真北科技Zhenbei
HomeCustomer stories
Customer Stories

Six industries, six real deliveries

Client names are industry-anonymised under confidentiality terms, but the scale figures, rollout scope and outcome metrics are real delivery results. We would rather show you what it looked like before and what it became.

Retail chain11 weeks (including legacy data migration)

42 subsidiaries and 17 business systems consolidated into one identity kernel

A retail group in East China

Each of the group’s 42 subsidiaries maintained its own account system, so staff had to remember five sets of passwords across systems; permission scopes were inconsistent, and the same role had very different rights in different systems. Group security demanded that “any access can be explained”, but no cross-system audit view existed.

42
subsidiaries brought under tenant isolation
17
legacy systems connected over OIDC
5 → 1
passwords each employee must remember
Real time
time to revoke access on departure
UIAM
Read the full story
Financial services9 weeks (including compliance review)

Let AI agents do the work — without handing them a master key

A licensed financial institution’s digital-workforce programme

The business wanted AI agents to take over customer enquiries and first-pass document review, but compliance had a hard line: an agent must never touch production data with a developer’s personal credentials, and every data access must trace back to “which user authorized which agent”. The earlier pilot had all agents sharing one service account, so when something went wrong there was no one to hold accountable.

0
agents sharing a credential
100%
of tool calls leaving an audit trail
Intersection
how delegated permissions are computed
Immediate
take-effect time of an authorization revocation
UIAMUniclaw
Read the full story
Vocational education7 weeks (including migration verification)

6 microservices folded into 1 container, with not one record of legacy data lost

A vocational education group

The original learning platform was split into 6 microservices calling each other over HTTP with URL-based configuration. One version upgrade meant coordinating 6 repositories and a dozen-odd containers — expensive to operate and hard to troubleshoot. Meanwhile the system had accumulated a large body of historical course and learner data, and the business was explicit: no rebuilding from scratch.

6 → 1
deployment units
146
syllabus nodes migrated
0
records lost in migration
Original password
how existing users log in
UnilearningUIAM
Read the full story
Beauty retail chain6 weeks (including WeCom service-provider authorization)

The customer assets of 800 advisors, visible to head office for the first time

A beauty retail brand

The brand had more than 800 in-store advisors, and customers were added to each advisor’s personal WeChat. Head office could not see the size or structure of the customer base, nor which channel spend was working. When an advisor left, the customers left with them, and there was no way to hand over the conversation history.

800+
advisors brought under unified management
100%
of conversations compliantly archived
End to end
channel-source attribution
37%
of common enquiries handled by the digital worker
UniscrmUniclaw
Read the full story
Advanced manufacturing10 weeks (including knowledge-base curation)

Twenty years of process documents, turned into an in-house expert that never leaves the intranet

An advanced equipment manufacturer

The company had accumulated more than twenty years of process documents, equipment manuals and fault-handling records, and a new hire needed six months or more shadowing senior engineers. A public-cloud AI approach had been tried before, but process parameters are core assets — data leaving the intranet was unacceptable.

0
data leaving the intranet
20 years
of accumulated documents in the knowledge base
6 months → 6 weeks
time for a new hire to work independently
7×24
availability of knowledge Q&A
UniclawUIAM
Read the full story
Software services3 weeks (tool rollout and process change)

Debugging sessions became team assets, cutting integration time in half

The R&D centre of a SaaS vendor

Front-end and back-end integration had long run on “try it again on your side”. Mobile issues were the hardest to localise: client logs were incomplete, intermittent production faults resisted reproduction, and a single API issue took two to three days of back-and-forth on average. The team also wanted to bring AI into triage but lacked structured traffic data.

50%+
average drop in integration time
100%
of production issues reproducible
3 shells
sharing the same kernel
JSON
default output for AI consumption
pxcNewTool
Read the full story
How We Report

How these numbers are measured

We do not try to make the cases look better than they are, so we spell out how the numbers are measured.

Client names are industry-anonymised

Under confidentiality agreements, client names are not published. The industry, scale and delivery scope are real.

Outcome metrics come from acceptance reports

The numbers in each case come from measurements taken at acceptance or client-side statistics — not estimates or projections.

What did not work is written down too

If a scenario did not go well, or we switched approaches in the end, the case says so instead of hiding it.

Get in touch

Hand the complexity of identity, agents and private domain to one governable kernel

Whether you are replacing an existing IAM, building an agent platform, or trying to make private-domain operations actually work — start with a 30-minute architecture call. We will first judge whether this is the kind of problem we are good at, and say so plainly if it is not.

Book an architecture callhi@zhenbei.tech
Delivery & servicesFour delivery models, one six-step pathSecurity & complianceWhere the boundaries are drawn, mapped to complianceEngineering notesTechnical decisions and what they costAPI & contractsRead the contracts before the sales call: OIDC endpoints, token exchange and agent tool-grant APIs