Beyond sharing our success story, the symposium unveiled groundbreaking innovations with significant implications for California’s public sector. Here are three key takeaways that every digital leader in government would benefit from understanding.

1. The Launch of SitecoreAI: Agentic Intelligence for the Public Sector

Sitecore unveiled SitecoreAI, a next-generation digital experience platform that positions artificial intelligence at the center of digital marketing and content delivery. Built on the foundation of Sitecore XM Cloud, this composable SaaS platform introduces the Agentic Studio, a collaborative workspace where marketers and AI work together through 20 AI-powered agents that automate workflows from campaign planning to content migration, production, and testing.

Why the California Public Sector Should Care

While the private sector focuses on “marketing,” government agencies confront parallel challenges in constituent communication, public information management, and service delivery. California is already leading the nation in the responsible deployment of AI for government services, as evidenced by Governor Newsom’s GenAI initiatives. SitecoreAI’s agentic intelligence addresses the actual operational needs of government agencies:

Content Operations & Public Information:

• Legislative and policy updates: When regulations change, AI agents can help identify all affected pages across your site, draft updates based on official documents, and flag content requiring human review, transforming a weeks-long manual process into days.
• Emergency content deployment: During wildfires, public health emergencies, or natural disasters, AI agents can rapidly deploy pre-approved content templates, update incident information across multiple pages, and ensure consistent messaging.
• Multilingual content at scale: California serves constituents in dozens of languages. AI agents can accelerate translation workflows while ensuring culturally appropriate terminology, though human review remains essential for accuracy and nuance.
• Content migration: Moving from legacy systems to modern platforms  typically requires months of manual work. AI agents can automate much of this heavy lifting.
• Routine updates: AI can handle repetitive tasks like updating contact information, office hours, or program deadlines across multiple pages.
• Improved search results: AI-powered search that understands intent (e.g., “apply for assistance” vs “eligibility requirements”)

2. Moving Beyond the Website: Content Discovery in the Age of AI Summaries

CEO Eric Stine emphasized a fundamental shift: “We’re living in the world beyond the website. Discovery is no longer driven by search; it’s powered by attention. Brands earn that attention in social media feeds and AI-generated summaries when they show up in the right moment with the right message.”

Californians are increasingly finding government information through AI-powered search summaries, voice assistants, and social media rather than directly navigating to agency websites. This shift has massive implications for how public sector organizations deliver critical information.

At SymSoft, we’re already helping California agencies prepare for this multi-channel reality by building a platform-agnostic, API-first content infrastructure that’s optimized for AI discoverability, while maintaining the accuracy and accountability that government requires.

3. The “Fans First” Philosophy: Eliminating Friction in Government Services

Keynote speaker Jesse Cole of the Savannah Bananas brought his “Fans First” philosophy to the symposium, emphasizing that every moment of friction, from sign-up to service, is an opportunity for your audience to walk away. Sitecore’s product vision centers on eliminating these friction points to build trust and loyalty.

Government services have historically been synonymous with frustration: confusing forms, broken links during emergencies, information buried in PDFs, and websites that don’t work on mobile devices. But it doesn’t have to be this way.

The “Fans First” philosophy translates directly to “Constituents First” for government:

• Every additional click to find wildfire evacuation information is a potential life-safety issue.
• Every confusing form reduces participation in critical benefit programs.
• Every broken link during an emergency erodes trust in government.
• Every accessibility barrier excludes vulnerable Californians who need services most.

Our recent work with Cal FIRE and other state agencies exemplifies this principle. We unify fragmented information across multiple sites, enhance search functionality, incorporate engaging visuals, and optimize an interactive map relied upon by millions of users during critical events to eliminate friction in delivering government services.

The SymSoft’s take

The innovations unveiled at Sitecore Symposium 2025 extend far beyond marketing automation and content management. They represent a fundamental shift in how governments can engage and serve their constituents in a GenAI-first world.

