What we look for at Halfmeyer Ventures

The short answer

We look for four things at once:

• Stage signal: a pre-seed or seed team with a working product, prototype, or validated problem–solution fit (not idea-only for investment). Build method is not the filter (how the product was built).
• Software surface: category-defining technology where software, data, or automation is the core of the product, usually in our focus sectors.
• Operator founders: people who value design craft, engineering rigour, and fast iteration, and who will use a studio partner rather than park a quiet check (founders who get the most from us).
• Compounding fit: a company where our operator network, capital, and GTM playbooks can compound from day one, not only after Series A.

If those four lines match how you build, you are in scope. For the family-office posture, see family office at Halfmeyer. For the broader who-we-back profile (ticket, geography, cold pitch path), see our Berlin pre-seed investor guide. For what a venture studio means in our practice, see what a venture studio is at Halfmeyer Ventures (including studio vs VC).

Stage and proof we need

Investment is for pre-seed and seed. Idea-only decks without a working product, prototype, or validated problem–solution fit are usually a pass for investment (pre-seed vs idea-only). A product built with AI coding assistants or a no-code platform still counts as a working product (how the product was built). Advisory and incubation can start earlier when the engagement model fits; name that model in the deck or link (not raising yet).

Pre-revenue is common at pre-seed when there are early users, pilots, LOIs, or a clear commercial wedge. Paying customers are not a hard filter (pre-revenue; proof that replaces revenue). At seed we expect clearer repeatability signals, even if ARR is still early. If metrics are thin, state what you validated and what this round will prove. Vague “we will figure GTM later” without a wedge usually is a pass.

Solo founders are welcome (solo founders). First-time and repeat founders are both in scope (first-time and repeat). Part-time founders can fit at pre-seed when execution velocity is real and there is a clear post-close full-time transition plan (part-time at pre-seed). Disclose employment, weekly hours, and the go-full-time timeline in the deck.

Software surface (what we underwrite)

Halfmeyer Ventures is a Berlin-based family office and venture studio. At pre-seed and seed we deploy €25,000–€200,000 when the investable product surface is software, data, or automation: the thing that ships, captures work, and sells. Focus sectors are health tech, AI, e-commerce, and automation; adjacent verticals can fit when that same software bar clears (focus vs adjacent; B2B and B2C). Cold pitches are welcome; we respond within 48 hours.

As of 2026, founder-facing software-versus-deep-tech explainers commonly contrast market-risk diligence for software with technical-feasibility diligence for hardware and deep tech, and they often steer capital-intensive builds toward specialty deep-tech funds. That framing leaves vague whether a Berlin family office and venture studio will still underwrite an early software-led deck at a €25k–€200k studio ticket when hardware appears in the stack. Our bar is the software product that owns a buyer workflow or data loop, not a hardware-only or deep-tech bet without that surface. Hardware can sit in the stack when the investable surface is software; it does not replace it. We do not publish robotics-ban lists or deep-tech TRL scorecards.

• Strong fits: software, data, or automation as the core that ships and sells; clear buyer and workflow; a wedge that compounds with design and engineering craft.
• Usually a pass: capital-intensive hardware-only or deep-tech without a software surface, token-first crypto/Web3, pure listing marketplaces without a technology moat, and pure services wrapped as product (clear passes).
• Sector theses: health tech, AI (thin wrappers), e-commerce (software-led), and automation (software-led).

Disclose what the software does, which buyer workflow or data loop it owns, how hardware (if any) serves that surface, how you charge for it (pricing at pre-seed), who the first buyer is (buyer size), traction at your stage (traction by stage), and what stays yours if a large platform ships a shallow version of it (incumbents and the competition question). Want a Berlin studio check into software-led companies without a hardware-only specialty path? Submit your deck (view-only; 48-hour response).

Incumbents and the competition question (what if a large platform builds this)

Halfmeyer Ventures is a Berlin-based family office and venture studio, and at pre-seed and seed we deploy €25,000–€200,000 into software-led teams without requiring a moat that already exists. A proprietary data flywheel, a switching cost measured in months of retraining, or a regulatory approval is not something most companies at this stage have, and we do not read for it. What we read on the competition question is whether you know what your buyer uses today, why they would switch, and what would still be yours in two years if a large platform shipped a shallow version of your product next quarter. We publish no defensibility score, no required moat category, and no replication-time threshold. Cold pitches are welcome; we respond within 48 hours.

