Define
On a short scoping call we agree which questions Fin must nail, which channels and languages, and the accuracy bar. That call is also what your fixed quote is built from.
Product · Packaged Chatbot Fin in a Box
The bot in the corner is the product.
Fin in a Box is exactly what it sounds like — the assistant in the corner of this page, packaged for your business. Grounded in your content, evaluated against your real questions, deployed to your channels, and handed to your team to own. Live in 2–3 weeks, not a six-month platform migration.
- Live in 2–3 weeks
- Grounded in Your docs, not guesses
- Yours to own Prompts · evals · code
01 Try it now
Don't take our word for it. Open Fin.
The assistant in the corner of this page — "Fin" — is the product, running live. Open it and ask it about our services, pricing, or how a build works. What you get for your business is the same engine, grounded in your content instead of ours.
Claude-poweredAnthropic, swappable
GroundedAnswers from real content
BilingualEnglish & Arabic, live
Routes to humansBooks a call, no dead ends
Your own infraNo third-party SaaS
Logged & reviewedEvery chat, for tuning
02 What's in the box
Scope, ingest,
eval, launch.
Ground
We pull in your docs, site, and knowledge base, build the retrieval layer, and wire guardrails, refusals, and human handoff — so Fin answers from your material, not the model's open memory.
Test
We run Fin against a real question set, tune until it clears the bar, and shadow it on live traffic before it answers anyone for real. You get to see the numbers before you go live.
Ship & own
Fin goes live on your channels, conversations log to your dashboard, and the whole thing is handed over — prompts, evals, code — for your team to run and extend.
03 Built for trust
Guardrails, refusals,
and a human on the line.
A bot that answers is only worth shipping if it knows when not to. Fin is engineered to stay inside your content and route out cleanly when it can't help.
Retrieval, not guesses
Every answer is drawn from your own content through retrieval — docs, site, knowledge base — and cites where it can. Fin doesn't reach into the model's open memory to fill gaps.
- RAG-grounded
- Cited sources
- Your content only
Knows when to stop
When Fin can't answer from your material, it says so — "I don't know, let me connect you to a human" — instead of inventing. Refusal behavior is designed in, not bolted on.
- Refusal logic
- No hallucinated answers
- Scope boundaries
Human on the line
Tricky, high-stakes, or emotional conversations route to a human with full context — book a call, open a ticket, or pass to live chat. No dead-end contact forms.
- Context-rich handoff
- Calendar booking
- No dead ends
Proof, not hope
An eval suite we agree on up front, run against a real question set, so you can prove Fin is accurate rather than hope it is — before launch and continuously after.
- Agreed eval bar
- Real question set
- Ongoing monitoring
PII & PDPA aware
Fin runs on your infrastructure and accounts. We configure data residency for embeddings and logs, run a PDPA review of the conversation flow, and set retention and purge rules for anything personal.
- Your infra
- PDPA review
- Retention & purge
Handed to you
At launch you get the prompts, the eval suite, and the code. Nothing critical is trapped on our side and there's no paywall around your own prompts. The retainer is a choice, not a lock-in.
- Prompts & code
- No lock-in
- You own it
04 Where Fin earns its keep
SEA runs on WhatsApp. So does Fin.
The same grounded brain that lives in a website widget deploys to WhatsApp Business and your help desk — one brain, every front door, unified logs. A few conversations teams across the region keep having by hand:
WhatsApp supportAnswers on the channel buyers use
Order & status queriesGrounded in your systems
Pre-sales Q&AProduct, pricing, how-to
Distributor & reseller helpSame answers, every partner
Multilingual + SinglishEN · 中文 · Malay · Arabic (RTL)
Clean human handoffBooks the call, no dead ends
Proof
An agent that replaced a triage workflow.
- −71%Triage time
- 14→1Person workflow
- End-to-endAudit-traceable
05 How you pay for it
Two ways to work.
No mystery pricing.
We don't post a single fixed from-price, because the honest number depends on your channels, languages, and the questions Fin must nail. We quote it after a short scoping call — then you decide how much of the ongoing work you want us to hold.
Launch
A fixed-scope build, quoted after your scoping call and priced before we start. Scope, ingest, eval, launch — Fin live on your channels in 2–3 weeks, handed over with prompts, evals, and code for your team to own.
Operate & Improve
An optional monthly retainer for teams who'd rather we keep Fin tuned — new content ingested, evals re-run, refusals and handoffs refined as your questions evolve. Cancellable, not a lock-in. See how engagements are structured.
Try it, then talk to us
Ask Fin first.
Then book the call.
Open the assistant in the corner and put it through its paces — that's the product, running live. Then tell us the conversation your team keeps having by hand, and we'll scope Fin in a Box for it on a short call. Prefer the full service view? See AI chatbots.
06 Fin in a Box FAQ
The questions buyers
actually ask.
What exactly is Fin in a Box?
It's the same assistant you can see in the corner of this page, packaged as a fixed-scope build for your business. We ground it in your content, evaluate it against a real question set, deploy it to your channels, and hand you the whole thing — prompts, evals, and code — to own. It is not a SaaS subscription with your prompts locked behind a paywall.
How fast can Fin go live?
A first production Fin is usually live in 2–3 weeks — grounded in your content, evaluated against real questions, shadowed on live traffic, and handed to your team to own and extend.
How is Fin in a Box priced?
There's no single fixed from-price. We quote a fixed-scope Launch build after a short scoping call, once we know your channels, languages, and the questions it must nail. An optional monthly Operate & Improve retainer keeps it tuned as your content and questions change. See our pricing breakdown for how engagements are structured.
Will Fin hallucinate or make things up?
Fin is grounded in your own content — docs, site, knowledge base — using retrieval, not the model's open memory. It answers from your material, cites where it can, and is built to say "I don't know, let me connect you to a human" rather than invent. We agree an accuracy/eval bar before launch and monitor it after.
Can Fin run on WhatsApp?
Yes. SEA runs on WhatsApp, so the same grounded brain deploys to WhatsApp Business alongside your website widget and help desk (Intercom, Zendesk, Freshdesk). One brain, every front door — with unified logs.
Do we own Fin after launch?
Yes. Fin runs on your own infrastructure and accounts. At launch we hand over the prompts, the eval suite, and the code, so nothing critical is trapped on our side. The Operate & Improve retainer is optional, not a lock-in.