Section 01
Caspers: Brand Value & Why It Failed
What the brand built, what sank it, and why the failure was structural, not conceptual.
Series A Raise
€5M
April 2022 · Slingshot Ventures
Annual Meals at Peak
230K
Across BE, NL, FR
YoY Revenue Growth
210%
Before bankruptcy (Jan 2024)
What Caspers Was
Caspers was founded in Ghent in 2020, during the COVID lockdowns. Unlike typical dark kitchens, it adopted a hybrid model: transparent street-facing locations with open windows, rider waiting areas, and self-service ordering technology. At its peak, it operated 10–12 virtual brands per location, Gloria (pasta), Barney's (burgers), Tuktuk (Asian cuisine), Bertha (Flemish cuisine), and others.
Caspers operated across Ghent, Antwerp, Brussels, Leuven, Kortrijk and Mechelen, giving the brand direct recognition in Belgium's largest urban delivery markets.
Why Caspers Failed
- The hybrid storefront model was too expensive. Visible high-street locations meant retail-level rents, but without retail-level revenue.
- Platform commission pressure. Takeaway.com and Deliveroo taking 15–30% per order left no margin to cover fixed costs.
- A last-minute investor pulled out. Cash flow collapsed. Spookje BV filed for bankruptcy in January 2024.
The failure was structural, not conceptual. The Belgian market did not reject the ghost kitchen model. It rejected the capital-heavy version of it.
What Still Survives
Three things outlasted the bankruptcy. Brand recognition, four years of delivery history among urban consumers aged 25–40 in Ghent, Antwerp, and Brussels. Digital footprint, established platform ratings and reviews for Gloria, Barney's, and Tuktuk that are partially recoverable. Positioning, the "honest ghost kitchen" concept was genuinely differentiated and well received.
What GFV's Model Changes
| Caspers (old model) | GFV X (new model) | |
|---|---|---|
| Kitchen ownership | Leased & fitted by Caspers | Existing restaurant kitchens |
| Upfront capital | High, per location | Zero |
| Fixed costs | Rent, equipment, staff | None |
| Scale risk | High, each location a liability | Low, partners absorb cost |
| Brand | Strong in Belgium | Acquires Caspers brand equity |
Section 02
Belgium Delivery Market
Three-platform market with one dominant player, and why timing has never been better.
Platform Active Users, Q4 2024
Practical takeaway: Takeaway.com is the anchor platform for any Caspers relaunch. Uber Eats is the secondary. Deliveroo is optional, worth maintaining existing presence, but not the launch priority.
Platform Commission Pressure
Typical cost stack for a virtual brand kitchen at 30% platform commission, before rent, marketing, or utilities.
The March 2026 BTW Shock
Belgium doubled the VAT rate on restaurant food on 1 March 2026, ending a COVID-era measure that operators had built their entire pricing model around.
Before March 2026
6%
VAT on restaurant food
From March 2026
12%
VAT on restaurant food
−€3,000 / month
Margin erosion for a typical mid-size kitchen (80 covers/day · €25 avg), without any change in costs
The timing is not accidental. The Caspers relaunch is entering a market where operator motivation to find new revenue streams is at its highest point in years. Incremental delivery revenue with minimal additional overhead is exactly what a kitchen under BTW pressure is looking for.
The Competitive Field
Caspers' collapse did not create a domestic successor, but the vacuum has been partially filled by Taster (Paris, $55.6M funded), now holding 7th largest delivery market share in Belgium. Its brands, Out Fry, Pepe Chicken, Starsmash, have no pre-existing Belgian recognition.
| Taster | GFV + Caspers | |
|---|---|---|
| Brand recognition (Belgium) | Zero, built from scratch | 4 years of platform history |
| Belgian consumer familiarity | None | Ghent, Antwerp, Brussels, Leuven |
| Funding | $55.6M | Asset-light, no owned infrastructure |
| Local partner | Unknown | Agenterix (12yr HORECA operator) |
Operator Sourcing
The Operator Sourcing Problem
Why Belgium is not moving at UK pace, and what it structurally takes to change that.
United Kingdom
90
locations signed in one month
60+ individual operators in 3 weeks
+ 1 deal with 27 locations
Belgium
Slow
operator acquisition significantly behind
Not a pace problem.