As of 2026, the defensibility literature moved quickly and landed above our stage. The common reframing is that AI collapsed the value of things that are hard to do (writing software, maintaining integrations, shipping a clever feature) and left only things that are hard to get: proprietary data produced by how your own product makes people behave, workflow depth and system-of-record position where switching becomes an organisational decision rather than a technical one, regulatory approvals described as twelve to twenty-four month barriers, distribution embedded in procurement, and brand recognised as the category default. The published tests are blunt: if a well-funded competitor could rebuild the product in six months there is no moat, and the asset on the slide should be one a foundation model lab would need more than eighteen months to replicate. Guides written for AI decks in 2026 add a dedicated slide answering why a large model provider will not ship your product, and reject “we are more focused”, “customer intimacy”, and “model-agnostic” as non-answers. A second literature runs alongside it and treats an empty competition slide as fatal: angel-side red-flag rubrics score “we have no competitors” as a pass on its own, one 2026 breakdown of pre-seed pass reasons attributes six percent of passes to that claim, and founders are told to name direct rivals, indirect alternatives, and the status quo. Some 2026 pre-seed deck structures then drop the competition slide altogether and call the stage a bet on founder and thesis.

Every asset on the first list takes elapsed real-world time to accumulate, which is precisely what a company with a prototype and two design partners does not have yet. So a founder is asked to publish a moat they cannot own at this stage, warned that saying nothing about competition is disqualifying, and told in the same breath that the slide is optional. The sharper strand of the 2026 commentary is about mechanism rather than category: the risk is less that a large company copies you and more that a large company can bundle the job into a contract the buyer already pays for, so the thing to avoid is being shaped like a feature of somebody else's product. The same commentary is honest about the other direction, that a platform entering a category often validates it and usually ships a shallow version, which leaves room for depth. None of that says what happens when the deck reaches a €25,000–€200,000 cheque from Berlin.

• The alternative your buyer uses today, named: at pre-seed it is rarely another startup. It is a spreadsheet, an agency, an internal script, a legacy suite the buyer already licenses, or nobody doing the job at all. Say what that costs them now and what happens if they keep it. “No competitors” reads as unfinished research, not as open space.
• A wedge, not a landscape: one segment, one job, one reason to switch this quarter. A feature grid against every tool in the category tells us less than a single honest comparison against the real alternative and the horizontal default. Where the wedge meets market size, we read the arithmetic rather than a headline number (how big the market has to be).
• What compounds while you operate: not what you already own, but what gets harder to copy the longer the product runs. Data your product generates because of how it is designed, an integration into the system of record, an installed workflow, an approval on the roadmap. Naming the loop and the first evidence it is real is enough here; owning it is not a pre-seed requirement (proof that replaces revenue).
• The platform answer, said out loud: if a large platform or model provider could ship a shallow version of what you do, say so and say what survives it. Naming the risk and answering it reads as a founder who has thought about the next two years. Silence reads as one who has not. For AI products this is a different question from prompt depth (thin wrappers) and from what the model layer costs to rent (model provider dependency).
• What shipping first is worth: being first is not a moat and we do not price it as one. Being first with a buyer who kept using it, and being able to say why, is evidence. So is a competitor you lost and understood.

We publish no defensibility scorecard, no required moat category, no minimum switching cost, and no replication-time threshold a product has to clear before a first read. A large incumbent in your category is not a pass here. A product shaped like a feature of somebody else's product usually is, which is the same bar already published for prompt-thin AI wrappers and for pure listing marketplaces without a technology surface (clear passes; marketplaces). One or two slides carry this at pitch stage (minimum deck); the landscape, the churn story, and the switching evidence go deeper inside the lean 1–2 week post-call review (diligence window). Patents, trademarks, and what any of them protect belong with your counsel, not with us. Where a wedge has to become a defensible product, studio design and engineering compound with the check (beyond the check). Want a Berlin studio check that reads your wedge rather than your moat slide? Submit your deck (view-only; 48-hour response).

