Én Space cuts conversation handling time by 96% with Filum AI Agent
F&B event-catering case study: Én Space automates 76% of customer-care messages and 70% of lead-capture flow with Filum AI.
About Én Restaurant & Event Space Én Restaurant & Event Space is a Ho Chi Minh City restaurant and event-catering brand, operating since 2015. Én serves corporate events, weddings, conferences and milestone gatherings — from small in-house dinners to large off-site events host…
About Én Restaurant & Event Space
Én Restaurant & Event Space is a Ho Chi Minh City restaurant and event-catering brand, operating since 2015. Én serves corporate events, weddings, conferences and milestone gatherings — from small in-house dinners to large off-site events hosted at the client's location.
Én's strength isn't just a versatile menu and professional catering — it's the ability to read each client's needs and deliver an end-to-end event experience: menu consulting tailored to the event type, coordination with MCs, decor, and sound-and-light partners.
In our industry, if you respond slower than 3 hours, you've almost certainly been dropped from the shortlist.

Context: Én Space operated a 3–6 month B2B sales cycle on Facebook Messenger manually, losing leads whenever reply time slipped past 3 hours — the industry-standard threshold for getting dropped from a buyer's shortlist. Intervention: Deployed Filum AI Agent + Automation + xProfile across Messenger, Instagram and Zalo OA within 2 weeks. Outcome (measured 2 months post-rollout): 96% reduction in conversation handling time, 76% of messages resolved end-to-end by AI, 70% of the lead-capture flow automated.
Methodology: data pulled from the Filum AI Agent dashboard for 01 Mar – 15 Apr 2026 (N = 12,847 conversations across Messenger, Instagram and Zalo OA). "AI-resolved" is defined as a conversation closing without human handoff. Baseline numbers come from Én Space's Q4/2024 chat export, processed through Filum CS Research's standard analytics pipeline. This case study was reviewed and approved by Én Space on 18 April 2026 before publication.
1 · Challenge — when response speed decides who wins
Before 2025, Én's sales team operated under the defining constraint of the event industry: a 3–6 month sales cycle for large corporate bookings, forcing them to track dozens of prospects in parallel — without a proper pipeline tool. Customer data lived scattered across the Facebook page and chat tools that weren't built for event services. Every missed message was a qualified lead quietly lost to a competitor.
"In our industry, if you respond slower than 3 hours, you've almost certainly been dropped from the shortlist." — Én Space representative
Speed wasn't the only gap. Én had no listening channel for end-users: feedback only came from the booking contact (PIC), while the voices of the people who actually attended the events — the richest signal for improving quality — were never captured. Internal feedback was stuck too: staff heard complaints at events but had no standard process to escalate them. Post-contract operations were manual: every detail on decor, MC briefs and setup was passed verbally, prone to error.
2 · Solution — one platform, four problems solved in parallel
From 2025, Én Space partnered with Filum.ai to deploy an integrated customer-experience platform addressing four pain points at once: slow response, fragmented data, manual operations, and no end-user voice.
24/7 response with a multi-channel AI Agent
Within 2 weeks of kickoff, Filum AI Agent went live across Én's customer channels: Facebook Messenger, Instagram and Zalo OA. The AI runs 24/7, intakes 100% of inbound conversations, answers common questions instantly, and captures prospect information automatically. Whether a customer messages at midnight or dawn, they get an immediate reply — eliminating the "slower than 3 hours" problem at the first contact layer.
Implementation detail: the AI Agent was trained on 3 months of historical conversations (Q4/2024), fine-tuned for 5 primary intents — menu consulting, venue/date availability, pricing, setup timeline, complaint handling. Average latency is 1.2 seconds per reply; automatic fallback to a human agent when AI confidence drops below 0.72.
Automated lead capture and sales handoff
Filum Automation and Ticket were configured to automate 70% of the flow from AI intake to qualified lead handover to sales. Customer data is written directly into the internal system, ending lead leakage. After a contract is signed, the system auto-generates tickets and operational checklists for the decor, MC and setup teams — closing the loop from sales to operations.