A structural market access problem.
Belgium's total addressable pool of kitchen partners is structurally smaller than the UK. But the gap is wider than market size alone explains. There are specific barriers, and each one has a countermeasure.
Why Agenterix
Five Problems. One Partner.
Relaunching Caspers in Belgium requires solving five problems simultaneously. All five sit within Agenterix's capability.
The Market Knowledge Problem
On 3 April 2026, GFV posted a "Restaurant Growth Partner - Belgium" job listing. That posting is evidence that GFV has already concluded it cannot execute the Belgium relaunch without a local partner. The hiring process takes 4–6 months. That person starts from zero. Agenterix is available now, with twelve years of Belgian HORECA experience.
The Operator Qualification Problem
Standard selection processes use checklists, food safety, equipment, delivery capacity. These are necessary but insufficient. Nusret reads a kitchen differently: physical layout, team organisation, platform rating as a signal of operational culture. A kitchen with a rating below 4.2 has structural problems that a virtual brand will amplify, not resolve. That assessment takes twenty minutes on site.
The First-60-Day Churn Problem
The highest-risk period for any kitchen adding a virtual brand is the first 30–60 days. Without structured guidance, many kitchens fail to adapt and exit early. Every kitchen that churns in month two represents lost recurring revenue, wasted onboarding cost, and damaged Caspers platform ratings. Agenterix's 60-day support model turns this risk into a foundation.
The Automation Problem
Running a virtual brand alongside an existing kitchen adds complexity: order routing across multiple platforms, inventory tracking for brand-specific ingredients, multi-brand coordination during peak hours. Agenterix builds this infrastructure using Make and n8n, automation designed around operational problems Nusret ran manually for six years as a Domino's operator. He was the coordination layer between kitchen and courier. He is now building the system that replaces what he used to do by hand.
The Concept Architecture Problem
Wrong concept matching is one of the primary causes of virtual brand underperformance. Nusret can walk into a kitchen, read the existing menu, assess the equipment, evaluate the culinary capability of the team, and determine which Caspers concept fits best. Where no existing concept fits, Agenterix develops a new virtual brand concept from scratch.
Full Belgian Market Coverage
Dutch-speaking Flanders and French-speaking Wallonia are effectively two separate markets with different consumer behaviour and different operator cultures. Agenterix operates in both languages. A single Belgian partner covering the full national programme is simpler for GFV than managing separate regional relationships, or defaulting to English and losing the trust that comes from speaking to operators in their own language.
The Partner
Nusret Bayindir spent twelve years inside Belgian HORECA before founding Agenterix, six as a professional chef in restaurant kitchens, followed by six as the owner of a Domino's franchise in Flemish Brabant. As a franchise owner, he managed 18 staff, held full P&L responsibility, operated his own delivery fleet of cars, motorcycles, and bicycles, and at peak owned two locations simultaneously. He negotiated directly with Takeaway.com and Uber Eats as a platform partner, not as a user, but as an operator whose livelihood depended on understanding how their algorithms actually worked.
Because he ran three delivery platforms simultaneously, Domino's proprietary system, Takeaway.com, and Uber Eats, he built his own real-time coordination system from scratch: tracking devices and courier platform applications, synchronised by hand, to keep kitchen output aligned with courier dispatch and minimise order wait times. He was the coordination layer between kitchen and delivery. Agenterix is the system that replaces what he used to do manually.
His professional kitchen background gives him a capability no operational consultant can replicate: he can assess a kitchen's culinary capacity from the inside. Walk in, read the menu, evaluate the equipment, talk to the team, and determine which Caspers concept that kitchen can actually produce at quality and speed. And where no existing concept fits, he can build one. He was a chef before he was a franchise owner. The combination of culinary judgment and operational experience is what makes concept matching reliable rather than optimistic.
As a Domino's franchisee, his own kitchen was evaluated against the same criteria he now applies to others: hygiene, organisation, staff standards, platform performance. He has sat on both sides of that assessment table. His qualification judgements come with a credibility that an external consultant, however well-trained, cannot claim.
