Zip Forward: How AI Agents are Driving Change

Rujul Zaparde, CEO, Zip

Why is it so important that so many people from procurement are here today?

The theme of this year's Zip forward is Agents of Change, and so it's just so incredible to have hundreds and hundreds of customers and community members here with us to really dig into how AI is redefining procurement.

Why is the future of procurement so exciting? 

Procurement for so many years has been bogged down in many ways by working through all of the repeatable, high-throughput manual tasks that everyone in procurement needs to do.

Those tasks really hold people back from doing the more strategic work, driving real value. I think deploying AI agents into the procurement workflow for the first time is really freeing up our customers. And we see this in the ROI. Our customers who are using agents are freed up to go focus on more of the strategic work. The much more repetitive, manual tasks – validating, checking the intake data is correct, comparing contracts and order forms – can, for the first time, be handled by AI, freeing people to work on the more strategic tasks. 

How are you developing this software and where’s it going over the next five years?

When we looked at our data last year, our customers processed 20 million reviews through Zip and 96% of those were manually approved entirely by human beings. 

By 2030, we expect to be processing over a billion reviews a year. The goal we've set for ourselves internally is that 90% of those reviews should be assisted by AI. So, that’s the opportunity we have ahead of us.

Kat Devlin, Head of Procure-to-Pay, Travel & Expense, OpenAI

Tell me about OpenAI. Obviously it's a really exciting time to work in ai, so what is your role there? 

At OpenAI, my focus is really on procurement transformation. So setting up the procurement of the future as part of our finance of the future initiatives. So my goal is to make it easy, faster, and safer to buy things in the company.

Which agents are you finding the most beneficial to work with?

My team is the first reviewer of the procurement process. So, when a requester initiates a request, my team is tasked with reviewing the quality of the data – doing a sanity check to make sure that this makes sense; that these are the right reviewers; that the contract or documentation attached aligns to the needs of the request. That used to take 20-40 minutes to assess. So, the first agent I had them build for us was an intake validation agent that summarises all of that information, provides a recommendation and flags discrepancies across the board. What used to take us 20 minutes is now five minutes of taking the recommendation and making decisions.

Tell us about OpenAI's partnership with Zip

OpenAI is heavily invested in AI technology. When I started there about a year ago, Zip was not yet really embracing the concept of AI to move the needle. It had some things that were automated, but it was not really leveraging AI to help make decisions, to help with those complex pieces of procurement that require judgement. 

They came back to me a few months later with this concept of agents and the ability for us to really reduce the mental load of our end users and help them make better decisions faster by summarising these complex deals, by giving them insights into risk reviews and sharing concepts about the deal that would've been missed in a human review.

Nick Heinzmann, Head of Research, Zip

What has Zip been showcasing at Zip Forward 2025?

We’ve been showcasing a changing procurement function. We see this in our research results – that the skills, operating models and technology priorities for how we build out the future of the function are all shifting rapidly. 

We said on the first day that procurement is at an inflexion point and our own research shows that procurement needs to adapt and help build the bridge to what's next in order to help guide the enterprise towards a more AI-enabled future. 

Will asking agents to carry out negotiations end up hindering relationships with suppliers?

It's actually very interesting when you look at the research on this. There are some aspects of negotiation where machine to machine can be better because machines can be fairer than humans. Often humans try to one-up each other and you will end up with a less equal relationship. Someone is always going to win out. You can get to an optimal outcome faster that's better for both sides. 

Increasingly, machine-to-machine negotiations might be embraced by some people who want to level the playing field, especially when you have a power difference between a large buyer or a large supplier and vice versa. 

That focus on development and relationship management will be more about, ‘how do we help you perform better as a supplier? What do you need? Do you need investment from us? Do you need ideas or access to different people in the business?’ That's a much more productive focus that gets better outcomes for both sides than trying to beat each other down at the negotiation table.

Gabrielle Toof, Director, Procurement and Strategic Sourcing, Circle

Which AI agents are you most likely to work with at Circle? 

Any agents that are really going to support the quality of data. I truly believe – and will scream it from the rooftops – that procurement is a data centre. The data that we collect is fed throughout the organisation, whether that's accounting for their accruals and amortisations of pre-paids, for finance, for budgeting or if it's for the business just to understand their performance. 

All of that is centrally housed within procurement and it's our job to be good data stewards. AI has the opportunity to unlock that and bring it to the next level through really data scrubbing and data validation.

Teddy Shih, Product Marketing Manager, Zip

What’s next for Zip? 

You’ve heard a lot of buzz around AI and Zip is really dedicated to that. I think we mentioned from the outset in our keynote that Zip is the AI leader in procurement orchestration. 

We have agents from beginning to end in the entire process journey to really think about how we can make people's lives better, faster and more efficient. That's where we're headed in terms of product innovation, customer adoption and overall change management to ensure it makes all our lives easier in the future.

How have agents changed how customers approach integration?

With App Studio, customers are now able to integrate much faster to their existing tools, whether that's an ERP, P2P, CLM. We have the connectors coming straight out of the box to be able to plug and play in such a way that agents can access data and processes in Zip, into a different system and back to Zip. 

That’s how we’re thinking about the integration strategy. And, in the future, we're thinking about building a marketplace, an ecosystem where partners and other apps can build on top of it – similar to the App Store on your iPhone, where you can have third-party developers thinking about how their user experience can be made better. 

