Partly: The AI Foundation Model Betting It Can Fix the Auto Parts Supply Chain

Partly is a New Zealand-born AI company that builds Interpreter, which it calls the world's only foundation model purpose-built for automotive parts.

The pitch is simple: repair shops constantly order the wrong parts, and a model trained on the world's parts data can stop that.

In June 2026 the company closed a $50 million Series B at a $500 million valuation, led by DST Global Partners.

It simultaneously moved its headquarters from New Zealand to Austin, Texas to attack the US collision repair market.

Key takeaways

• Real traction in a neglected niche. Partly says thousands of repair businesses use its platform, and it has signed more than fifty manufacturer agreements.
• Serious money, serious backers. Total funding is about $92 million, with DST Global, Blackbird, Octopus Ventures, and Square Peg on the cap table.
• The US is unproven ground. Partly's revenue is undisclosed, its benchmarks are self-published, and it now goes head to head with entrenched giants like CCC and OEConnection on their home turf.

Partly at a glance

Background and origins

Levi Fawcett, a former Rocket Lab engineer who had already founded four companies during his university years, started Partly in Christchurch in 2020.

He co-founded it with Nathan Taylor, Mark Song, Evan Jia, and Tony Austin, the last a former eBay Europe executive who helped launch Amazon in Australia.

The founding insight came from Fawcett's own time around cars: buyers almost never know if a replacement part will actually fit their vehicle.

At the time, roughly 98% of parts orders were handled by phone, with a human "parts interpreter" translating a vehicle description into a part number.

Partly's answer was to digitize that interpreter, which is exactly what it later named its AI model.

Fawcett has since become one of New Zealand's most visible founders, making Forbes 30 Under 30 and the EY Entrepreneur of the Year shortlist.

The wider leadership team now mixes automotive and big-tech alumni, including executives from Airbnb, eBay, Microsoft, and CARIAD, Volkswagen's software arm.

Funding and valuation

Partly's funding story tracks a steady escalation, from a scrappy pandemic-era round to one of the larger AI raises to come out of Australasia.

The Series A was, at the time, New Zealand's largest ever, and the June 2026 Series B nearly tripled the implied valuation.

The pre-seed and seed were both led by Blackbird, with early angels including Rocket Lab founder Peter Beck and Figma co-founder Dylan Field.

The Series A landed in December 2022, led by Octopus Ventures, a deliberate choice as Partly expanded into Europe.

The Series B brings total funding to about $92.4 million, per trade reporting. Partly has not published official cumulative figures.

DST Global's involvement matters for signaling: the firm's portfolio includes Anthropic, Meta, Alibaba, Airbnb, and Spotify.

The round also fits a wider pattern of capital rotating into applied, industry-specific AI, which we mapped in our guide to where deep tech funding is actually going in 2026.

Revenue has never been disclosed. The best public proxy is dated: by late 2022 Partly was reportedly processing around NZ$150 million in orders annually, a platform-volume figure rather than revenue, from a secondary source.

The technology: what Interpreter actually is

Interpreter is not a chatbot wrapper. Partly describes it as a multi-modal model that processes technical diagrams, damage photos, and text to resolve one question: exactly which parts fit this exact vehicle.

It is a textbook example of the specialized, vertical models we covered in The State of AI Infrastructure 2026: narrow domain, proprietary data, purpose-built architecture.

That question is brutally hard. There are thousands of vehicle manufacturers, tens of thousands of part makers, and hundreds of millions of unique vehicle configurations, each cataloged inconsistently.

Given a VIN, Interpreter accounts for trim level, engine type, market, and production date, and returns fitment data plus details most catalogs miss, such as installation hardware, supersessions, and finish specs.

The underlying data assets are Partly's Universal Vehicle Database (UVDB) and Global Auto Parts Catalog (GAPC), which by 2022 already spanned 42 million catalog parts across 4,000 brands.

Training draws on government records, licensed manufacturer data, physical vehicle tear-downs, and five years of human feedback, formalized in more than fifty manufacturer agreements.

Partly's product chief Joachim Ritter, a founding member of VW's CARIAD, frames it as a language problem: systems and manufacturers do not share a common language, and Interpreter is the translation layer.

Products, platform, and pricing

Partly sells the same core data brain through several doors, from raw APIs to finished apps.

The developer surface is public: API documentation covers vehicle lookup, fitment, and parts data.

Pricing is not. There is no published price list; enterprise deals go through contact sales, and the collision-shop product currently sits behind a waitlist.

That opacity is normal for enterprise infrastructure, but it makes independent value-for-money assessment impossible. Treat any cost expectations as negotiable.

Traction, customers, and deployments

Partly claims thousands of businesses across the repair industry use its platform, with customer logos including Toyota, Hyundai, and JLR dealer operations.

The earliest marquee customer was eBay, which used Partly to structure sellers' parts data against a vehicle database.

By 2022 the roster also included the United Nations and unnamed Fortune 500 companies, with most customers in Europe.

In late 2025 Partly pushed into UK dealerships, claiming early adopters processed orders nine times faster. That figure is company-stated and appeared in sponsored content, so weight it accordingly.

The company reports having grown to about 160 employees across more than 20 countries, and plans to more than double headcount from its new Austin base.

The competitive landscape

Partly is entering a US market where parts ordering already runs through a handful of entrenched platforms.

The prize is large: Partly pegs the US collision repair market at more than $100 billion, served by roughly 250,000 repairers. Independent estimates of the sector's annual revenue vary, with some placing collision repair alone above $50 billion.