Pricing at pre-seed (per seat, usage, or outcome)

Pricing is a read, not a requirement. Halfmeyer Ventures is a Berlin-based family office and venture studio, and at pre-seed and seed we deploy €25,000–€200,000 into software-led teams without requiring a particular pricing model. We publish no preferred model, no minimum price point, no minimum contract value, and no date by which a company has to leave per-seat pricing. What we read is whether you can name the unit you charge for, whether the product can count that unit today, and whether you know roughly what one unit costs you to deliver. Cold pitches are welcome; we respond within 48 hours.

As of 2026, pricing is the loudest commercial question in software, and the published answer is written for companies that already have billing history. A monetization survey run in April and May 2026 across roughly 230 software companies put hybrid pricing (a subscription base with consumption charged on top) at 37 percent adoption, up from 25 percent twelve months earlier, and asked participants which model investors now prefer: hybrid came first at 35 percent, then outcome-based at 26 percent and usage-based at 24 percent, against 10 percent for flat fees and 5 percent for per-seat. Industry forecasting from the same period expects roughly 70 percent of software vendors to move off pure per-seat pricing by 2028, and AI credit models were already live at 29 percent of the surveyed companies, with a further 33 percent planning to add them within a year. Read as a scorecard, all of that tells a founder with a prototype and two design partners that the pricing is wrong before the first call.

The same research says something quieter. The companies at our end of the market do the opposite of the chart: businesses below 5 million US dollars in ARR gravitate hardest to flat-fee subscriptions (37 percent), while the ones still holding pure per-seat pricing are the large companies above 150 million US dollars in ARR (29 percent). Nobody in the sample reports being content, and the most frequent complaint is too little expansion revenue. The reference class also revises constantly: a pricing index covering the top 500 software and AI companies logged more than 1,800 pricing and packaging changes during 2025, about 3.6 per company in a single year, with roughly three in four companies changing something in the last twelve months and credit models on that index rising from 35 companies to 79. The model a founder is asked to commit to in a deck is the line the market’s own reference companies rewrite several times a year, with far more evidence than a pre-seed company has.

So we do not read pricing as a commitment. At pre-seed it is the most reversible decision in the deck and one of the cheapest to correct. We read the reasoning and the test underneath it:

• The unit, named: what one paid instance of your product actually is (a seat, a run, a resolved case, a processed document, a booked appointment, a shipment), why a buyer recognises it as the thing they are buying, and whether the product can count it today. A value metric nobody can count is a slide, not a price.
• Where the number came from: not a benchmark table. A conversation where a price was said out loud, a paid pilot, a quote you lost, a renewal you won, or the invoice your buyer already pays for the workaround. “€X per month” with no origin tells us nothing. “Three of five agreed at €X, and the two who declined wanted the other unit” tells us what you know.
• What one unit costs you to serve: at least directionally, and honestly where a model provider meters every request or a person still sits in the loop. The question here is the pricing metric rather than a margin benchmark: what the model layer costs to rent is separate (model provider dependency), and selling an outcome instead of a licence changes the cost structure behind it (services-as-software).
• Who signs, and out of which budget: whether the price fits a budget line that already exists or needs a new one approved, and how long that approval takes. A price invented for a buyer who cannot authorise it is a go-to-market problem wearing a pricing costume. Where the buyer’s calendar owns the answer, say so (procurement cycles).
• What you already changed, and what it taught you: the strongest pricing answer at this stage is a price that has moved once with a reason attached. You raised it and nobody left. You moved from seats to runs because the buyer counts runs. You closed a free tier because it bought no information. One honest iteration carries more than a pricing page that has never been tested.

