technology and Tencent — people who had run deals before writing software, which is why we treat the maintenance of definitions, not the feature list, as the real failure mode.

We do not sell a platform story. Three deliverables carry the engagement: a data dictionary signed by business, risk and finance; a process engine reconfigured to your investment-committee logic; and a response time written into the contract.

10+

Years in asset management systems

500+

Institutions served

14,000+

Funds covered (managed basis)

5.6RMB tn+

Assets covered (managed basis)

Start from the nature of the capital, not the feature list

A provincial SOE group answering to a supervisory filing office and a market GP measured on DPI need opposite data constraints. Establish which one you are before evaluating any system.

Six capabilities, one delivery chain

We commit only to what can be accepted against: blueprint, data dictionary, process configuration, metric definitions, permission model, response times. Each appears in the work-breakdown structure.

How we deliver

Five steps, each with something accept-able

Most failed implementations cut the first incision at "install the system" rather than "settle the definitions". Our first two steps contain no code.

01 · 2–3 weeks

Current-state assessment

Interviews across business, risk and finance; a map of data flows and where they break.

  • Break inventory
  • Definition conflicts
  • Filing burden modelled

02 · 3–4 weeks

Target blueprint

Process, field ownership and metric definitions agreed and signed by three functions.

  • Data dictionary v1
  • Process swimlanes
  • Phased roadmap

03 · 8–16 weeks

Platform build

PEPM configured to your approval logic; legacy data cleansed and migrated.

  • Workflow engine
  • Historical migration
  • Field-level permissions

04 · Ongoing

Governance & integration

One supervised outlet for external filings; DingTalk, Feishu, WeCom and LP/GP portals.

  • Single filing outlet
  • Mobile parity
  • Disclosure templates

05 · Long term

Operations

One year of free service after acceptance, 24/7 response, issue located within an hour.

  • SLA in contract
  • Quarterly definition review
  • Metric library evolution

AI & agents

Put AI where an answer can be checked

The company reports the first AI-agent investment-management system in the sector. The judgement was in scene selection: report extraction, indicator assessment, behaviour logging and effort costing all have objectively checkable answers, so the benefit is measurable on day one. "AI decides the investment" cannot be verified, and should not be sold.

Every AI module therefore sits inside an existing process node and emits reviewable structured fields — not a chat box.

Financial-statement OCR

Multi-format extraction puts portfolio-company and sub-fund reporting into one definitional table.

Risk-model judgement

Custom indicators with 18 warning metrics recalculated daily and retained for audit.

Desktop behaviour control

Monitoring of sensitive operations, meeting compliance-evidence needs in lending and quasi-financial work.

Effort accounting

Automatic time capture makes deal-team cost measurable for the first time.

Market position, with the qualifiers attached

The figures below are customer-coverage counts from 2023–2025 third-party rankings: how many ranked institutions run PEPM. Rankings are re-cut annually and positions move; this is not a claim about current standing or outcomes.

RankingCoverageBasis
Zero2IPO China VC Top 10024 / 100Ranked institutions running PEPM
ChinaVenture Top 50+CoveredSystem adoption among leading market managers
State-owned capital Top 5017 / 50Provincial, municipal and district platforms
Best exit institutions Top 10CoveredShared source for exit modelling and post-investment data
Guidance funds & provincial FOF30%+ SOE shareCompany-disclosed basis; leading among central-province SOEs

Institutions publicly referenced

Names are drawn from the company’s public sites. Listing order carries no meaning about scale or precedence.

国新元禾控股百度资本毅达资本金沙江创投广投集团安徽国控中信建投海通证券深创投集团厦门建发元禾辰坤交子金控上市科创基金

Case pages are anonymised by default; named clients require written authorisation.

Four sides that have to be equally strong

A weak corner surfaces in year two. A feature-rich platform with no data dictionary becomes manual reconciliation the first time a supervisory filing is due.

01

Product: rebuilt to your logic

Highly customisable; approval flows adjusted to your committee rules; public cloud, private cloud or isolated intranet; no breakpoints in end-to-end data flow.

02

Technology: a governable base

Data-dictionary driven with user-defined fields; AES-256 at rest with TLS 1.3 in transit; three-tier permissions at function, row and field level.

03

Service: response times in writing

24/7 emergency response, issue located within one hour, one year free after acceptance, supported from 15 offices.

04

Sector: people who read the filings

Experience across 14,000+ funds on platform, and fluency in the language of SOE groups, guidance funds, market GPs, quasi-financial lenders and energy assets.

When we say no

We will tell you not to buy: a single-fund GP under fifteen people is better served by spreadsheets and a shared drive; a budget that covers the standard edition with no data-governance headcount will decay within two years; and a request to replace investment judgement with automation — AI should compress the time spent gathering evidence and calculating, not the time spent being accountable.

Latest thinking

The problems we have been working on lately

Method and judgement from live engagements, for use while you shortlist and redesign.

System Selection: Definitions Before Features

Feature lists converge; field ownership, reporting exits and metric formulas do not. Four decisions, a meeting checklist, three cases to avoid.

Platform selection20 Oct 202610 min read

Post-Investment Metrics That Compute Themselves

A post-investment metric survives only when numerator and denominator come from objects already in the system. How to sort them, and what to do with the rest.

Post-investment6 Oct 20269 min read

Three-Tier Definitions: the Order That Works

Provincial SOE supervision platforms stall on build order: entity ledger, holdings, metric definitions, reporting mapping, dashboards last.

State capital22 Sept 202610 min read

Get started

Assess the current state before you buy a system

45 minutes with one sector advisor and one architect. We walk pre-investment through disclosure, mark the breaks, and return a phased route with a budget band.