C Chowdhree

AI consulting — London & remote

AI that does the actual work.

We design, build and ship AI-powered tools for small teams, startups and non-profits: voice assistants, document pipelines, admin automation, private deployments. Anthropic Claude, OpenAI models, Gemini models and open-weight models (DeepSeek, GLM, Qwen, Llama) — chosen per job. For companies that need private, secure AI, open-weight models run on infrastructure you control, and your data never leaves it.

[email protected] Meet Aida — our voice AI →
  • Anthropic Claude
  • OpenAI
  • Gemini
  • DeepSeek
  • GLM
  • Qwen
  • Llama
  • Voice AI
  • On-prem
  • Cloudflare

Selected work

Aida

Voice AI consultant

A 20-minute spoken interview that maps how a business really runs — in 23 languages — and ends with a structured SWOT read-back. Live today at aida.chowdhree.com.

Talk to her →

Masjid Al-Ummah

Tech incubator & community platform

A £4m next-generation masjid in Watford, built as a tech incubator for the community: website, donations, the 313-supporter campaign, and AI tooling for madrasah and youth-programme administration.

See it live →

Automation & agents

Ops tooling

Scheduled agents that scrape, summarise, reconcile and file — running on cron, watching their own logs, recovering when a provider rate-limits them.

What we automate →

What we do

01

AI strategy & diagnostics

We map your processes and find the workflows where AI pays for itself. You get a costed 90-day plan — or an honest verdict that nothing pays yet.

02

Custom AI build & ship

Working tools in weeks, not months: voice interfaces, document intelligence, admin automation. Lightweight, tested stacks, cheap to run.

03

Private & on-prem AI

Open-weight models — DeepSeek, GLM, Qwen, Llama — running on infrastructure you control, so data never leaves. For legal, finance, healthcare and charities handling sensitive information.

04

Voice AI & multilingual agents

Real-time spoken agents in 20+ languages: multilingual receptionists, discovery interviews, field reports for crews who don't write English.

05

AI imagery & creative

AI-generated brand imagery, campaign visuals and product shots that match your style — generated on demand, edited in minutes, licensed clean.

06

Property & architecture visuals

AI building renders and 3D walkthroughs of developments before ground is broken. Material for planning committees, investors and fundraising.

07

Fundraising & donor platforms

Campaign sites, supporter numbering, giving flows and live progress. Our 313-supporter campaign helped fund a £4m community hub.

08

Marketing & web design

Fast, high-performance marketing sites that rank and convert. Flagship example: the Masjid Al-Ummah site and campaign platform.

About us

Decades of shared operating experience sit at the core of the team — not a consultancy career, an operating one. We have carried P&L and delivery responsibility in institutional finance and banking, in very large enterprise operations, and as SME owner-operators. We have founded venture-backed startups with multiple successful exits, developed real estate and property, built data analytics platforms, and shipped video games and advertising. That range is the point: the same operational disease looks different in a courier depot, a letting agent and a studio, and we have seen all three.

How we work comes from a strategy canon, not from whatever shipped last week. Rumelt's kernel for structure (diagnosis, guiding policy, coherent action). Goldratt's Theory of Constraints to find the one bottleneck. Ohno's Toyota Production System to name the waste before pricing it. Deming on the system versus the staff. Lean value-streams for end-to-end time. Google SRE's toil test for what automation should take. Grove's leverage to rank by impact. Christensen's Jobs to Be Done to match AI to the job, not the job title. Wardley Mapping for build-buy-stop. Cynefin for the kind of problem we are in. Meadows for loops and delays. McChesney's lead measures. Gawande's checklists. Kim's flow. These are applied with AI, not recited in slideware.

What we ship is tested, cheap to run and documented. Every number carries its provenance. Every build either gets handed over with docs your team can run, or stays with us as a managed service. No hidden costs, no surprise invoices, no lock-in: models are chosen per job and swappable when prices move.

The canon

The books we build strategy from. Every entry is applied in real work — diagnostics, builds and operations — not cited for decoration.

  1. Good Strategy / Bad Strategy — Richard RumeltDiagnosis → guiding policy → coherent action, in that order.
  2. The Goal — Eliyahu GoldrattFind the one constraint; subordinate everything else to it.
  3. Toyota Production System — Taiichi OhnoName the waste before pricing it.
  4. Out of the Crisis — W. Edwards Deming94% of problems belong to the system, not the staff.
  5. Lean Thinking — Womack & JonesValue-add vs non-value-add time across the whole stream.
  6. Site Reliability Engineering — Beyer et al., GoogleThe test for automatable toil: manual, repetitive, tactical, scales linearly.
  7. High Output Management — Andy GroveRank actions by leverage, not by how busy they look.
  8. Jobs to Be Done — Clayton ChristensenMatch AI to the job, not to the job title.
  9. Wardley Mapping — Simon WardleyBuild, buy or stop — decided by where a component sits in its evolution.
  10. Cynefin — Dave SnowdenMatch the intervention to the kind of problem it actually is.
  11. Thinking in Systems — Donella MeadowsFind the loop and the delay, not the latest symptom.
  12. The 4 Disciplines of Execution — Chris McChesneyLead measures and a scoreboard, or the plan dies in a drawer.
  13. The Checklist Manifesto — Atul GawandeHandovers get an explicit protocol, not a memory.
  14. The Phoenix Project — Gene KimWIP limits and flow, so the queue stops being the constraint.

How it works

  1. Call. A 30-minute conversation about what is slow, messy or expensive in your organisation.
  2. Scope. A one-page plan: what we will build, what it costs, when it ships, and how we will know it worked.
  3. Ship. A working tool in weeks. You use it from week one, not after a six-month programme.
  4. Run. We monitor, maintain and improve it — or hand it to your team with documentation.

Tell us what's slowing you down.

We'll tell you honestly whether AI can fix it — and what it would take.

[email protected]