We publish no preferred pricing model, no price floor, no minimum contract value, no gross-margin threshold, and no deadline for leaving per-seat pricing. Per-seat is not a pass here, and charging nothing yet is not a pass either (pre-revenue). What usually is a pass: a revenue model that only works at a price no buyer has ever been asked to pay, and a founding side that cannot say what one unit of the product costs to deliver. One deck line carries this at pitch stage (minimum deck); pricing history, discounting, and contract shape go deeper inside the lean 1–2 week post-call review (diligence window). What ends up in a customer contract belongs with your counsel, not with us. Where a price has to be tested rather than argued, studio go-to-market compounds with the check (beyond the check). Who the first buyer is (SMB, mid-market, or enterprise) is a different question (buyer size). Want a Berlin studio check that reads your pricing test rather than your pricing page? Submit your deck (view-only; 48-hour response).

Buyer size at pre-seed (SMB, mid-market, or enterprise)

Buyer size is a disclosure, not a filter. Halfmeyer Ventures is a Berlin-based family office and venture studio, and at pre-seed and seed we deploy €25,000–€200,000 into software-led teams whether the first buyer is an SMB, a mid-market operator, or an enterprise committee. We publish no required enterprise logo, no ACV floor, no sales-cycle maximum, and no rule that a company must already sell product-led or sales-led. Cold pitches are welcome; we respond within 48 hours.

As of 2026, founder-facing B2B GTM benchmark guides write the buyer-size question as a Series A screen. One H1 2026 report, last reviewed 1 September 2026, splits the bar by annual contract value: SMB below about $10,000 ACV is asked for CAC payback under 12 months, a 14-to-30-day sales cycle, and a 25-to-35 percent win rate on qualified opportunities; mid-market between about $10,000 and $50,000 ACV is asked for 12-to-18-month payback, a 30-to-90-day cycle, and an 18-to-25 percent win rate; enterprise above about $50,000 ACV is given 18-to-24-month payback, a 6-to-12-month cycle, and a 12-to-18 percent win rate. The same page says a Series A partner in H1 2026 underwrites CAC payback under 18 months, a Magic Number above 0.75, and net revenue retention above 110 percent, against a 2024 KeyBanc/Sapphire median payback of 20 months new-only and 23 months fully loaded. A May 2026 essay on SMB software names the category most institutional investors skip, because a $150-per-month customer is $1,800 ACV and a $5,000 acquisition cost stays underwater for years. That file is written for a company that already has a sales motion, a payback number, and a data room. None of it names a Berlin family office and venture studio that reads a cold deck at €25,000–€200,000 before those bands exist.

So we do not read buyer size as a gate. At pre-seed it is a description of who pays first, not a forecast of who will pay at Series A. We read the motion underneath it:

• Who the first buyer is, named: an owner-operated SMB, a mid-market team, or an enterprise committee, who actually signs, and out of which budget. Buyer type (B2B or B2C) is a different question (B2B and B2C). A named German Mittelstand customer or DAX logo is a different question (Mittelstand).
• What they pay, or would pay: labelled measured versus asked. We publish no ACV floor. A price that only works at an ACV no buyer was asked to pay is the same pass as an untested price (pricing at pre-seed).
• How long the last close actually took: calendar, not credibility. A twelve-month enterprise process is not a pass, the same way a hospital cycle is not a pass (procurement cycles). A two-week SMB close is not a requirement.
• What this round funds about that motion: one buyer size is normal at pre-seed. Part of what a €25,000–€200,000 check buys is making that motion repeatable or finding the second buyer. One design partner carrying the revenue is not a concentration filter (customer concentration).
• The motion, said out loud: product-led, founder-led, or sales-led. We publish no preferred motion. A product-led story with nobody who has used the product, or a sales-led story that only works if a logo that does not exist appears, costs credibility.

We publish no required enterprise logo, no ACV floor, no sales-cycle maximum, no CAC-payback ceiling, no win-rate band, no Magic Number, no net-revenue-retention target, no pipeline-coverage multiple, and no product-led versus sales-led mandate. An SMB first buyer is not a pass here, and an enterprise first buyer is not a requirement. What usually is a pass: a deck that only works if a named enterprise logo that does not exist appears, and a go-to-market that only works at an ACV no buyer has been asked to pay. One deck line carries this at pitch stage (minimum deck); cycle length, discounting, and expansion go deeper inside the lean 1–2 week post-call review (diligence window). Where a first motion has to be tested rather than argued, studio go-to-market compounds with the check (beyond the check). Want a Berlin studio check that reads your first buyer rather than your Series A ACV band? Submit your deck (view-only; 48-hour response).