At SymSoft, we’ve spent nearly two decades helping California’s public sector stay ahead of emerging technologies and evolving citizen needs. Today, we’re partnering with agencies such as CDTFA, CAL FIRE, and DWR to deliver essential information through AI-powered summaries and voice assistants, extending service delivery far beyond the traditional website homepage.

Whether your agency is navigating a legacy CMS, planning a major digital modernization, or seeking to understand how Sitecore’s latest innovations can advance your mission, our team is ready to help.

Recently, Governor Newsom announced California’s first-in-nation deployment of Generative AI (GenAI) technologies to improve government efficiency. As part of this effort, SymSoft Solutions is proud to highlight our successful initial implementation of Axyom Assist – our AI-powered assistant for customer service representatives at the California Department of Tax and Fee Administration (CDTFA).

Aligned with the state’s ambitious GenAI initiative, SymSoft is helping California agencies harness generative AI to improve citizen services, support state staff, and enhance operational efficiency.

Transforming Taxpayer Services at CDTFA

In collaboration with CDTFA, SymSoft Solutions has successfully implemented its Axyom Agent Assist solution to empower contact center staff with AI-augmented capabilities. This implementation supports CDTFA’s mission of administering more than 40 tax and fee programs and handling over 800,000 taxpayer inquiries annually.

Axyom Assist works alongside the Customer Service Agents like a knowledgeable teammate. It listens to calls and transcribes conversations in real-time, automatically detecting taxpayer needs while providing relevant information at agents’ fingertips – right when they need it. Beyond this automatic assistance, agents can also proactively ask their own questions during calls to receive instant answers drawn from trusted sources. After calls, the system creates quick conversation summaries so agents can focus on people, not paperwork.

Powered by AWS Bedrock and Anthropic’s Claude large language models, Axyom Assist demonstrates how generative AI is enhancing public sector services—improving accessibility, empowering staff, and streamlining operations.

“As California continues to lead the nation in the responsible implementation of generative AI technologies, SymSoft Solutions is committed to delivering AI solutions that combine powerful capabilities with appropriate safeguards,” said Savita Farooqui, Emerging Technologies Lead and Founder of SymSoft Solutions. “With Axyom Assist, we’re helping government agencies transform operations, empower their teams, and serve citizens with greater efficiency and care.”

“This implementation reflects SymSoft’s broader vision for AI-powered transformation in the public sector,” added Bhavik Patel, CEO of SymSoft Solutions. “We’re proud to work with CDTFA to modernize service delivery while maintaining the highest standards of security, accuracy, and accountability.”

A Broader Vision: Axyom Assist, A Comprehensive GenAI Solution Suite

SymSoft Solutions has expanded its generative AI vision with Axyom Assist — a powerful, flexible product suite designed to help government agencies and regulated industries transform the way they serve the public. More than just an agent support tool, Axyom Assist leverages the power of generative AI and emerging Agentic AI capabilities, to achieve three critical goals: improving the citizen experience, empowering government staff, and streamlining operations.

With Axyom Assist, agencies can create conversational user interfaces that understand context, remember past interactions, and provide accurate, personalized information in real time. By integrating Agentic AI, these systems go a step further — autonomously retrieving and synthesizing information, generating compliant responses, and proactively guiding staff in various scenarios, and through complex regulations. Behind the scenes, Axyom Assist uses generative and Agentic AI to analyze structured and unstructured data, surface insights, automate routine tasks, and help agencies anticipate service needs before they arise.

About SymSoft Solutions and Axyom

SymSoft Solutions is a Sacramento-based firm specializing in digital transformation and user-centric solutions for government agencies. With a focus on innovation, security, and compliance, SymSoft delivers cutting-edge technologies that enhance public sector operations and citizen engagement.

In 2023, SymSoft launched Axyom, a specialized division focused on emerging technologies such as Artificial Intelligence and Distributed Digital Trust.

Axyom Assist, our flagship GenAI product suite, is designed to transform customer service with innovative AI technologies that improve citizen services, support staff, and enhance operational efficiencies across government agencies and regulated industries.

