AI projects, built properly.
Custom AI products, integrations, and architecture-level advisory by someone who ships under the Papi Labs banner. Done with your team, not done to them.
Three things you get that most AI builds miss.
-
01
Architecture first.
Seven years of shipping software means I work at the architecture layer: requirements, system design, data flow, integration, risk, and review. AI-assisted tooling does the typing. I read the code, review it, and own the result.
-
02
Built for real use, not demos.
Every project ships into someone's hands. 100+ AI tools and products built over the last three years — GuideKit (live with hundreds of active users), OtterLog (in beta), ClipShield (in market). I know what makes the leap from prototype to production.
-
03
Done with you, not done to you.
Your team isn't passive while I build. Workshops, pair sessions, architecture reviews — they come out of the engagement more capable than they went in. The deliverable is a working product and a team that can extend it.
From idea to shipped product.
Discovery first
A free 20-minute intro call to understand what you're trying to build, who it's for, and whether building is the right move. Some projects are better not built. I'll tell you which, honestly.
Then a scoping conversation
A 45-minute Project Discovery call to define scope, timeline, engagement shape, and success criteria. We agree on what done looks like before any code is written.
Engagement shapes I work in: fixed-scope project builds, fractional product / CTO retainers, architecture advisory, AI strategy consulting, and end-to-end MVP development. Most engagements blend two or three of these — sized to what your project actually needs.
Recent AI products from my own portfolio.
The most honest proof of how I build is the things I have built. All live (or in active beta) under the Papi Labs banner.
A Chrome extension that teaches AI tools at the exact moment you need them. Tested prompts, 15-minute lessons, multilingual. Live with hundreds of active users, growing weekly.
Meeting notes that never leave your computer. Records, transcribes, and summarises locally. No cloud, no account, no telemetry. Mac and Windows.
Local-LLM PII protection. Spots and shields sensitive content before it ever leaves your machine into an AI conversation. Processing happens on-device.
Saman Taherimoud.
AI Product Builder and Solutions Engineer based in Newcastle. Seven years shipping software, four years building specifically with LLMs since the davinci-3 era. Founder of papilabs, the AI-native venture studio behind GuideKit, OtterLog, and ClipShield.
I work architecturally. The line-by-line authoring is AI-assisted, the same way senior engineers have always delegated boilerplate and scaffolding to their tooling. I read the code. I review it. I own it.
See the full picture at samantm.com ↗Every project is scoped on the discovery call, with a custom proposal, a clear deliverable, and a fee that matches the work. Mission-led and impact-driven organisations: there's always room to find a way.
Got an AI project in mind?
The 20-minute intro call is for figuring out fit. Tell me what you're trying to build, and I'll be honest about whether I'm the right person to help you build it.