How the product was built (AI tools, no-code, or custom code)

How your product was built is not a filter on first read. Halfmeyer Ventures is a Berlin-based family office and venture studio, and at pre-seed and seed we deploy €25,000–€200,000 into software-led teams whether the current product came from AI coding assistants, a no-code platform, contractors, or hand-written code. What we underwrite is the software surface that ships and sells, plus a founding side that can explain what the product does and change it. Cold pitches are welcome; we respond within 48 hours.

As of 2026, market guidance reads this question from the wrong end of the funding path. A widely reported 2025 accelerator cohort disclosure put a quarter of one batch at roughly 95 percent AI-generated code, and founder-facing diligence guides now ask AI-built startups for an ownership audit, an architecture diagram, a dependency and security scan, a rebuild plan, and inference cost per active user before a serious institutional conversation. No-code migration guides publish rebuild bands (commonly framed from about €20,000 for a simple application to €80,000 and above for an integration-heavy platform) and treat a stack rented entirely from one closed platform as a later-stage valuation liability. That advice is written for the round after ours. It does not say who reads a cold pre-seed deck while the product still runs on a prototype stack.

• What we read: the product does the job for real users, someone on the founding side can walk the architecture, the data, and the failure modes at the level the stage requires, and code, repositories, platform accounts, and IP sit with the company (who owns the build).
• Stack honesty beats stack purity: name the tooling, the parts you would replace, and what this round changes about build capacity. A deliberate speed decision with a named next step is a strong answer. A rebuild nobody has scoped or funded is a weak one.
• Usually a pass: a demo that only survives a scripted path, nobody who can explain or change what shipped, repositories or platform accounts held by an outside builder, or generated code that no one on the team has read.

Generated code and platform tooling are not a red flag by themselves. A product nobody can explain is. We publish no AI-density threshold, no approved stack list, no rebuild deadline, and no audit or penetration-test requirement before a first read (security and compliance posture). One deck line on how the product is built is enough at this stage (minimum deck), and a short demo link carries more than a stack diagram (demo videos and product access). Product and code questions go deeper inside the lean 1–2 week post-call review (diligence window); transaction-facing reads of AI-assisted codebases sit in our Evaluation practice instead (AI-assisted codebases). Where a validated prototype has to become a production stack, studio design and engineering compound with the check (beyond the check). Want a Berlin studio check that reads the product, not the toolchain? Submit your deck (view-only; 48-hour response).

Security and compliance posture (SOC 2, ISO 27001, questionnaires)

Security and compliance posture is a disclosure question at pitch stage, not a gate. Halfmeyer Ventures is a Berlin-based family office and venture studio, and at pre-seed and seed we deploy €25,000–€200,000 into software-led teams without requiring a SOC 2 report, an ISO 27001 certificate, a penetration test, or a completed security questionnaire before a first check. What we read is which controls exist today, which buyers will ask for more, and who on the team owns the answer. That is a separate question from build method (how the product was built). Cold pitches are welcome; we respond within 48 hours.

As of 2026, the market publishes a stage ladder calibrated to the rounds after ours. Compliance guidance puts pre-seed and seed at near-zero spend and basic hygiene only (multi-factor authentication on production, secrets kept out of the repository, encryption in transit and at rest, a written list of who can reach customer data), a SOC 2 Type 1 around Series A (commonly framed at two to five months and a five-figure audit and platform bill), and Type 2 plus ISO 27001 later, while practitioners argue the real trigger is the first security questionnaire from a buyer you can name in writing rather than a hypothetical future one. The demand side moved faster than that ladder. Since the NIS2 transposition deadline in October 2024, regulated European buyers carry supply-chain security duties toward their suppliers, ISO 27001 is the evidence European procurement recognises most readily, and suppliers into banking, energy, and health report first questionnaires and contractual security clauses landing through 2025 and 2026. None of that says who reads a cold pre-seed deck from a company with no certificate, no penetration test, and no questionnaire history.