The move from franchise operator to AI automation founder is not a career pivot. It is the same problem, solved at a different level. Every system Nusret built manually, tracking devices, platform dashboards, courier synchronisation, was an attempt to close the same gap: between what delivery platforms demand and what a real kitchen can sustain under pressure. Agenterix exists because he spent six years building those solutions by hand, hit the ceiling of what manual coordination could achieve, and built the company to do what he no longer could alone. Agenterix is not a technology firm that studied the restaurant industry. It is a restaurant operator who built a technology firm, because he ran out of ways to solve the problem without one.
Agenterix was not built for this industry. It was built from inside it. That is the difference between a partner who understands operators in theory and one who has been the operator, running the kitchen, negotiating the platforms, absorbing the margin pressure, and building workarounds until better tools became possible. Six years as a professional chef. Six years as a franchise owner running three delivery platforms simultaneously. Now automating exactly what he used to do by hand. Bilingual in Dutch and French.
Nusret did not discover the virtual brand model. The virtual brand model is the formalisation of what he was already doing: multiple concepts, single kitchen, multiple platforms, optimised coordination, built manually, under pressure, for six years. He was not learning this model. He was living it before it had a name. Agenterix is what happens when that operator decides to build the infrastructure instead of just surviving inside it.
Section 03
Target Cities & Operator Profiles
Ghent, Brussels, and Bruges: each a distinct market dynamic and launch strategy.
Phase 1 · Priority Start
Ghent
Highest probability of fast reactivation
Phase 1 · Parallel Track
Brussels
Largest market, most competitive
Phase 2
Bruges
New territory, tourist-driven opportunity
| Restaurant | Cuisine | Rating | Reviews | Signal |
|---|---|---|---|---|
| Ghent | ||||
| Seoul South Station | Korean | 4.4 | 620 | Highest volume in city |
| Poké Hub | Poke | 4.4 | 380 | Best Newcomer 2025 winner |
| MG's Delicious | Fast Food | 4.3 | 78 | Consistent performer |
| Brussels | ||||
| Sensei Sushi Jette | Japanese | 1st | high | Best of Brussels 2025 winner |
| Better Than Hungry | Thai/Asian | 4.7 | 130 | Strongest rating in sample |
| Barouche | Pita/Kebab | 3.9 | 39 | Average performer |
Key signal: Asian cuisine concepts dominate top ratings in both cities. Caspers' TukTuk (Asian) and Gloria (Italian) are the strongest relaunch candidates based on demonstrated consumer demand. Operators below 4.2 are not viable kitchen partners, platform ratings this low reflect structural operational problems a virtual brand will amplify.
Operator Profile We Are Looking For
Across all three cities, the kitchen partners most likely to succeed share a consistent profile: established operations with documented delivery history, identifiable off-peak kitchen capacity, cuisine compatible with the Caspers brand family, and, in April 2026, margin pressure. That last criterion is easy to meet.
The operators least likely to succeed are those adding virtual brands as an afterthought to an already at-capacity kitchen. That is where platform ratings get damaged, and where a second chance on Takeaway.com is structurally very difficult to recover.
Section 04
Agenterix's Role
Five capability areas, applied to the Ghent and Brussels pilot, with Bruges as the subsequent phase.
Kitchen Sourcing & Qualification
Agenterix builds the qualified pipeline through on-the-ground outreach, not digital databases. The owner of a family brasserie or kebab restaurant is not on LinkedIn. They are not in Apollo. The only sourcing method that works in Belgian HORECA is physical: visit the kitchen, speak to the owner directly, in their language. The qualification process is a physical assessment against four criteria: hygiene standards, kitchen organisation as a proxy for operational discipline, staff stability as a proxy for management authority, and current platform performance. A kitchen with a rating below 4.2 is not a candidate. The initial conversation is structured around a Business Impact Analysis, a number, not a pitch.
Concept Matching & Development
Once a kitchen passes qualification, Agenterix determines which Caspers virtual brand concept to place in it, evaluating the kitchen's existing menu, equipment capability, and team skills against the production requirements of each concept. Gloria (pasta), Barney's (burgers), Tuktuk (Asian), and Bertha (Flemish) have different production profiles. Wrong concept matching damages both kitchen performance and platform rating. Where no existing Caspers concept fits, Agenterix develops a new virtual brand concept from scratch.
