Case studies
Our work, old and new.
This page covers our flagship agentic engagement, the AI-era crews we're running now, and a selection from the .NET and MS SQL era that we still patch every month. We've rebranded, but the track record stands.
- Years delivering
- 18+
- Continuous since 2008
- AI-era engagements
- 10+ live
- Selected projects shown below
- Sectors served
- 6
- Healthcare to data centre
Flagship, live
Industrial agentic AI for WaterDoctor.
2024 to now, WaterDoctor, live
WaterDoctor: AI agent stack for biological water treatment
An embedded crew of four agents and two engineers runs adaptive treatment loops on WaterDoctor's biofilm reactors, through the AquaMind agent stack and supporting systems. wGrow is WaterDoctor's strategic technology partner. The work is auditable and readable by regulators, with a weekly release.
Read the case studyReported outcomes
- Pollutant discharge
- −50%+
- Water exchange
- −90%
- Energy per unit
- −30%
Aeration setpointIllustrative
Agent output
- ingestor.agent sensor batch received
- eval.agent no drift against baseline
- controller.agent aeration change, inside limits
- reporter.agent figures linked to raw data
- controller.agent proposed change over 15% of baseline
- engineer approval requested, change held
AI-era engagements
Newer crews, under the same delivery rules.
Alongside WaterDoctor, these are three engagements where the deliverable is an agent or a fine-tuned model running in production.
-
Live, private LLM
Private LLM for TV-script production
An embedded engagement with an overseas television-series studio. We fine-tuned a DeepSeek base model on the studio's 3,000+ in-house scripts and shipped a private, on-prem LLM for their writers' room, calibrated to house voice, format and continuity.
DeepSeek, script-corpus fine-tune, private deploy
Overseas TV series studio
-
Live, internal, open-source sample
Server-ops agents on Claude Code
Internal crews built on Claude Code for server management, covering log triage, deployment runs and incident commentary, inside a private, hardened Telegram bot. The reference architecture is open-sourced as ClaudeAgentBOT, so buyers can evaluate the pattern before they engage.
Claude Code, Telegram bot, Linux ops
Internal tooling and public sample
ClaudeAgentBOT on GitHub -
Live, gen-AI pipelines
eCommerce product generation: image and video
Generative pipelines for product research and listing content: image variants, lifestyle stills, and short-form video for marketplaces and ad creative. We use them in our own commerce R&D and on selected eCommerce engagements where speed to creative is the bottleneck.
Diffusion image, short-form video, prompt and eval harness
Internal R&D and eCommerce engagements
From the longer book
Eighteen years of systems we still patch.
Healthcare, legal, government, finance, eCommerce, manufacturing and infrastructure. Most are still running. Here is a selection, and the full archive is on its own page.
-
Healthcare, AWS
Patient Management System
A multi-clinic patient and medical records platform, covering appointment scheduling, lab integration, drug inventory and billing, with field-level encryption and audit trails. Still in production.
AWS, MS SQL, field-level encryption, PDPA
Singapore medical group
-
Government, grants
Grants Management System
End-to-end grants administration for a Singapore grant-making body, with an applicant portal, expert review, claims handling and procurement-grade audit controls.
.NET, MS SQL, audit logs, PDPA
Grant-making body
-
eCommerce, named
Vitasg.SG: health supplements
An end-to-end commerce stack for Singapore-based Vitasg.SG, with Shopee and Lazada integration and regional expansion into Malaysia and Indonesia.
TECOM, Shopee, Lazada, Google SEM
Vitasg.SG
-
Infrastructure, SS 564
Green Data Centre Reports
SS 564 certificate reporting for a data-centre operator, with IoT sensor ingest, PUE calculations and automated submission packs that stand up to an audit.
SS 564, IoT, PUE, MS SQL
Data centre operator
Selected work since 2008
See all our engagements Twenty engagements across six sectors.Brief us
Tell us about
the work.
Tell us what you need. A wGrow engineer will reply within one business day.
Brief us