For more information about SymSoft Solutions and our Axyom Assist product suite, visit www.symsoftsolutions.com and axyomassist.com.

Contact Information:

Media Contact: Savita Farooqui
Email: info@symsoftsolutions.com
Phone: (916) 567-1740

SymSoft Solutions is a California Department of General Services (DGS) certified CMAS, TDDC MSA Tier 3, and Small Business (SB) vendor.

In this blog post, I will cover a range of methods by which LLMs and downstream applications can be evaluated. The goal is not to cover specific benchmarks or metrics but to discuss common underlying methods undergirding the benchmarks. Our goal is not so much to draw conclusions as to provide the information needed to make an informed decision.

Human Evaluation

Human evaluation is the method of evaluation where language model output is passed to human evaluators to rate. Essentially, they take a small survey and decide how much they like the output. Sometimes this is based on common sense, such as when rating fluency or determining if text is offensive. Other times a reference answer is passed to the evaluators as well, especially if the evaluator may not know the correct answer.

Human evaluation is often done through crowdsourcing microwork platforms like Amazon’s Mechanical Turk. No special input is required for the machine. When we are measuring we often use Likert-like scales, ask binary questions or ask humans to identify parts of a response.

Pros:

• Highest accuracy: Humans are adept at understanding context and nuances, ensuring alignment with desired objectives.
• Diverse metrics: Human evaluation allows for a wide range of qualitative and quantitative metrics.

Cons:

• Subjectivity: Different evaluators may have varying preferences and opinions, which can affect consistency, particularly in assessing potentially offensive content.
• Cost: Paying human evaluators is significantly more expensive than using automated methods.
• Speed: Manual reviews are slower compared to machine-based evaluations.

Programmatic Evaluation/Unit Testing (HumanEval Benchmark)

For evaluating code, programmatic methods like unit testing check whether the generated program correctly performs specified tasks. The popular HumanEval benchmark runs LLM-generated code within a sandbox with a specified set of unit tests and/or expected correct answers. If the LLM-generated code is correct, we can count this as a win for the LLM. If not, that indicates that more work needs to be done. Usually, for LLM-generated code, code is generated multiple times and the highest scoring code is ultimately chosen.

More broadly, similar approaches can be applied to test AI’s ability to interface with tools or systems. We can create small automated programs that the LLM can interface with and check whether it is able to complete specific tasks.

Pros:

• Naturalistic Setting: Provides a realistic evaluation environment where the AI must perform tasks similar to real-world applications.
• Automation: Allows for fully automated testing, reducing the need for human intervention.
• Task-Specific Efficiency: Works exceptionally well for evaluating specific tasks where outcomes are clear-cut and measurable.

Cons:

• Task Dependency: This method is highly dependent on the type of task and works best for well-defined, specific tasks such as coding.
• Specialized Requirements: Non-coding evaluations may require specialized programs or tools, adding complexity to the evaluation setup.
• Scope Limitations: May not be suitable for more general or open-ended tasks where outputs are not easily validated through unit tests.

ELO System (Chatbot Arena)

Regardless of the testing program, some common flaws remain. Any fixed dataset we choose can be gamed, perhaps not deliberately but implicitly. To account for this, a separate evaluation dataset is ideally held secret and not used until we are ready to ship the product. This is also a problem when it comes to public leaderboards. In addition to explicit training on the test set, public leaderboards allow LLM authors to pick models that seem to perform well on the test set setting up an unfortunate regression to the mean in practice.

Due to the overall weaknesses of each of these evaluation methods, ChatbotArena was created to formalize side-by-side comparison of various chat-based language models. Users of this site pose a question to two anonymized LLMs and then vote on which LLM answered better. This allows for the creation of an ELO score from head-to-head rankings which can then be used to evaluate chatbots in a variety of contexts.

Pros:

• Impossible to Game: People creating LLMs have no idea what users of ChatbotArena will ask and will not be able to game the system.
• Captures subtleties: Subtleties of what make a chatbot “good” may not be accurately captured by any given test but can still be felt. A number of chatbots do well on benchmarks but nevertheless lack creativity or basic appeal in responses.