First-60-Day Operational Support
Weekly operational reviews online (60 min), platform performance, order accuracy, customer satisfaction, emerging issues before they compound. Multi-platform order optimisation. Customer satisfaction monitoring with direct feedback to the kitchen team. Demand forecasting and staffing planning to prepare kitchens for volume peaks before they happen. This support is provided to the kitchen operator as a separate service, the fee structure will be determined together in the next stages.
Automation Infrastructure
Agenterix builds lightweight automation workflows using Make and n8n: order routing logic across platforms, inventory tracking for brand-specific ingredients, coordination alerts between kitchen stations during multi-brand service, and performance dashboards for both the kitchen operator and GFV. This is not theoretical infrastructure. Nusret spent six years manually running exactly this coordination system as a Domino's operator, synchronising kitchen output with courier dispatch across three platforms in real time. The automation removes the friction that makes adding a virtual brand feel like more work than it is worth.
Full Belgian Language Coverage
Agenterix manages the full Belgian programme in Dutch and French. Operator outreach in Ghent and Bruges in Dutch. Brussels in both languages. All operational support, training, and documentation in the operator's language. In Belgian HORECA, trust is built in the first language. An English-speaking intermediary creates a persistent friction that Agenterix eliminates entirely.
Section 05
Pilot Proposal
A structured test, not a soft launch. Three to five kitchens, clear success criteria, a validated Belgian playbook.
3–5 Kitchen Pilot
Ghent + Brussels
Find the operators most likely to succeed, run the programme properly, including full first-60-day support, and use the results to build the case for a national rollout. Not twenty kitchens to see what happens. Three to five done right.
The pilot runs over a 90-day period, kitchen onboarding in weeks 1–4, full 60-day operational support in weeks 5–12, and a structured debrief with performance data and market findings at the close.
Operator Viability Data
Which kitchen profiles perform and which do not, a qualification model that applies beyond Belgium.
Brand Reactivation Data
Whether the Caspers name generates measurable uplift in Takeaway.com discoverability vs a new brand starting from zero.
Belgian Market Playbook
Which concepts work where, which kitchen types convert fastest, which platform dynamics are Belgium-specific. The foundation for scaling.
What Success Looks Like at Day 90
The pilot has four measurable outcomes. If all four are met, Belgium is ready for national rollout. If any are not, the findings tell us exactly why, and what to fix before scaling.
- Platform ratings maintained. All pilot kitchens hold a Takeaway.com rating of 4.2 or above through the full 90-day period. A drop below this threshold signals a concept or operational mismatch, and is itself a finding.
- Zero early exits. All onboarded kitchen partners remain active at day 90. Churn before the 60-day support period ends is the primary risk this pilot is designed to prevent.
- Incremental revenue confirmed. Each kitchen can document measurable order volume attributable to the virtual brand, separate from their existing business.
- A replicable playbook. A documented operator qualification model, onboarding protocol, and concept matching framework, ready for GFV to use for national rollout without reinventing the process city by city.
What We Need From GFV
Confirmed brand concept list
Which of the original Caspers virtual brands (Gloria, Barney's, Tuktuk, Bertha, or others) are being brought back, and with what menu scope. This determines which kitchen profiles Agenterix is screening for.
Platform account access or introductions
Specifically for Takeaway.com Belgium, where Caspers' historical ratings may be partially recoverable. The faster platform presence is established, the faster the pilot generates meaningful data.
A preferred start timeline and any existing kitchen contacts from the Caspers acquisition
If GFV has retained relationships with any of the original Caspers kitchen partners, those are the fastest path to a first qualified pilot kitchen. And knowing whether the target is Q2 or Q3 determines how Agenterix structures the sourcing timeline in Ghent and Brussels.
The commercial framework, pilot scope, roles, and fee structure, will be defined together on that next step. This brief intentionally leaves it open: the right structure is better built in conversation than defined on a page.
The Caspers opportunity in Belgium is real. The timing, BTW shock, operator margin pressure, GFV's Caspers acquisition creating a head start, is better than it has been in years. Taster, already holding 7th largest delivery market share in Belgium, sold 11 million meals in 2025 and is expanding across 250 locations in 70+ cities. The window is open. It will not stay that way. Once you have reviewed it, we can align on the pilot structure via a short call or over email, whichever you prefer.