• Controls that exist today, named: multi-factor authentication on production and cloud accounts, secrets out of the repository, encryption in transit and at rest, a written list of who can reach customer data, and a current list of the services that process it. This is free groundwork rather than an audit, and naming it carries more than a paragraph about “enterprise-grade security”.
• The buyer demand you have actually seen: whether an enterprise or regulated European buyer sits in the pipeline, whether a questionnaire, procurement portal, or security addendum already landed, and which framework that buyer named (European procurement usually asks for ISO 27001, North American buyers usually ask for SOC 2). One deck line is enough at this stage (minimum deck).
• Incidents, stated: an incident that was detected, contained, and reviewed is an answerable fact and not a pass. An undisclosed one that surfaces later ends the conversation, for the same reason an undisclosed instrument on the cap table does (cap table at first read).
• An owner and a cost: who decides security questions today, what the next step costs, and whether this round funds it. “We take security seriously” with nobody named and nothing scoped reads as nothing.

We publish no certification, penetration-test, questionnaire, or framework requirement, and no date by which one has to exist. A compliance data room is not needed to pitch us (no data room to pitch). A report bought before any buyer asked for it is a runway decision we do not price as credibility, while a stalled deal with a named buyer, a scoped readiness plan, and a line in use of funds is a legitimate ask (state the ask). Health products carry their own data-protection disclosure (health data and GDPR at pitch stage), and AI products carry a regulatory-classification question alongside this one (the EU AI Act at pitch stage). Products that ship a client, firmware, or a connected device carry a separate manufacturer-law question under the Cyber Resilience Act (CRA at pitch stage). Contract clauses, framework scope, and what any of it obliges you to do belong with your counsel, not with us. Security and architecture questions go deeper inside the lean 1–2 week post-call review (diligence window), and studio engineering can compound with the check when hardening is real work (beyond the check). Want a Berlin studio check that reads your controls rather than your certificate wall? Submit your deck (view-only; 48-hour response).

Cyber Resilience Act at pitch stage (SBOM, CE marking, vulnerability reporting)

The Cyber Resilience Act is a disclosure question at pitch stage, not a gate. Halfmeyer Ventures is a Berlin-based family office and venture studio, and at pre-seed and seed we deploy €25,000–€200,000 into software-led teams without requiring a Software Bill of Materials, an EU Declaration of Conformity, a CE mark, a notified-body assessment, or a finished CRA classification memo before a first check. What we read is what actually ships to a user’s machine or device, whether a buyer has already asked about the Act, and who owns vulnerability intake if anything you ship looks like a product with digital elements. That is a different question from organisational certificates such as SOC 2 or ISO 27001 (security and compliance posture). Cold pitches are welcome; we respond within 48 hours.

As of September 2026, the market treats the Act as a raise and procurement gate written for companies already placing products on the EU market. Regulation (EU) 2024/2847 entered into force on 10 December 2024. The European Commission states that reporting obligations apply from 11 September 2026 and that the main obligations, including essential cybersecurity requirements and CE marking, apply from 11 December 2027. On 27 July 2026 the Commission published practical guidance for manufacturers, developers, and businesses of all sizes. Founder-facing German startup briefs describe a 24-hour early warning and a 72-hour full notification for actively exploited vulnerabilities, and they put the top fine band at EUR 15 million or 2.5 percent of global turnover. The same briefs commonly treat hosted software with no client-side binary as outside the Act, and treat a downloadable agent, a mobile or desktop client, firmware, or a connected device as in scope as a manufacturer. Investor-side due diligence notes from 2026 now include CRA readiness next to AI governance, and startup compliance guides tell founders to show a current SBOM, a signed Declaration of Conformity, and a documented vulnerability process before a serious conversation. That file is written for the round after ours, and for a company that is already shipping into the Union market. It does not say who reads a cold pre-seed deck after the reporting clock has started and before the 2027 conformity clock.