Cons:

• Lack of Task Specificity: Chatbot Arena users are not evaluating on any specific task. While this can be mitigated with an inhouse system, the default system allows users to pose just about any question which may make it hard to distinguish between LLMs that are each good at one specific task.
• Volunteer Labor / Cost: To evaluate an LLM system, you can rely on the volunteers on ChatbotArena or try to set up your own. In the former case, you will be selecting for a very specialized group of people and in the latter case you will likely need to spend a fair bit of money to overcome network effects.
• Subjectivity: Ultimately, ChatbotArena scores are subjective. While we can hope for a wisdom of crowds, it is important to note that ELO systems exacerbate the subjectivity found in Likert-type scales by diluting the number of participants and allowing for free reign by users. How much this matters is ultimately unknown.

Multiple Choice Questions (MMLU, various)

Another way that LLM based models can be evaluated is by giving them multiple choice tests. The multiple-choice nature of tests makes them easy to grade. Examples of this type of test can be found in a number of BigBench benchmarks and the popular MMLU.

These sorts of questions are easy to grade and allow for automated testing whenever new model builds are created. The realm of automated testing doesn’t just allow for multiple choice questions – other questions with similar definite answers also work well in this regard. The answer can just be evaluated by simple comparison of sampled answers but in the case of multiple-choice questions, the generated next-token vector gives an effective probability of each of the multiple-choice answers. The cross entropy of this probability vector can be compared to the one-shot encoded multiple-choice answer.

Pros:

• Accuracy and speed: Combines the precision of human evaluation with the speed of machine processing.
• Automation: Can be quickly executed with each new model release.

Cons:

• Limited scope: Questions must have definite answers, which may not fully test a model’s general capabilities.
• Non-naturalistic problems: Many questions may not reflect real-world usage scenarios.
• Memorization issues: Models might recognize and reproduce answers from their training data.

NLP similarity metrics

In terms of questions with definite answers that are more complex than multiple choice, n-gram similarity metrics such as BLEU can be employed to tell whether a generated answer has the same words or pairs of adjacent words or triples of adjacent words as a sample answer. While this is sensitive to phrasing, use of n-gram metrics can distinguish between wrong word orders while still giving partial credit. These sorts of similarity metrics – ROUGE and BLEU – are present in a number of BigBench metrics and are often used to evaluate machine translations.

Pros:

• Speed: Allows rapid machine-based testing.
• Natural language compatibility: Suitable for evaluating natural language question/answer pairs.
• Interpretable and deterministic: Produces clear, repeatable results.

Cons:

• Sensitivity to phrasing: Lacks semantic understanding and may penalize correct answers due to synonym usage.
• Potential for rejecting valid answers: May overlook correct responses that are phrased differently from the reference.

Neural Machine Models

BLEURT is a neural model trained to replicate human ratings of text quality. It leverages transfer learning to evaluate outputs on novel datasets, providing a balance between human evaluation and traditional n-gram metrics. Other similar metrics can be created for a number of datasets and have been tried. These models increase complexity compared to n-grams but hopefully can leverage semantic understanding for higher accuracy.

Pros:

• Speed: Faster than human evaluation, allowing for quicker assessments.
• Efficiency: Converts a small amount of human evaluation data into a more robust model that can generalize to new datasets.
• Accuracy: Can potentially achieve high correlation with human judgments by learning from human-rated examples.

Cons:

• Interpretability: The model’s decisions can be opaque, making it hard to understand why a particular rating was given.
• Overfitting: There is a risk that the model might overfit to the training data, reducing its effectiveness on new, unseen data.
• Technical Complexity: Implementing and fine-tuning neural models like BLEURT can be technically challenging and resource-intensive.

