Originally Published on Procurement Magazine By Aaron McMillan on July 04, 2025

Two months after securing Series A funding rounds of US$33.5m, LightSource CEO Spencer Penn is reflecting on a period of rapid transformation. The procurement technology company has doubled its headcount, whilst building credibility with enterprise customers and top-tier investors including Lightspeed and Bain Capital.

But for Spencer, the real excitement lies in what he sees as a "second step change" in AI technology that's moving beyond simple generative outputs to agentic workflows – a shift he believes will fundamentally alter how procurement professionals work.

Talking to Procurement Magazine, the Co-Founder discussed AI, working during such challenging times in global trade and why he believes there are more magicians than wizards.

It has been three months since the funding round, how has LightSource's and its AI strategy evolved since then?

It's been a very exciting three months for us. We went from being a team of less than 20 to being nearly 40 today. We've scaled the organisation and made some critical hires, and having that kind of financial backing brings us tremendous credibility with customers. We’ve gone from being effectively in stealth mode to suddenly having US$33.5m and a substantial valuation—this is one of the largest Series A's of the year at US$130m valuation from amazing investors like Lightspeed and Bain Capital.

I think the AI space has had a second step change since the beginning of the year, particularly everything related to agents and agentic workflows. Generative AI gives you an output from a prompt, but with agentic workflows you input a prompt and get a set of actions or a sequence of actions. We're seeing people move from having LLMs to help them do their work to having small armies of AI interns. The only problem is that if they're left to their own devices, they'll produce rubbish, so they still need substantial human interaction.

When I went to the DPW conference, every single software vendor had the words AI and/or agentic AI on their banner. There's tremendous AI washing in the space now – companies that built infrastructure 30-40 years ago claiming to be AI native which, by definition, can't be true. There's a phrase that I really love: there are a lot of magicians, but very few wizards.

We've been working with larger and larger enterprises on more substantial and thornier problems and we've been very proactive about using AI in two different ways. One is in our own organisation—how do we use it to be more productive ourselves? But also within our product at every single turn—how can we make the product more capable for our users?

Can you explain how LightSource leverages AI and automation to transform procurement from a tactical function into a strategic advantage?

There's this classic chart showing people spend 80% of their time doing tactical work and 20% doing strategic work. The tactical work needs to happen—placing purchase orders, following up with vendors and negotiating pricing. But if you're stuck in the tactical, you can never think about next year or next quarter, which is the strategic level thinking.

You have to eat your vegetables to get dessert, but the dessert is what will lead you to long-term success and is also frankly just a more interesting version of the procurement function. I would not be interested in procurement at all if it was just literally doing human-based robotic process automation, but by hand. Automation allows people to take that 80% of very tactical work that's very repeatable and replace it with technology.

Traditional technologies are actually best at that kind of work. There's a spectrum between agentic workflows, which are much more deterministic and true AI agents. The more deterministic the task—the more transactional and non-strategic—the more you find deterministic systems work well. When it's more strategic, agents can help you be more creative in solving complex tasks.

I've seen overpromise and underdeliver on actual AI products. I call this the AI capability overhang—AI has gotten so much better, but products built using AI are lagging behind the technology's progress. Companies have AI strategies they want to pursue, but the application space is quite limited. I think as technology providers, if we are really ahead of the curve, we have to do a better job of bringing the level of our products up to the true capability of the AI progress that we've seen today. And I don't think it's really happened yet.

What are the biggest challenges organisations face when adopting AI-driven automation in procurement and how does LightSource address these?

You have to start with personal productivity. If people in an organisation can't use AI in their individual tasks, it's very hard to imagine organisation-wide AI adoption. There's a mandate to find AI technologies and make them useful, but IT teams don't always have incentive structures built in to be adventurous. If something goes wrong with AI and you have a data leak, you lose your job. But if you're six months late to the party, you were just being appropriately cautious.