So we do not read the Act as a certificate wall. At €25,000–€200,000 we read four facts:

• What ships to the user, named: hosted software only, or a downloadable agent, a mobile or desktop client, firmware, or a connected device. Label it as your current view, not as a legal classification. One deck line is enough at this stage (minimum deck).
• The buyer ask you have actually seen: whether a customer or a procurement team already asked for an SBOM, a CE mark, CRA conformity, or a vulnerability-disclosure contact. “No one has asked” is an answer. Inventing a request is not.
• Who owns intake: if anything you ship looks like a product with digital elements, who would notice an exploited vulnerability and what they would do in the first day. “We will figure CRA later” with a downloadable client already in users’ hands is a weaker answer than “hosted only, no client binary, no buyer has asked.”
• What this round funds: if CRA work is real work, put a line in use of funds (state the ask). A technical file bought before any buyer asked is a runway decision we do not price as credibility.

We publish no SBOM format, no CE-mark deadline, no notified-body route, no classification template, and no rule that a hosted product is in or out of scope. A CRA technical file is not needed to pitch us (no data room to pitch). Scope, manufacturer status, reporting duties, CE marking, and notified-body routes belong with your counsel, not with us. This is a different question from organisational certificates (security and compliance posture), from accessibility rules under the European Accessibility Act (EAA at pitch stage), from AI Act classification (the EU AI Act at pitch stage), and from medical-device rules (MDR and software as a medical device). Product and security questions go deeper inside the lean 1–2 week post-call review (diligence window), and studio engineering can compound with the check when hardening is real work (beyond the check). Want a Berlin studio check that reads what you ship rather than your CE mark? Submit your deck (view-only; 48-hour response).

European Accessibility Act at pitch stage (WCAG, EN 301 549, accessibility statement)

The European Accessibility Act is a disclosure question at pitch stage, not a gate. Halfmeyer Ventures is a Berlin-based family office and venture studio, and at pre-seed and seed we deploy €25,000–€200,000 into software-led teams without requiring a WCAG audit, an EN 301 549 certificate, a published accessibility statement, a VPAT, or a finished EAA scope memo before a first check. What we read is whether the product looks like a consumer-facing service the Act names (for example e-commerce, consumer banking, or e-books), whether a buyer has already asked about accessibility, and what exists versus what this round funds. That is a different question from manufacturer cybersecurity under the Cyber Resilience Act (CRA at pitch stage) and from organisational certificates such as SOC 2 or ISO 27001 (security and compliance posture). Cold pitches are welcome; we respond within 48 hours.

As of September 2026, founder-facing SaaS and compliance guides treat the Act as live enforcement and as a raise gate. Directive (EU) 2019/882 applies from 28 June 2025 to listed products placed on the EU market and to listed services provided to consumers, including e-commerce, consumer banking, e-books, electronic communications, and access to audiovisual media. Germany implements it as the Barrierefreiheitsstärkungsgesetz (BFSG). German practice notes commonly treat purely B2B services as outside the BFSG when they are not directed at consumers, describe a microenterprise exemption for services (fewer than ten people and annual turnover or balance-sheet total of at most EUR 2 million; the exemption does not cover in-scope products), and put the top BFSG fine band at EUR 100,000. Vendor guides then ask for a WCAG 2.1 AA or EN 301 549 audit, an accessibility statement, and a remediation log before a serious conversation, and they warn that hiring a tenth person or crossing EUR 2 million can end the service exemption. That file is written for companies already selling a listed consumer service in the Union. It does not say who reads a cold pre-seed deck while the team is still under ten people and before any audit exists.

So we do not read the Act as a certificate wall. At €25,000–€200,000 we read four facts:

• What you ship, named: a consumer-facing e-commerce, banking, or media service the Act lists, a listed product, or a B2B workflow that is not sold to consumers. Label it as your current view, not as a legal classification. One deck line is enough at this stage (minimum deck).
• The buyer ask you have actually seen: whether a customer or a procurement team already asked for WCAG, EN 301 549, an accessibility statement, or a VPAT. “No one has asked” is an answer. Inventing a request is not.
• What exists versus planned: a statement, an audit, a remediation log, or nothing yet. Do not invent a certificate. “We will figure EAA later” next to a live consumer checkout is a weaker answer than “B2B only, no consumer storefront, no buyer has asked.”
• What this round funds: if accessibility work is real work, put a line in use of funds (state the ask). An audit bought before any buyer asked is a runway decision we do not price as credibility.

