01
Brief
30 MIN30-minute call. We listen, ask hard questions, and decide together if we're the right team for the job.
YOU GET
- Mutual NDA
- Scope clarified
- Go / no-go call
NOW SHIPPING · V1.0
Engineered intelligence for ambitious teams
AI platforms, LINE-native commerce, realtime sports infrastructure, and automation pipelines — built to ship, not to demo.
SCROLL
Building with teams across SaaS · sports · commerce · operations
Court Republic3x3 Bangkok LeagueSiamRetailNorthStar LogisticsFieldOpsVertex StudioInsightAIHatchwaveCourt Republic3x3 Bangkok LeagueSiamRetailNorthStar LogisticsFieldOpsVertex StudioInsightAIHatchwave
FIND A PROJECT / 00
Search anything we've shipped.
⌘ K
AI · RAG · WIDGET
ChatBot platform with custom RAG & embeddable widget
ReactpgvectorOpenRouter
AI · RAG · INTERNAL OPS
BigBrain — AI 2nd brain for IT & Ops teams
Next.js 15FastAPIpgvectorBGE-M3
LINE · AI · THAI
LINE Official IT Support agent (Thai)
LINE APIOpenRouterServiceNow
LINE · COMMERCE · PROMPTPAY
LINE commerce chatbot with Flex Messages & QR PromptPay
LINE FlexPromptPayNode.js
SPORTS · OBS · LIVE
Live broadcast scoreboard — OBS-ready, any sport
WebSocketsOBSReact
SPORTS · SCHEDULE · FANS
Public tournament schedule & standings site
Next.jsPostgreSQLWebSockets
3X3 · MULTI-COURT · STREAM
Multi-court 3x3 basketball platform — full broadcast kit
WebSocketsYouTube LiveRedis
OPERATIONS · AUTOMATION
Queue automation with n8n, Power Automate & Grafana
n8nPower AutomateGrafana
CAPABILITIES / 01
A full stack for intelligent products.
We design, build, and run the infrastructure modern teams need — from conversational AI to operational automations and live data pipelines.
Custom AI & RAG systems
Conversational agents grounded in your data. We build the ingestion, vector store, retrieval, and front-end widgets — embedded anywhere your users live.
OpenRouterpgvectorembedding pipelinesstreamingfunction calling
$ nexable rag ingest --source ./docs --vector pgvector
→ embedding 1,284 chunks · model: text-embedding-3-large
→ indexed to production · 312ms p95 retrieval
$ nexable agent deploy
→ agent support-th-v3 live · widget embed ready
Workflow automation
From ticket triage to nightly reports — we wire the systems your team uses into one orchestrated flow.
n8nPower AutomateGrafana
RUNS · LAST 12 HOURS
LINE & messaging commerce
Flex Messages, carts, PromptPay payments — full LINE OA storefronts and AI support agents.
Live scoreboards & broadcast kits
Custom scoreboards for any sport — OBS streaming overlay, mobile score operator, public schedule site. Built per tournament, themed to your league.
Operations & observability
Queue automation, monitoring dashboards, alerting — so your business actually runs while you sleep.
HOW WE WORK / 02
Brief to deployed in five steps.
No discovery decks, no waterfall handoffs. The same senior engineer who scopes your project is the one who ships it — and the one on call when you need them.
02
Architecture
48 HOURSA build plan with stack choices, milestones, and risks. Fixed estimate. No fishing expeditions.
YOU GET
- Build plan doc
- Fixed quote
- Timeline
03
Build
6–12 WEEKSWeekly demos in your Slack. You see it shaping up — no surprise reveals, no end-of-project disasters.
YOU GET
- Weekly demos
- Slack channel
- Live staging
04
Ship
1 DAYProduction deploy with observability built in: dashboards, alerts, runbooks. Boring on purpose.
YOU GET
- Live URL
- Dashboards
- Runbooks
05
Operate
ONGOINGWe stay on. Tuning, scaling, on-call support, monthly check-ins — for as long as you need us.
YOU GET
- Monthly check-in
- On-call support
- Iteration
0+
Production systems shipped across AI, commerce, and live ops.
0.00%
Uptime across deployed services in the last twelve months.
0ms
Median p95 retrieval latency across our RAG deployments.