LLM Machine Grading (GPT-4/Claude 3)

This approach involves using general-purpose language models like GPT-4 or Claude 3 to evaluate answers against a reference answer through a prompt-based system. Models like GPT4 and Claude 3 have been tested and found to have good agreement with human graders which makes these an appealing option for many use cases. However, if you are using GPT4 or Claude for your task, both are known to prefer their own outputs over others. While this may not be a huge issue, it is likely just the tip of the iceberg of subtle biases given the closed-source and uninterpretable nature of these tools. There is a lot we just don’t know about this method and

Pros:

• Ease of Implementation: APIs for GPT-4 and Claude 3 make it straightforward to set up and use.
• Accuracy: In some test cases, these models have shown superior grading accuracy compared to human evaluators. (For full transparency, this is somewhat speculative and other studies suggest there may not be as close of an alignment with human graders as was previously thought.)
• Speed: Automated grading is much faster than manual human evaluation, enabling quicker turnaround times.

Cons:

• Compounding Errors: Problems where GPT-4 and Claude 3 cannot provide a satisfactory answer can give misleadingly rosy outlooks.
• Lack of Interpretability: The reasoning behind the model’s grading decisions can be unclear.
• Cost: Using advanced models like GPT-4 or Claude 3 can be expensive, particularly at scale.
• Inconsistent Score Ranges: Likert type scores output by GPT-4 do not correlate well with actual performance in terms of measurable metrics (found by looking at texts with spelling mistakes)
• Bias Toward Own Outputs: These models may show a preference for responses similar to their own generated text, which can introduce bias.
• Novelty: These methods are relatively new and may still have undiscovered limitations or require further validation.

Baseline Metrics

This section is intended as a brief catch-all. LLMs applications can also be tested with a number of internal metrics such as perplexity or consistency. These metrics are largely applicable to base models and have some correlation with accuracy as can be seen through decreasing perplexity graphs.

Pros:

• Unlabelled Data: No labelled data is required to determine metrics such as perplexity.

Cons:

• Lack of applicability: While perplexity is an important metric to judge various LLMs, the relationship between perplexity and downstream tasks is far from straightforward.

Conclusion

We hope this blog post has given you some more insight into the world of LLM benchmarking. Each of the methods covered can give some insight into the performance of an LLM or downstream LLM applications. Ideally, by experimenting with different models and datasets using these methodologies, you can draw your own well-informed conclusions.

At SymSoft Solutions, we have successfully implemented Azure AD B2C across multiple projects, leveraging both standard user flows and custom policies to enhance the identity user experience. Azure AD B2C enables seamless integration with external identity providers such as Microsoft, Google, GitHub, and more via Single Sign-On (SSO). Additionally, it serves as a cloud-based user directory, securely storing local user information within the b2c-extension-application. 

Integration with Applications 

Azure AD B2C is easy to integrate with JavaScript-based front-end frameworks using the MSAL (Microsoft Authentication Library). This allows applications to redirect users or present authentication forms in a popup for a smoother login experience. 

At the end of any user flow, Azure AD B2C issues a JWT (JSON Web Token), which serves as an identity token. Additionally, it can provide access tokens for API authentication. These tokens allow web applications to securely interact with backend APIs using MSAL or OpenID Connect (OIDC) authentication in .NET-based applications. 

User Flows and Custom Policies 

Azure AD B2C offers two approaches to defining authentication experiences: 

User Flows (Predefined Policies) 

User flows are built-in, configurable policies that define common authentication scenarios. They allow developers to: 

• Customize branding (company logos, themes, UI styling). 
• Modify authentication attributes collected during sign-up or sign-in. 
• Configure claims returned in the JWT token. 

These predefined flows are ideal for applications that require quick setup and basic identity management without extensive customization. 

Custom Policies (Identity Experience Framework – IEF) 

For more complex authentication requirements, custom policies offer greater control over every step of the identity journey. Unlike user flows, these policies are defined using XML files that refer to each other in a hierarchical structure. 

Custom policies allow: 

• Fine-grained control over authentication logic and user interactions. 
• Integration with legacy identity providers (IDPs) via SAML, OAuth2, and REST APIs. 
• Implementation of advanced security measures, such as risk-based authentication. 