Our responsibility at LightSource is making AI useful and application-oriented. If you tell someone to use ChatGPT to make their job easier, you're asking procurement people to suddenly have this totally new skill: being creative about technology use. That's very different from the actual job itself.

Our job as tool providers is to understand both the technology space and the domain, and marry those together. We don't boil the ocean on all AI capabilities, but focus on real pain points faced by direct materials teams and work backwards to figure out the right AI technologies. If you get ChatGPT or Claude, you need to be basically a technologist to make use of it. With LightSource, you're solving real business problems with the latest technology, but we've built the systems and infrastructure that make it seamless out of the box.

How does LightSource evolve in real time to help companies complete global trade in such a challenging environment?

We always start with solving problems, not bringing technologies in search of problems. Right now, one of the big challenges is the evolving global trade and tariff environment. We recommend a three-step process: identify, quantify and mitigate.

First, identify what components or materials within your spend landscape will be impacted by tariffs. Second, quantify the actual impact—the value at risk. Tariffs might be 30% one day, 150% the next week, then 40% the following week. We encourage people to have a view around risk that includes both expected impact and maximum value at risk.

Third, mitigate—figure out how to improve tariffs' impact on your bottom line through tariff engineering, substantial transformation in different domiciles, reclassifying HTS codes, or changing product specifications. The pace of change is different from previous administrations. People don't have years to prepare—they have weeks, sometimes months.

What this means for our product is that we have to meet the moment. We're lucky we've built a flexible system that can include metadata like HTS codes and tariffs as part of total cost of ownership analysis. A tariff is no different from logistics or surcharges—it's one of hundreds of possible factors for sourcing decisions.

Direct materials are the items most affected but somewhat ignored by many vendors. How does LightSource provide support for this most vital component?

Direct materials typically represent the cost of goods sold—it's in your gross profit versus indirect, which falls below the line as SG&A expense. From a financial perspective, direct materials is one of the earliest things you can improve about overall company profitability.

People focus on indirect because it can be easier and more consistent across companies. Every company buys laptops and office chairs versus not every company buying caustic soda or carbon ceramic brake calipers. Solutions have been built primarily for indirect, then marketed to direct, but they end up being glorified Excel spreadsheets.

Every category past tail spend has a unique standard way people build pricing structures. We've solved this at LightSource by building a quoting engine with Excel's flexibility but ripe with automation. We're really the first non-point solution focused on the full spectrum of direct and strategic indirect.

We break down the world into category content—category-specific metadata, bid sheets, suppliers and analysis. This creates a flexible system segmented into category-specific slices so direct materials teams finally have a tool that helps them do their work versus another version of Excel.

What do you see as the next big opportunities for AI and automation in procurement and how is LightSource preparing to capitalise on these?

There have been two thoughts about AI's future in procurement: replace people or enable people to do more. At LightSource, we focus on augmenting people. Procurement manages such large amounts of money that the salaries are small relative to the value created or lost through performance. The answer isn't to replace procurement, but augment and help them work faster and better.

Over the next 12 months, generative AI is getting better, specifically at context window size. You can now paste entire documents and ask for contract analysis. Through vector databases and RAG, systems can have context windows including your company's full corpus of information.

The second thing I'm watching is everything related to agents and agentic workflows. There's tremendous false information—companies rebranding workflows and RPA as agentic when they're not. What makes something truly agentic isn't a multi-step sequence, but having a meta structure where it doesn't have a defined path, has a goal state, and iteratively takes action, senses outcomes, adjusts expectations and replans until achieving the goal.

We're well positioned because we're AI native, building from the ground up in an AI world. We registered LightSource.ai in 2020 before people knew what that meant. Our team is everything—I came from Tesla and Waymo, Peter led McKinsey's procurement practice for a decade, our CTO Arjun came from Google X and Google Research where the transformer model was created.

We're a technology business first and foremost. We've grown purely through word of mouth and referrals because in the short term, technology adoption is a popularity contest, but in the long term, it's based on actual performance. Companies using LightSource will outperform competition significantly, finding not just better outcomes, but better quality of life for their people.