An integration test is an unlikely place to begin a venture capital story. It has no rocket launch, no laboratory revelation, no founder ringing a bell. It asks whether pieces of software actually work together. Yet that everyday nuisance offers a useful way into IA Seed Ventures, a Berkeley investment firm whose early bets included AtomicJar, the company behind Testcontainers.
- An accelerator’s investor network became a seed fund.
- AtomicJar turned a recurring developer headache into a business Docker acquired.
- The firm now asks AI founders to explain why better models will leave their companies valuable.
There is a certain charm to an investor interested in the plumbing. A demonstration shows what technology can do under chosen conditions. A working business has to keep doing it on an ordinary Tuesday. The distance between those two things is where this firm’s stated interest in resilience becomes more concrete.
A fund with an accelerator’s memory
IA Seed Ventures emerged in 2021 from The Batchery, an accelerator organized around investors who also worked with founders. That origin matters because the scarce resource in an early company is often judgment: someone who can recognize a sales problem disguised as a product problem, or spot a technical dependency before it becomes a hiring emergency.
The connection gave the new firm an existing community from which to draw opportunities and experience. Ed Dua, Leonid Igolnik and Peter Burghardt were among its launch General Partners. Rather than treating the network as an address book, the useful way to understand it is as a collection of people who have already encountered different versions of a founder’s problem.
That is also its proposed distinction in a crowded seed market. Founders can approach angels, accelerator funds or specialist venture firms. IASV’s case rests on combining early capital with people who know operating work. Its history makes that proposition plausible; the practical test for a founder is whether a particular introduction or conversation helps resolve a particular bottleneck.
The test that escaped the laptop
AtomicJar makes the proposition tangible. Testcontainers lets developers start temporary, containerized dependencies for testing: a database, for example, rather than an imitation of one. The point is to discover integration trouble while writing software, before that trouble reaches the people using it.
Docker acquired AtomicJar on December 11, 2023. Its announcement described an open-source framework already used at organizations including DoorDash, Spotify, Uber and Netflix. Those are users of the technology, not customers of IA Seed Ventures. Keeping the relationship straight helps explain what the investor actually backed: a company solving a problem for developers who had other work to finish.
The episode suggests a useful investing instinct: look for repeated friction in an established workflow. Developers did not need persuading that broken integrations were inconvenient. They needed a better way to catch them. For a founder, that is a more useful starting point than the size of the audience that might admire a clever demonstration.
Another early investment, Synthesized, works on a neighboring problem: giving engineering teams realistic test data while protecting sensitive information. Its products generate and prepare data for software testing. The connection is useful. Getting software to behave reliably involves both the systems it talks to and the information it encounters. These are separate businesses addressing work that continues after the launch presentation ends.
What survives a better model?
The language on IASV’s website has moved toward AI. Its launch announcement emphasized fintech, SaaS and crypto/Web3; its current homepage highlights HealthTech, vertical AI and robotics. The change in emphasis is visible. The more revealing detail is how the firm says it evaluates AI companies.
In a public post, IASV names founder expertise, proprietary data or a potential data flywheel, evidence of product-market fit, and continuing value as foundation models improve. That last criterion introduces an uncomfortable question. If next year’s model can do more of the work, what exactly will a customer still need from this company?
“Ability to deliver value, as foundational models improve.”IA Seed Ventures · published AI investment criteria
Consider an illustrative distinction. One application offers convenient access to a model. Another handles a specialized workflow, understands its constraints and learns from repeated use. Both may deliver a good demo. The second has more to explain about its continuing usefulness. This is an interpretation of IASV’s criteria, rather than a claim about any particular portfolio company.
- 01Do you know the domain?
- 02What data advantage can compound?
- 03Where is the evidence of demand?
- 04What improves when models improve?
There is also a human consequence to the AI thesis. IA Seed Ventures participated in Flashpass’s seed financing, a relationship publicly confirmed by lead investor J2 Ventures. Flashpass develops workforce training, connecting skills and certifications with employment needs. That investment broadens the picture: the opportunities surrounding AI include helping people adapt to changes in work, alongside building tools that automate the work itself.
The money behind the introductions
There is a financial business underneath the conversations. IASV manages venture funds, connecting limited partners’ capital with young companies. Its Fund II filing describes management fees and performance fees or allocations. The arrangement gives the firm a different role from a software vendor: it is selecting and supporting investments on behalf of capital providers.
The February 5, 2026 amendment for IASV Seed Ventures Fund II reported $2,731,410 sold toward a $10 million offering, with 17 investors. These are dated fundraising figures. The offering amount describes the target; the amount sold describes the reported progress. Neither number is a startup’s valuation or the firm’s revenue.
For an investor evaluating the firm, portfolio labels deserve the same attention as financial labels. IASV’s website explicitly separates past investments made by its General Partners. A partner’s earlier experience can help explain expertise and access. A fund’s own investments answer a different question about where its capital went. Treating those records separately makes the company’s proposition easier to assess on its actual terms.
A pitch you can use
For a founder considering IASV, the most useful preparation follows directly from that public filter. Show familiarity with the buyer’s work. Explain the painful step your product removes. Bring evidence that somebody wants the result. Then describe what remains distinctive when the next model release makes impressive capabilities cheaper and more common.
The fit becomes harder when the case depends entirely on today’s model limitations, or when enthusiasm outruns evidence of demand. An operator network can sharpen decisions, but it cannot supply customer willingness by itself. AtomicJar gives this story its neatest example: a specific inconvenience, a tool people used, and a buyer that wanted testing closer to development. A small problem on the screen can occupy a large place in the working day.