While both user flows and custom policies support attribute customization, session management, and UI customization, custom policies provide significantly more flexibility at the cost of increased complexity. 

A Few Interesting Custom Flows 

Here are a few unique custom flows we implemented in Azure AD B2C to solve specific business challenges: 

Impersonation for Customer Support

Some business scenarios require customer support agents to impersonate users for troubleshooting and better assistance. Since impersonation is not a built-in Azure AD B2C feature, we implemented a custom policy to enable this. 

Implementation: 

• Customer support agents log in via an external IDP (Microsoft). 
• The system verifies their credentials and retrieves their role and permissions from the b2c-extension-app. 
• Agents enter the email address of the customer they want to impersonate. 
• The system fetches customer information from the user directory and grants access via a modified JWT token. 

This solution ensures security, traceability, and compliance while maintaining a seamless support experience. 

Legacy IDP Authentication via REST API

Organizations migrating from a legacy identity provider often need to validate users against the old system before transitioning to Azure AD B2C. 

Implementation: 

• We built a REST API integration with the legacy IDP. 
• During login, Azure AD B2C calls the API to verify legacy user credentials before issuing a token. 
• Once validated, users are migrated to Azure AD B2C seamlessly. 
• Optionally, you can have the users forcefully reset their password after migration to Azure AD B2C to meet the standard password policies. 

This method enables a smooth and secure migration without disrupting user access. 

Custom Email Verification via SendGrid

By default, Azure AD B2C uses Microsoft’s email verification for account sign-ups. However, some businesses prefer using third-party email providers like SendGrid for greater control over branding, analytics, and deliverability. 

Implementation: 

• We replaced the built-in email verification process with a custom SendGrid integration via a RESTful API call. 
• It requires to have a SendGrid account setup in advance along with an API key stored in the Azure Policy Keys which is used to authenticate any HTTP requests made to it. 
• This allowed personalized email templates with the help of dynamic templates and place holder variables. 
• SendGrid supports multiple languages, improves tracking of email delivery and provide analytics regarding the engagement. 

This approach ensures better user engagement and customized user experience through email. 

Advantages and Limitations

Advantages 

• Azure AD B2C is cloud-native solution which makes it highly scalable to handle millions of users while meeting the industry standards of compliance and security. 
• User management and authentication is easier as Azure B2C handles most of it for you and provides you with the options of integrating with identity federations and enterprise providers. 
• Azure AD B2C requires no on-premises identity management solutions and requires minimal setup and development effort as it provides easy to configure authentication flows. 
• Azure AD B2C is cost efficient as you are charged based on Monthly Active Users (MAU) and is easy to integrate with applications developed using the modern technical stack. 

Limitations 

• Azure B2C provides limited customizations within the user flows and restricts the developers to use at most one external API call. 
• Working with custom policies requires deeper understanding of XML configurations, orchestration steps, technical profiles and claim transformations. 
• Debugging any errors within XML or issues during the user journeys is very difficult due to limited and non-descriptive error messages and logging. 
• Documentation from Microsoft do not talk in detail about the advanced and more complex scenarios, making Azure B2C seem more like a black box.

Conclusion 

Azure AD B2C is a powerful, scalable, and secure identity management solution for B2C applications, providing SSO, API security, and seamless integrations with third-party providers. 

For simple authentication needs, user flows are quick and easy to implement. However, for businesses requiring advanced customization, custom policies offer greater flexibility—but at the cost of higher complexity and longer development time. 

By carefully choosing the right approach—user flows for standard implementations and custom policies for unique identity experiences—organizations can maximize the benefits of Azure AD B2C while mitigating its limitations. 

Global Accessibility Awareness Day (GAAD) is an annual event celebrated on the third Thursday of May, dedicated to raising awareness about digital accessibility. It’s a day to reflect on the importance of creating inclusive digital experiences that cater to individuals of all abilities. As we mark GAAD, let’s delve into one of the fundamental aspects of accessibility: color contrast.