Meteb Alfayez, Procurement Director, Cribl

Which Zip agents are you finding most beneficial?

The first agent we have is the price negotiation agent, which seems to be a hot product. The second agent is the hotel contract agent. 

We use the price negotiation almost on every software today. It takes data from reliable third-party sources that have done 50-100 negotiations on the same company name, then brings that back, takes our contract data and then checks it against each other and then gives us an average on that price. 

Speaking on ROI, in terms of money we're projecting to see US$3m in savings annually. And in terms of time, we look at it as three hour savings on each request, from saving us from going benchmarking, pinging other third-party tools, networking – all done in 30 seconds within the press of a button. 

Hotel contracts are very complex to read. There's room taxes, food and beverage, all kinds of percentages. The hotel contract agent takes the contract, reads it, gives you a summary and then guides you on how much you should raise the purchase request for. It then helps you understand the contract commercials in a minute or two, including the reading time. 

These are the two most important agents. They’re being used frequently and are highly adopted in our company.

Aside from saving time and money, what other benefits do such agents bring? 

Building these agents might take you time the first time because you want to make sure the data is right and that it's working correctly. But, after you build the agent, it's a matter of pressing a button that will give you the answer. Pressing a button might be difficult for some people, but for us it's been very, very easy – it takes a second. 
When I think about procurement long term, my vision is to think about it as air traffic control in an airport. I want to focus on strategy, creating relationships and being a proactive team. The agents can run the operation, run the motion of that air traffic tower, so I can buy time for my folks. They can focus and spend it with our internal customers. 

I always say, the long-term vision is we don't want to replace people with AI agents. We want to empower them and buy them more time.

Luhua Xu, Sr. Product Marketing Manager, Zip

Tell us more about AI agents and why companies should be using them?

I can give some specific examples of our agents. One that is most welcomed by our customers is the Price Negotiation Agent. This agent is special because it actually gives you hard dollar savings instead of just time savings. Cribl is using our Price Negotiation Agent and seeing 12-15% cost savings per request, which is going to give them probably US$3m in cost savings every year. The agent compares your quotes to your historical invoices and benchmarks and compares it to industry benchmarks as well, so you see the cost savings from each of the requests and actually realise that. It carries a magic power among our customers. Everybody loves it. 

I also love OpenAI, which has been using our risk agents and they give it a nickname called Risk Buddy. It scans the vendor against 12 different dimensions of risk every time. It saves about two to three hours per request and, on aggregate, probably 3,000 hours by the end of year. So, it really is a productivity boost and it gets so much adoption across different teams. Risk Buddy is one of our favourites here.

In addition to saving time and money, what else can be achieved with these agents?

I would say the other big thing we’re addressing is risk. 

A lot of our agents are focused on contract reviews. In the past, people were probably comparing contracts line by line, side by side, with 50 pages of documents. Now, you can actually use our agents to catch risks upfront and in less time. That is really important for compliance, especially at companies like OpenAI, which has a very rigorous due diligence process as you can imagine. So, that agent and that contract review process catches a lot of compliance issues.

What feedback have you been getting on your agents?

I cannot say enough good words about our agents. Whenever I talk to customers one to one or in group settings, people are so excited about our agents because they're seeing real ROIs and they’re asking, “when can we get more agents? What are some other things on the roadmap?” So, you can see that it's actually changing their whole environment. I love this from OpenAI: Kat [Devlin, Head of Procure-to-Pay, Travel & Expense] told me OpenAI is the fastest-growing company in the world and they can’t afford for procurement to slow down. Zip’s AI agents are what’s getting them there.

Julia Li, Product Manager, Zip

Can you tell me what the biggest changes have been to supplier management this year?

Our mission as a team focused on supplier management is just to make it as easy as possible for buyers and suppliers to work together. And as you can imagine, supplier management really happens throughout the entire end-to-end purchasing process. 

The themes that we're really focused on for this upcoming year and that we have been as well for this past year. The first is around making the supplier lifecycle management a lot more compliant and also accurate, up to date. We've done that by actually incorporating more dedicated workflows for supplier onboarding updates and offboarding, being able to automatically trigger those when helpful, being able to have the right approval controls in place and ensure proper ERP sync as well.

Which of Zip’s agents are you particularly excited about?

I think one in particular that people are really excited about is the SOC 2 summary agent. As you may know, a SOC 2 can be 100-plus pages. This AI agent will help summarise all of those SOC 2 findings in an actionable way against SOC 2 trust criteria and any other criteria you might have as a company. 

We also have another agent that is particularly exciting for customers: our adverse media agent. You can monitor different public news sources and other forms of data to check for negative coverage to make sure you’re mitigating reputational risk to your company. 

The last one I'll highlight is our W-8 and W-9 validation agent, which can extract information from a W-8 or W-9 and verify it seamlessly against submitted or collected supplier information. 

You can run all of these agents out of the box, but the beauty is that they're also customisable, so you can also tailor them to your company's use case. 

What, in your eyes, is the future of procurement? 

I'm very excited about AI agents and seeing how we can embed those not just on the buyer side for our customers, but also on the supplier side. 

For example, the W-9 agent – if we were able to incorporate that on the supplier side and check for something like signature completion, it's actually more efficient if you tell a supplier that and have them run it. Then, you’re not getting an incomplete W-9 and having to send it back to the supplier and have them fix it. Doing that on the supplier side is more efficient and we would love to enable that as well.