The team behind a 4,811★ open-source library.
Now building AI-native products.
GetWidget started as a Flutter UI kit. Eight years, 4,811 GitHub stars, and 23,000 monthly pub.dev downloads later, we pivoted to AI. Same engineering team, same discipline, different problem. We design and ship AI agents, automation workflows, and production AI systems.
Started in Flutter.
Moved into AI when the work changed.
2017. A small team in Bengaluru starts building a Flutter UI kit. The problem is simple: Flutter components are inconsistent across projects, and every new mobile app wastes the first two weeks reinventing buttons, cards, and nav bars. The library ships as open source on GitHub and pub.dev.
2018–2022. The library grows. Developers across India, the US, and Europe add it to their production apps. By 2022, the GetWidget Flutter UI kit sits at 1,000+ widgets, over 4,000 GitHub stars, and tens of thousands of monthly downloads. The engineering team — still mostly the same people — gets good at one thing: building reusable, well-tested UI infrastructure that doesn't break in production.
2023. The team starts shipping AI features on top of their Flutter work — LLM-backed search, document summarization, voice interfaces. The pattern becomes clear: clients want the AI layer, not just the UI layer. The highest-value problems aren't "build me a button" anymore. They're "route this support ticket," "extract these invoice fields," "draft this outbound message." Work that used to take rule-based systems and headcount now ships in six weeks with an agent and an eval suite.
2024–present. GetWidget Labs pivots to an AI-native product studio. The Flutter UI kit stays live — it's still the company's most-downloaded open-source project and the engineering discipline that sharpens our production instincts. But the new work is AI: agents, automation workflows, Claude and OpenAI integrations, production systems with eval suites and cost reporting. The pivot isn't a rebrand — it's the same team following the highest-value problem.
Honest version of the company today: Bengaluru engineering team, remote-first, registered US mailing address in Dallas (mail forwarding only — no team on-site), serving clients in the US, UK, UAE, and India. No sales team. Discovery calls are run by engineers.
Four principles, not a manifesto.
Each one comes from a mistake we made.
These aren't aspirational values. They're constraints we imposed on ourselves after shipping work that didn't hold up — or that we oversold.
Outcomes, not demos
Every engagement has a metric. We don't ship pilots that die in staging — the audit phase scopes the work, the pilot phase ships behind a flag, and the metric drives whether we continue. No outcome, no phase 2.
You own the work
Code, prompts, eval suites, and data all transfer to you at project end. We retain the right to reuse operator patterns (how to build a support triage agent) but not your stack, your data, or your architecture.
Honest discovery before any build
The $3K audit phase exists so we can tell you not to build. A walk-away clause is in every contract. If the data doesn't move the metric, we'll tell you at the kill-point rather than quietly build past it.
Model-agnostic by design
We pick the right model per workflow — Claude for long-context reasoning, GPT-4 for structured outputs, Llama for cost-sensitive batch on your infrastructure. The model is one variable in the code; the workflow logic is portable.
Small, senior, Bengaluru.
No offshoring of the difficult work.
We don't staff engagements with junior resources and have a senior review at the end. The engineer on your discovery call is the engineer building your system. Senior AI engineers stay on the work — that's how we stay accountable to the eval suite.
Bengaluru-based engineers who have shipped production Flutter code since 2017 and AI systems since 2023. Small enough that everyone knows every live engagement.
Senior means: you own the eval suite, not just the code. You ask about the metric before you ask about the model. You know when to say the pilot should stop.
Discovery to pilot to continuous — the same core team. We don't hand off to a delivery team after the sale. The person who scoped the work builds the work.
Registered US mailing address for clients who need a US point of contact. No team on-site. Engineering is remote-first from Bengaluru. Discovery and delivery calls are on Zoom.
Open source since 2017.
Still the team behind it.
The GetWidget Flutter UI kit has 4,811 GitHub stars and 23,000 monthly downloads on pub.dev. It's not a side project — it's what taught us how to build infrastructure that developers trust in production. The discipline behind it (eval before ship, owned by the team that built it, documented to the point of being replaceable) is the same discipline we apply to AI work.
Where to go next.
Depending on what you need.
If you want to understand the AI services, see the case studies, or talk to the engineering team directly — these are the fastest paths.
Talk to the engineers.
$3K audit, walk-away clause, 1–2 weeks.
Every engagement starts with a two-week discovery audit — we map your highest-ROI workflows, score them by ROI, risk, and time-to-ship, and deliver a 90-day roadmap with per-workflow model picks and cost projections. Walk-away clause baked in: if the audit says don't build, you keep the deliverable and pay nothing more.