AI Briefing: Here are two startups that want to help build brands AI agents
In recent years, ad agencies have embraced the AI narrative to demonstrate continued relevancy, as such disruptive technologies threaten their historic value proposition: manpower.
Now, AI startups — helmed by agency alums and former execs from Big Tech — are starting to enter the fray with further competition in the promise of “AI agents,” and some brands are starting to listen.
Last week, a new startup called Palona launched with $10 million in seed funding and plans to build AI agents that help consumer brands with sales efforts, customer support and interactive information.
Palona’s co-founders — former execs from Meta, Google and Samsung — hope to give AI agents higher emotional intelligence while also customizing agents with brand-specific guidelines and safety guardrails. Palona also developed what it describes as an AI “supervisor” to help control hallucinations and make sure conversations stay within a brand’s guidelines. Early customers of Palona include the West Coast pizza chain Pizza My Heart (for ordering pizzas) and the home security firm Wyze (for learning about camera and subscription options).
“Our product objective is gentle persuasion, not aggressive persuasion,” said Palona co-founder and CEO Maria Zhang, speaking of branded AI agents’ conversations with consumers. “We made it very clear, and also all of our customers want gentle persuasion. No one wants that annoying kind of salesperson.”
One early Palona client is MindZero, a South Carolina-based wellness center. It’s recently begun using an AI agent to cut response times from a day down to mere seconds, said MindZero CEO David Semerad. Along with answering customer questions, the AI agent also integrates with MindZero’s booking system. Semerad thinks it’ll be especially helpful as MindZero scales from 1 to 10 locations across the U.S. in the next year.
Palona is just one of several startups that have launched in recent weeks to help brands build AI agents. Another is Anthrologic, founded by former execs at MediaMonks, the next-gen agency helmed by industry veteran Martin Sorrell.
However, rather than just build agents, the startup also offers other data solutions and AI advisory services. One early client is the fashion brand Anthropologie, which is working with the startup to enhance downstream use of data while using AI agents to improve employee workflows.
By focusing on specific projects rather than massive budgets, Anthrologic aims to take a more targeted approach with clients rather than asking clients to cut large checks for everything. Tyler Pietz, Anthrologic’s co-founder, said traditional three-year roadmaps with systems integrators — and the expensive budgets that come with them — don’t work with the current AI landscape.
“Why would you ever throw $350 million or $500 million into a campfire and hope the smoke will make you healthier?” Pietz said. “What’s very clear is that every single week, there is going to be something that changes the entire game. Not always at the highest level, but with a new approach to this or that.”
Working with small teams can help brands navigate the bleeding edge of innovation, said Jon Halvorson, Mondelez’s global SVP of consumer experience and digital commerce. They’re also often less threatening to existing partners than large consultancies that might be competing with agencies for business. However, he noted it doesn’t mean clients should choose to work with just startups.
“Any time you go through these massive change-management projects, you can’t always have all the talent in-house on day one,” Halvorson said. “A lot of the specialized talent in this field doesn’t want to be in your house on day one. So you have a need for fractional expert resources.”
Speed, efficiency and trust are all key when it comes to AI, Halvorson said: “Every single day when the world is changing, what you need is one trusted source to be in your ear.”
Preparing companies for AI agents isn’t always an easy task. Depending on the applications, it could require major overhauls of how companies collect, clean, store and access their data. Other tasks include new considerations for data governance to define who should get access to various data sources.
“The models that come out and people get all jazzed and excited about the use cases,” said Erin Foxworthy, Snowflake’s industry lead for marketing & advertising. “What I think misses a lot in the industry is the data layer. You want to lean into a foundation of the data that is neutral and agnostic so that you’re not tied into something that you don’t know what’s going to change later down the line.”
The week in AI search
It was another week for generative AI search across tech giants, smaller rivals, and even a new startup.
OpenAI previewed its long-awaited “Operator” AI agent tool with research preview partners across e-commerce, travel, news, and — eBay, Etsy, Target, Instacart, StubHub, Booking.com — and some publishers like Thomson Reuters, The Associated Press, and Axios.
Perplexity announced a new integration with Crunchbase to let users search financial data, conduct market research and look for sales leads. Other updates this week include today’s release of a new mobile assistant for Android and Tuesday’s debut of a new search API called Sonar.
Other updates this week came from the browser Brave, which released a new feature called Rerank that lets users personalize their search algorithm to prioritize and deprioritize page results. Meanwhile, Microsoft previewed new AI search features that’s rolling out first to Windows 11 users before later expanding to public availability.
Prompts and Products — Other AI news and announcements
- More brands are bring AI to their Super Bowl ads, with GoDaddy planning to promote its Airo platform and Meta buying an ad for its Ray-Ban Smart glasses.
- LinkedIn became the latest tech giant to face an AI-related lawsuit, after LinkedIn Premium customers alleging the Microsoft subsidiary trained AI models using private messages without consent.
- President Donald Trump reset AI regulations in the U.S, after revoking a 36-page executive order on AI signed by then-President Joe Biden in 2023. In its place, Trump issued a new one-page EO promoting American innovation while urging AI systems to avoid “ideological bias or engineered social agendas.”
- Infinite Reality, an immersive tech company, announced plans to acquire the e-commerce startup Obsess, which helps brands make virtual shopping experiences using AI.
- Mozilla and EleutherAI published new research about using open source datasets to train LLMs. Published in collaboration with more than two dozen researchers, the paper provides best practices, policy recommendations and aims to show how open datasets can make AI models more fair, more transparent and more accountable.
- Anthropic debuted a new feature called Citations, a new API that lets its Claude models ground answers in source documents.
Stories from across Digiday
- WTF are AI agents?
- Agencies have mixed feelings about using AI tools for product placements and influencer marketing
- LADBible Group CEOs plan for growth: £200m, IP, M&A and more
- Assessing the most likely outcomes of Google’s pivotal ad tech antitrust trial
- The TikTok outage caused TikTok Shop sales to spike, not sink
- Marketers cautiously resume TikTok spending after shutdown, while some continue enacting ban measures
- Creators are split on whether to keep using TikTok’s editing app CapCut post-shutdown
More in Media
IAB Tech Lab seeks to step up its standards-setting in 2025 with new tools for CTV, AI and a move away from web browsers
Organization plans to release 31 new specifications or spec updates in 2025, including new signals for ads in live streaming programming and an API for publishers to pass content to LLMs.
What Forbes, Dotdash Meredith, BuzzFeed and other publishers are saying about AI in 2025
Publishing executives from Forbes, Dotdash Meredith, BuzzFeed and other companies detail how they’re using AI in 2025, from how they’re building AI tools and using them internally and externally, to the guardrails they have in place and the future of AI.
How Jessica Chan, Perplexity’s one-person team tasked with building relationships with publishers, gets it done
Jessica Chan is a one-woman team at Perplexity with quite the task at hand: convince publishers that Perplexity has something to offer them, at a time when media companies are increasingly seeking litigation against AI tech companies for copyright infringement.