Listening to end-users via on-site QR surveys
Én expanded its listening surface with QR-code surveys placed at event venues. Every piece of feedback — however small — is captured and analyzed, giving management a 360° view of service quality that extends well beyond the PIC's perspective.
Unified customer data + proactive upsell
With Filum xProfile and Campaign, customer records are unified into a single profile. Én runs loyalty programs for returning corporate customers — clients who book roughly 3 times a year — sending targeted offers and upsell messages at the right moment, turning existing customers into recurring revenue.
3 · Results — before and after rollout
The metrics below were measured in the first full quarter post-rollout (Q1/2026), compared against a Q4/2024 baseline — the last quarter Én operated entirely manually on Messenger.
| Metric | Before (Q4/2024) | After (Q1/2026) | Δ |
|---|---|---|---|
| Average conversation resolution time | 7h 46m | 15 min | −96% |
| Share of messages resolved end-to-end by AI | 0% | 76% | +76 pp |
| Flow automation coverage | ~10% | 70% | +60 pp |
| Volume of conversations handled directly by staff | 100% | 25% | −75 pp |
"Before, we were always overloaded. Now the AI carries most of the frontline work, so we can actually focus on customers." — Én Space representative
4 · Discussion — implications for long-cycle B2B sales teams
Én's story shows that a lean F&B & Event team can operate at the scale and speed of a much larger brand — if they pick the right tools. The real bottleneck for long-cycle B2B service teams isn't the closing conversation; it's the first-response layer. Small teams that can't be available 24/7 lose deals at the shortlist gate, well before price or proposal ever enters the conversation.
Vietnam's F&B industry is projected to grow 9.6% in 2025(1). Competitive pressure at that pace demands not just better core service, but smarter operations at every customer touchpoint. The way Én deployed AI at the frontline — not to replace agents but to free them for higher-value conversations — is a model that generalizes to many other long-cycle B2B sales teams.
5 · What's next
After the first cycle, Én Space and Filum are expanding AI scenarios: upgrading the Lead Capture script and training the AI Agent to filter lead quality to improve sales conversion further. Phase 2 will focus on automating post-contract operations — decor coordination tickets, MC timeline, and event-crew check-in.
Want results like Én Space?
Explore more success stories at filum.ai/customers.
- Decision Lab (2025). Vietnam F&B Market Outlook 2024–2028. Projected compound annual growth rate (CAGR) of 9.6% for the 2024–2028 period.
- Q4/2024 baseline was derived from Én Space's Facebook Messenger and Zalo chat history export, processed with Filum CS Research's standard analytics pipeline — excluding spam and test messages.
- "AI-resolved" in this report is defined as a conversation completed without any agent intervention (no handoff or escalation). Conversations closed with a fallback to a human agent are counted in the remaining 24%.
- Financial figures (revenue, contract value) are not disclosed under the confidentiality agreement between Filum.ai and Én Space.
Top businesses have already seen transformational results with Filum AI
ZUS Coffee × Filum.ai — 1,000 stores, 3 markets: AI handles 55%+ of conversations
1,000 stores, 3 markets — AI resolves 55%+ of conversations, with the same 30-person team.
55%+
~90%
3 markets
Vietravel × Filum.ai — 20,000+ traveler voices, heard in full for the first time
20,000+ voices — the first time Vietravel could listen, tour by tour
50%+
20,000+
24 hours
Sidotech × Filum.ai — AI Agent Resolves ~90% of AfterSales Conversations, 24/7
AI Agent handles ~90% of AfterSales conversations — around the clock
~90%
24/7
>87%
Nông Nghiệp ADI × Filum.ai — AI Handles ~95% of Agri-Input Advisory Conversations in the First Month
Two specialized AI Agents that understand 'farmer-speak' — handling ~95% of fanpage conversations after just one month.
~95%
>75%
1 month
Marmoris Yachting × Filum.ai — Two Brands, Two AI Agents, ~60% of Conversations Automated in Six Weeks
Two premium brands, five chat channels, one workspace — and a customer care team freed for the work only people can do.
~60%
~87%
~98%
Vietnam Television — VTV1 (Cất Cánh program)
>30%
1,000+
First