Sona

Sona (getsona.com) is an AI-powered workforce management (WFM) platform specifically designed for “frontline” or “deskless” industries like social care, hospitality, and retail. Unlike traditional WFM tools that focus primarily on digitizing timesheets, Sona positions itself as a “Frontline Operating System” that uses “Agentic AI” to automate complex scheduling, forecasting, and communication tasks.

1. Positioning and Value Proposition

Sona targets large-scale, multi-site enterprises where labor management is complex and paper-heavy.

The “Self-Driving Car” of WFM: The founders use this metaphor to contrast Sona with legacy systems. While old tools simply digitize manual work, Sona aims to automate the decision-making process itself (e.g., automatically building the “perfect” roster based on demand).
Frontline-First: It emphasizes the employee experience, offering a “consumer-grade” mobile app that allows staff to claim shifts, manage leave, and communicate with management as easily as using a social media app.
Agentic AI Focus: Sona distinguishes its “Agentic AI” (which can take actions and provide nudges) from standard machine learning (which only shows data on a dashboard).

2. Strengths and Weaknesses

Strengths

Deep Sector Expertise: The founders previously built Catapult (a hospitality staffing app), giving them specialized knowledge in social care and hospitality.
All-in-One Integration: It unifies scheduling, time & attendance, HR, payroll, and internal communication into one platform, reducing the need for fragmented “point solutions.”
Reduction in Agency Spend: A standout feature is the Shift Filler, which acts as an internal marketplace. It helps companies fill shifts with their own staff first, significantly cutting down on expensive third-party agency fees (one customer reported a 63% reduction).
High-Tier Backing: Supported by Google’s Gradient Ventures, giving it strong technical credibility and capital for R&D.

Weaknesses

Niche Focus: While they are dominant in hospitality and care, the platform’s features are highly tailored to these sectors, which may make it less adaptable for desk-based corporate environments.
Implementation Complexity: As an enterprise-level “operating system,” replacing legacy systems across 300+ sites (like their client Loungers) requires a significant and potentially disruptive transition period.
Emerging AI Tech: Their “Agentic AI” claims are cutting-edge, but like all generative/agentic tech, it relies on high-quality data input to avoid errors in critical areas like payroll and compliance.

3. Current Competitors

Sona competes against three tiers of players:

Legacy Enterprise Systems: Large, older WFM platforms like UKG (Ultimate Kronos Group), Rotageek, and Planday (owned by Xero).
Modern Point Solutions: Apps that focus on just one part of the stack, such as Deputy (scheduling), 7shifts (hospitality-specific scheduling), or Workforce.com.
Generalist HR/Ops Platforms: Tools like Rippling or Hibob, which are moving more into the frontline space but often lack the deep specialized scheduling logic Sona provides for social care/hospitality.

4. Market Perception

Innovation Leader: Sona is perceived as a disruptor in the “left-behind” industries of care and hospitality. It was recently ranked in the Startups 100 index (2025) and has won awards like “Innovative Solution of the Year” for care homes.
High ROI: In the market, it is viewed as a “pain relief pill” for GMs and HR directors who struggle with labor shortages and thin margins, as it directly addresses EBITDA by optimizing labor costs.

5. Principal Players

Steffen Wulff Petersen (CEO & Co-Founder): Former investment banker (Goldman Sachs) and experienced entrepreneur with a focus on scaling tech companies.
Ben Dixon (CTO & Co-Founder): A lifelong engineer and author of books on software deployment; he leads the “Agentic AI” and technical architecture.
Oli Johnson (CFO & Co-Founder): Brings a finance and operations background to ensure the platform solves the bottom-line issues of enterprise clients.

(Note: There is an unrelated, inactive company called Sona.AI based in India; for current workforce management, you should exclusively look at getsona.com.)

Sona (getsona.com) differentiates itself by moving away from “traditional” machine learning (ML) and toward a vision of Agentic AI.1 While most workforce management (WFM) tools use AI for forecasting, Sona’s “Agentic” approach means the AI can take autonomous actions, such as filling shifts or adjusting rosters, rather than just providing data on a dashboard.2

Below is a breakdown of their AI features, purpose, and technical stack based on public disclosures.

1. AI Features & Purpose

Sona categorizes its AI features into three primary “Agents” and a core ML engine:

The Scheduling Agent (Auto-scheduling):3
- Purpose: Automatically builds “perfect” rosters that balance demand, staff preferences, compliance (visa limits, skill sets), and labor costs.4
- Action: It doesn’t just suggest a schedule; it proactively identifies gaps and uses “Shift Filler” to offer open shifts to eligible internal staff via the mobile app before managers even realize there is a shortage.5
The BI Analyst Agent (Reporting & Insights):
- Purpose: Allows managers to ask natural language questions like “What drove the difference in financial performance this week vs. last?“6
- Action: It conducts “Deep Research” by pulling data from scheduling, payroll, and engagement modules to generate a narrative report.
The HR Assistant Agent:
- Purpose: A 24/7 frontline support tool.
- Action: It is trained on a company’s specific employee handbooks, policies, and operating manuals.7 It answers staff questions (e.g., “How do I book maternity leave?” or “What is the policy on overtime?”) instantly, reducing admin for GMs.
Forecasting Engine (The ML Backbone):
- Purpose: Predicts future demand (sales, footfall, or care requirements).8
- Action: Uses historical data and external variables (weather, events, seasonal trends) to provide “intra-day” re-forecasting, allowing locations to adjust staffing levels in real-time.

2. The Technology Stack & Models

Sona’s architecture is “AI-native,” meaning it was built to support autonomous agents rather than having AI “bolted on” to a legacy database.9

Models: Proprietary vs. Third-Party

Proprietary ML for Forecasting: Sona uses bespoke machine learning models for its demand forecasting.10 Because scheduling requires highly specific, numerical, and time-series data (e.g., hospitality sales patterns), they train and deploy custom models for each location rather than using general LLMs for this task.
Agentic Reasoning (Third-Party LLMs): For its “Agentic” and “Analyst” features, Sona leverages Advanced Reasoning Models. While they do not explicitly name their provider in every document, CTO Ben Dixon has highlighted the breakthrough of 2024/2025 reasoning models (likely GPT-4o, Claude 3.5/3.7, or Llama 3 variants) as the engine that allows their agents to “think” and “act” across different system modules.
Integration Layer: Sona emphasizes an Agentic AI Interface (via APIs). This layer allows the LLMs to interact with their core business logic (payroll rules, compliance checks) securely, ensuring the AI can “take actions” (like publishing a shift) without violating labor laws.

Infrastructure Stack

Open APIs: Sona uses a “headless” approach where its core logic is accessible via APIs, allowing their AI agents to pull data from (and push actions to) third-party systems like Sage, Xero, or Fourth.11
Data Readiness: They emphasize a “single source of truth” architecture.12 By housing HR, Payroll, and Scheduling in one platform, they avoid the “data silo” problem that prevents other AI tools from being truly autonomous.

3. Key Technical Philosophy

Sona’s technical strategy is built on “Interconnected Systems Awareness.” Unlike a chatbot that just summarizes text, Sona’s agents are “aware” of the constraints across the entire stack.13 For example:

If a Scheduling Agent suggests a staff member for an overtime shift, it has already checked the Payroll module (to see if they are too expensive) and the HR module (to ensure their visa/certifications are valid).
Human-in-the-loop: Sona’s current implementation follows an “Approve to Automate” model. The AI makes the decision and provides a “one-click” approval for the manager, with the goal of moving toward full autonomy as trust in the model grows.14