Understanding Color Contrast

Color contrast refers to the difference in luminance or color between two elements in a design. It plays a crucial role in ensuring readability and usability for everyone, especially for individuals with visual impairments, color vision deficiencies, or other disabilities. For these individuals, insufficient color contrast can make content difficult or even impossible to perceive.

Impact of Color Contrast

According to the 2024 WebAIM Million report, lack of sufficient color contrast is one of the leading causes of accessibility barriers in digital content. Consider this: for someone with low vision, reading text with poor color contrast can be akin to trying to decipher blurred or indistinguishable characters. Similarly, individuals with color vision deficiencies (color blindeness) may struggle to differentiate between certain colors, further complicating their interaction with digital content.

Tips for Ensuring Adequate Color Contrast

Fortunately, there are simple yet effective measures we can take to address color contrast issues and enhance digital accessibility:

• Use Contrast Checking Tools: There are numerous online tools available that can help assess the color contrast of text and background combinations. Tools like WebAIM’s Contrast Checker or the Color Contrast Analyzer by The Paciello Group can quickly determine if your design meets accessibility standards.
• Follow WCAG Guidelines: The Web Content Accessibility Guidelines (WCAG) provide specific criteria for color contrast ratios that ensure content is perceivable by all users. Aim for a minimum contrast ratio of 4.5:1 for normal text and 3:1 for large text to meet WCAG standards. Use color contrast tools such as those mentioned in this article to ensure text meets the requirement.
• Choose Colors Wisely: Opt for color combinations that provide sufficient contrast. If you struggle in this area, tools like Accessible Color Palette Generator will generate an accessible color palette based on a starting color.

Need More Help?

At SymSoft Solutions, we understand the importance of accessibility and are dedicated to helping businesses and organizations create more accessible digital experiences. Whether you’re seeking guidance on color contrast optimization or comprehensive accessibility audits, full accessibility solutions, or looking to establish an accessibility program in your organization, Symsoft is here to support you every step of the way.

Together, let’s strive to build a digital landscape where everyone can access and engage with content without barriers. Happy Global Accessibility Awareness Day!

Date: May 9, 2024

WHAT YOU NEED TO KNOW: Leading companies will utilize cutting-edge technology for five specific challenges, working with state teams on how GenAI can enhance customer service, improve health care facility inspections for better care, reduce highway congestion, improve safety for vulnerable roadway users, and improve language access.

SACRAMENTO – California is continuing to implement Governor Gavin Newsom’s Executive Order on Generative Artificial Intelligence (GenAI), entering into agreements with five vendors to test, iterate and evaluate GenAI through proofs of concept (POCs). Vendors will utilize GenAI technologies developed by OpenAI, Anthropic, Google, Meta, and others to develop these solutions. The cloud environments used will be Amazon (AWS) Cloud, Google Cloud, Microsoft Azure and ServiceNow (SaaS).

The state will pay each proof-of-concept vendor $1 to test in the “sandbox”, a secure and isolated testing environment. This gives the state and the vendors time to see what works, what doesn’t, and make necessary adjustments during a 6-month window.

“We are now at a point where we can begin understanding if GenAI can provide us with viable solutions while supporting the state workforce,” said Amy Tong, Secretary of the California Government Operations Agency. “Our job is to learn by testing, and we’ll do this by having a human in the loop at every step so that we’re building confidence in this new technology.”

Download the complete GovOps press release for additional information.

SymSoft is thrilled to have been selected for the GenAI POC for the California Department of Tax and Fee Administration (CDTFA). As quoted in the press release:

“SymSoft is enthusiastic about the potential of Generative AI to elevate Customer Service and Customer Engagement,” said Savita Farooqui, Director of Research, SymSoft Solutions. Our approach not only enhances operational efficiency but also enriches the customer experience. By implementing robust AI governance and effective safeguards, we are dedicated to delivering AI solutions that genuinely benefit our customers.”