CCC alone anchors the US collision workflow, running estimating for the largest multi-shop operators and routing insurer claims.

OEC's CollisionLink has over 40,000 active collision repair centers and programs with nearly every automaker selling in the US.

Partly's wedge is that none of these were built as AI-native fitment engines. The incumbents' wedge is distribution: shops already live inside their software all day.

Where Partly wins: data depth on fitment, a genuinely global catalog, and a neutral "infrastructure, not standalone software" posture that lets it integrate rather than replace.

Where it loses: brand recognition in the US, insurer relationships, and the switching inertia of an industry that still faxes.

The big story: the $50M Series B and the move to Austin

On June 23, 2026, Partly announced its DST Global-led $50 million Series B at a $500 million valuation, paired with an immediate US launch.

This was not a satellite office. Fawcett and the core executive team physically relocated to Austin, and the company is hiring across engineering, product, and business development there.

Fawcett's framing was characteristically bold: nothing since the assembly line or EVs, he argued, has improved efficiency, profitability, and consumer value at once.

The skeptical read: $50 million is a modest war chest for cracking a market where CCC generates close to $950 million in annual revenue and serves 18 of the top 20 US auto insurers.

The valuation step-up, roughly 2.8x since late 2022 per trade reporting, also implies investors are pricing in successful US execution that has not happened yet.

Strengths, weaknesses, and risks

Strengths

• A genuinely hard, defensible data problem: fitment across hundreds of millions of vehicle configurations is not something a general-purpose LLM solves out of the box.
• Fifty-plus manufacturer agreements and five years of proprietary training data create a real moat if maintained.
• Blue-chip investor validation from DST Global, Blackbird, Octopus Ventures, and Square Peg, the kind of backers we profiled among the most active deep tech VC firms.
• A team blending automotive veterans (eBay, Solera, CARIAD alumni) with engineers from Rocket Lab, Google, and Microsoft.
• Multiple routes to market: APIs, shop-facing apps, dealership tools, and seller software.

Weaknesses and risks

• No disclosed revenue. Every traction metric is company-stated; there is no public evidence of the business's actual scale or unit economics.
• Self-graded benchmarks. The headline accuracy claims come from Partly's own evaluation, limited to small and medium damage repairs, with no independent replication.
• Incumbent distribution. CCC, OEC, Mitchell, and Solera own the workflows and insurer relationships Partly must integrate with or dislodge.
• Execution concentration. Moving the HQ and executive team to Austin bets the company on US expansion while stretching a 160-person team across 20+ countries.
• Opaque pricing. With no public price list and a waitlisted shop product, buyers cannot easily evaluate cost against incumbents.

Final Notes

Partly is one of the most credible attempts yet to bring purpose-built AI to a huge, stubbornly analog industry.

The data asset is real, the problem is real, and the backers are as serious as they come.

But as of August 2026, the review has to be honest about what is unproven: revenue, independent performance validation, and any track record in the US market it just bet the company on.

If you sell, buy, or catalog parts, Partly is worth a serious evaluation today. If you are watching as an investor or analyst, the next 18 months in Austin will tell you whether the $500 million valuation was foresight or froth.

Frequently asked questions

What does Partly do?

Partly builds AI infrastructure for the automotive repair and parts industry. Its core product is Interpreter, a foundation model that identifies exactly which parts fit a specific vehicle, powering repair estimates, parts ordering, and catalog management. The company also offers APIs, a collision-shop product, and the PartsPal catalog tool.

Who founded Partly and who owns it now?

Partly was founded in Christchurch, New Zealand in 2020 by former Rocket Lab engineer Levi Fawcett, with co-founders Nathan Taylor, Mark Song, Evan Jia, and Tony Austin. It remains a private, venture-backed company, with investors including DST Global Partners, Blackbird, Octopus Ventures, and Square Peg.

How much has Partly raised and what is it worth?

Partly has raised about $92 million in total, per trade reporting. Its June 2026 Series B was a $50 million round led by DST Global Partners at a $500 million valuation. The December 2022 Series A of NZ$37 million was New Zealand's largest at the time.

What is Partly's Interpreter AI model?

Interpreter is a multi-modal AI model purpose-built for automotive parts. It processes technical diagrams, damage photos, and text to determine the exact parts that fit a given VIN, accounting for trim, engine, market, and production date. Partly says it was trained over five years on manufacturer data, vehicle tear-downs, and human feedback.

Who are Partly's main competitors?

Partly's main competitors in US parts and repair workflows are CCC Intelligent Solutions, OEConnection (CollisionLink), Solera/Audatex, Mitchell, and PartsTrader. These incumbents dominate estimating and parts ordering, while Partly positions itself as an AI-native infrastructure layer that can integrate with existing systems.

How much does Partly cost?

Partly does not publish pricing. Enterprise customers go through its sales team, and the collision-shop product at partly.ai is currently waitlisted. Expect negotiated, usage- or seat-based enterprise contracts rather than off-the-shelf plans.

Is Partly legit and any good?

Partly is a legitimate, well-funded company backed by DST Global, Blackbird, and Octopus Ventures, with more than fifty manufacturer agreements and customers across marketplaces, dealerships, and repair shops. The main caveats are that its revenue is undisclosed and its headline accuracy benchmarks are self-published rather than independently verified.

Why did Partly move to Austin, Texas?

Partly moved its headquarters from New Zealand to Austin in June 2026 to attack the US collision repair market, which it values at over $100 billion with roughly 250,000 repairers. The CEO and core executive team relocated, and the company plans to more than double its workforce.