0×
Average operational throughput gain after we automate a workflow.
THE TOOLBOX / 04
Built on tools we trust in production.
We're opinionated about the stack — but pragmatic. These are the tools we reach for first because they hold up under load.
AI / RETRIEVAL
OpenRouterLangChainpgvectorPineconeHugging FaceLiteLLM
FRONTEND
ReactNext.jsTypeScriptTailwindFramer Motionshadcn/ui
BACKEND / DATA
Node.jsBunPostgresRedistRPCPrisma
LIVE / REALTIME
WebSocketsPusherOBS BrowserYouTube LiveTwitch API
AUTOMATION / OPS
n8nPower AutomateGrafanaPrometheusSentry
LINE / COMMERCE
LINE Messaging APIFlex MessagesPromptPay QRLIFF
SELECTED WORK / 05
All projects →Recent deployments.
Acme RAG · KB v4
Trained on 1,284 docs. Ready when you are.
What's our refund policy on enterprise?
From doc #88: enterprise refunds prorate within 30 days. Cite §3.2?
AI · RAG · WIDGET
ChatBot platform with custom RAG & embeddable widget
A full RAG pipeline with chunking, embeddings, retrieval-with-citations, and a one-line embed widget that drops onto any page.
BigBrain · Ops
RAG · 1.2k docs
วันนี้ผมมีงานอะไรบ้าง
3 tasks open · 1 due today
ตรวจสอบ Log Web-01DUE TODAY
Grafana CPU dashboard fixIN PROGRESS
Confirm before write · tool: create_task
AI · RAG · INTERNAL OPS
BigBrain — AI 2nd brain for IT & Ops teams
A self-hosted platform that unifies knowledge base, asset registry, and a lightweight task tracker behind a single conversational agent. Hybrid RAG (BGE-M3 + Postgres FTS, RRF-merged) keeps answers grounded; a tool-use loop with a confirmation modal lets the agent create and update records safely.
LIVE FROM THE STUDIO / 06
What we shipped this week.
DEPLOY FEEDLAST 72H
- 10:24deployed v3.2 to court-republic
- 08:15ingested 4,212 chunks · sukhumvit-line-oa
- 06:43p99 latency ↓ 28% · queue-automation
- YDAYshipped 3x3-platform
- YDAYreleased Flex carousel v4
- MONopened support-th-v3
Always shipping. Always learning.
We don't disappear after launch. Every project we run stays under our care — measured, observed, improved. The feed on the left is a snapshot of the last few days across our active stack.
11
DEPLOYS THIS WEEK
0
P0 INCIDENTS · 90D
2.1M
INFERENCES · 30D
99.97%
SLA · ROLLING 12M
“
We went from a brittle spreadsheet workflow to a live realtime platform our entire league runs on. Nexable didn't just build us software — they shipped us the unfair advantage we'd been trying to articulate for two years.
AP
Aran Phongphat
FOUNDER · 3X3 BANGKOK LEAGUE
QUESTIONS / 08
Things we get asked often.
Most production builds run 6–12 weeks from kickoff to live deploy. Smaller prototypes (a working RAG widget, a LINE OA agent) can ship in 2–3 weeks. We'll quote you a fixed timeline within 48 hours of our first call — no fishing expeditions.
Both. Most projects start as a fixed-scope build, then transition into a monthly retainer for ops, tuning, and new features. Some clients just want us to ship and hand off — that's fine too. We don't lock you in.
Yes — half of what we ship lives inside larger codebases. We integrate into your Git, your CI, your standup. We're senior enough to drop in without ramp-up, and small enough to stay out of the way.
Prototypes start around $10k. Production builds run $30k–$80k. Multi-phase platforms (3x3 league, custom RAG infra) sit at $80k+. We share a fixed estimate with the architecture doc — no surprises on invoice day.
Bangkok-based (GMT+7), with most clients in APAC and EU. We run async by default but join your standups when it matters. For US clients, expect 4–6 overlap hours a day.
Always. Send us yours, or use our standard mutual NDA. For projects touching PII, payment data, or health records — we'll walk you through our data-handling SOPs before we touch a line of code.
Have a system you wish already existed?
Tell us what's slow, fragile, or impossible today. We'll come back with a build plan within 48 hours.
Start a project →