Constructing an enterprise AI company on a "foundation of shifting sand" is the central challenge for founders today, in line with the leadership at Palona AI.
Today, the Palo Alto-based startup—led by former Google and Meta engineering veterans—is making a decisive vertical push into the restaurant and hospitality space with today's launch of Palona Vision and Palona Workflow.
The brand new offerings transform the corporate’s multimodal agent suite right into a real-time operating system for restaurant operations — spanning cameras, calls, conversations, and coordinated task execution.
The news marks a strategic pivot from the corporate’s debut in early 2025, when it first emerged with $10 million in seed funding to construct emotionally intelligent sales agents for broad direct-to-consumer enterprises.
Now, by narrowing its focus to a "multimodal native" approach for restaurants, Palona is providing a blueprint for AI builders on the right way to move beyond "thin wrappers" to construct deep systems that solve high-stakes physical world problems.
“You’re constructing an organization on top of a foundation that’s sand—not quicksand, but shifting sand,” said co-founder and CTO Tim Howes, referring to the instability of today’s LLM ecosystem. “So we built an orchestration layer that lets us swap models on performance, fluency, and value.”
VentureBeat spoke with Howes and co-founder and CEO Maria Zhang in person recently at — where else? — a restaurant in NYC in regards to the technical challenges and hard lessons learned from their launch, growth, and pivot.
The Latest Offering: Vision and Workflow as a ‘Digital GM’
For the tip user—the restaurant owner or operator—Palona’s latest release is designed to operate as an automatic "best operations manager" that never sleeps.
Palona Vision uses in-store security cameras to investigate operational signals — similar to queue lengths, table turnover, prep bottlenecks, and cleanliness — without requiring any recent hardware.
It monitors front-of-house metrics like queue lengths, table turns, and cleanliness, while concurrently identifying back-of-house issues like prep slowdowns or station setup errors.
Palona Workflow complements this by automating multi-step operational processes. This includes managing catering orders, opening and shutting checklists, and food prep achievement. By correlating video signals from Vision with Point-of-Sale (POS) data and staffing levels, Workflow ensures consistent execution across multiple locations.
“Palona Vision is like giving every location a digital GM,” said Shaz Khan, founding father of Tono Pizzeria + Cheesesteaks, in a press release provided to VentureBeat. “It flags issues before they escalate and saves me hours every week.”
Going Vertical: Lessons in Domain Expertise
Palona’s journey began with a star-studded roster. CEO Zhang previously served as VP of Engineering at Google and CTO of Tinder, while Co-founder Howes is the co-inventor of LDAP and a former Netscape CTO.
Despite this pedigree, the team’s first 12 months was a lesson in the need of focus.
Initially, Palona served fashion and electronics brands, creating "wizard" and "surfer dude" personalities to handle sales. Nonetheless, the team quickly realized that the restaurant industry presented a singular, trillion-dollar opportunity that was "surprisingly recession-proof" but "gobsmacked" by operational inefficiency.
"Advice to startup founders: don't go multi-industry," Zhang warned.
By verticalizing, Palona moved from being a "thin" chat layer to constructing a "multi-sensory information pipeline" that processes vision, voice, and text in tandem.
That clarity of focus opened access to proprietary training data (like prep playbooks and call transcripts) while avoiding generic data scraping.
1. Constructing on ‘Shifting Sand’
To accommodate the truth of enterprise AI deployments in 2025 — with recent, improved models coming out on an almost weekly basis — Palona developed a patent-pending orchestration layer.
Somewhat than being "bundled" with a single provider like OpenAI or Google, Palona’s architecture allows them to swap models on a dime based on performance and value.
They use a combination of proprietary and open-source models, including Gemini for computer vision benchmarks and specific language models for Spanish or Chinese fluency.
For builders, the message is obvious: Never let your product's core value be a single-vendor dependency.
2. From Words to ‘World Models’
The launch of Palona Vision represents a shift from understanding words to understanding the physical reality of a kitchen.
While many developers struggle to stitch separate APIs together, Palona’s recent vision model transforms existing in-store cameras into operational assistants.
The system identifies "cause and effect" in real-time—recognizing if a pizza is undercooked by its "pale beige" color or alerting a manager if a display case is empty.
"In words, physics don't matter," Zhang explained. "But in point of fact, I drop the phone, it at all times goes down… we wish to essentially determine what's happening on this world of restaurants".
3. The ‘Muffin’ Solution: Custom Memory Architecture
One of the crucial significant technical hurdles Palona faced was memory management. In a restaurant context, memory is the difference between a frustrating interaction and a "magical" one where the agent remembers a diner’s "usual" order.
The team initially utilized an unspecified open-source tool, but found it produced errors 30% of the time. "I believe advisory developers at all times turn off memory [on consumer AI products], because that can guarantee to mess every little thing up," Zhang cautioned.
To resolve this, Palona built Muffin, a proprietary memory management system named as a nod to web "cookies". Unlike standard vector-based approaches that struggle with structured data, Muffin is architected to handle 4 distinct layers:
-
Structured Data: Stable facts like delivery addresses or allergy information.
-
Slow-changing Dimensions: Loyalty preferences and favorite items.
-
Transient and Seasonal Memories: Adapting to shifts like preferring cold drinks in July versus hot cocoa in winter.
-
Regional Context: Defaults like time zones or language preferences.
The lesson for builders: If the perfect available tool isn't adequate in your specific vertical, you will need to be willing to construct your individual.
4. Reliability through ‘GRACE’
In a kitchen, an AI error isn't only a typo; it’s a wasted order or a security risk. A recent incident at Stefanina’s Pizzeria in Missouri, where an AI hallucinated fake deals during a dinner rush, highlights how quickly brand trust can evaporate when safeguards are absent.
To forestall such chaos, Palona’s engineers follow its internal GRACE framework:
-
Guardrails: Hard limits on agent behavior to stop unapproved promotions.
-
Red Teaming: Proactive attempts to "break" the AI and discover potential hallucination triggers.
-
App Sec: Lock down APIs and third-party integrations with TLS, tokenization, and attack prevention systems.
-
Compliance: Grounding every response in verified, vetted menu data to make sure accuracy.
-
Escalation: Routing complex interactions to a human manager before a guest receives misinformation.
This reliability is verified through massive simulation. "We simulated 1,000,000 ways to order pizza," Zhang said, using one AI to act as a customer and one other to take the order, measuring accuracy to eliminate hallucinations.
The Bottom Line
With the launch of Vision and Workflow, Palona is betting that the long run of enterprise AI isn't in broad assistants, but in specialized "operating systems" that may see, hear, and think inside a particular domain.
In contrast to general-purpose AI agents, Palona’s system is designed to execute restaurant workflows, not only reply to queries — it's able to remembering customers, hearing them order their "usual," and monitoring the restaurant operations to make sure they deliver that customer the food in line with their internal processes and guidelines, flagging at any time when something goes mistaken or crucially, is about to go mistaken.
For Zhang, the goal is to let human operators deal with their craft: "In the event you've got that delicious food nailed… we’ll inform you what to do."