We publish no WCAG version, no EN 301 549 certificate, no audit deadline, no in-or-out-of-scope rule, and no rule that a hiring plan or a seed cheque ends an exemption. An accessibility file is not needed to pitch us (no data room to pitch). Scope, B2B versus consumer, the microenterprise exemption, disproportionate-burden claims, and BFSG duties belong with your counsel, not with us. This is a different question from manufacturer cybersecurity (CRA at pitch stage), from organisational certificates (security and compliance posture), from AI Act classification (the EU AI Act at pitch stage), and from medical-device rules (MDR and software as a medical device). Product questions go deeper inside the lean 1–2 week post-call review (diligence window), and studio design can compound with the check when interface work is real (beyond the check). Want a Berlin studio check that reads the product you ship rather than your WCAG certificate? Submit your deck (view-only; 48-hour response).

Founders who get the most from us

Halfmeyer Ventures is a Berlin-based family office and venture studio. At pre-seed and seed we deploy €25,000–€200,000 when founders treat capital and studio operators as one offer: design, engineering, and go-to-market playbooks with the check, not a quiet allocation parked on the cap table. Cold pitches are welcome; we respond within 48 hours.

As of 2026, founder-facing “what VCs look for in a founding team” guides commonly pattern-match on multi-founder composition, technical co-founder presence, and domain pedigree, while operator-versus-VC choosers contrast hands-on help with episodic capital. Those frames leave vague whether a Berlin family office and venture studio wants founders who will use studio craft on product and GTM from day one, versus founders who want capital only and reject operator involvement. Our bar is operator-minded founders who ship with partners. We do not publish a co-founder-count or pedigree SLA (founder pedigree).

• Strong fits: clear buyer and workflow, a sharp wedge, evidence you iterate with users, honesty about what this round must unlock, and openness to design, engineering, and GTM help with the check.
• Weak fits: decks that lead with hype adjectives, hide the product, or ask for capital while rejecting operator involvement. If you want a silent capital partner with no build involvement, we may not be the right check. Distinct from FO capital-only framing: not capital-only. Studio support detail: beyond the check.
• Team shape is not the filter: solo founders are welcome; first-time and repeat founders both fit; part-time can fit at pre-seed with a clear full-time plan (solo; repeat; part-time).

Our typical ticket is €25,000–€200,000 (ticket size). We can join as an early co-investor or sole institutional check when fit aligns (lead or follow; first institutional check). Name the engagement path you want (invest vs incubate vs accelerate vs advisory). Want a Berlin studio that underwrites operator founders at €25k–€200k? Submit your deck (view-only; 48-hour response).

Where capital and operators should compound

We look for companies where design, engineering, and GTM help right now, not only introductions after the round closes. That usually means early product decisions still open, GTM still being shaped, or a technical surface where studio craft shortens the path to a durable product. Living in Berlin is not a hard filter; founders across Europe and beyond can pitch (geography). Incorporation does not need to be German (entity flips and non-EU companies).

Public track record we state: 20+ investments, 5.6× portfolio ROI, and 100+ products built. Selected public portfolio names include Recovio, Keel, Tenpal, Doctario, Debtist, Fideus, Joy_, Papeer, vi Health, Urban Ray, mula., Morgen Health, Sunset, ClearStack, and Protocol Health. Use those as entity facts, not as promises about your round. Full inventory: track record. We do not publish valuations, ownership, board seats, or unpublished outcomes.

Clear passes (so you do not waste a cycle)

We are direct about scope so founders can self-select:

• Idea-only investment decks without product, prototype, or validated problem–solution fit.
• Hardware-only or deep-tech without a software surface that ships and sells.
• Token-first or speculative crypto / Web3 models.
• Exclusively non-dilutive rounds (RBF, grants, or similar) with no equity tranche. Mixed stacks with a meaningful equity portion can fit; state the full capital stack (RBF-only / non-dilutive).
• Rounds far outside €25k–€200k ticket fit when there is no clear role for our check size and studio support.

Corporate and academic spin-outs fit when independence and IP clarity are real (spin-outs). Instruments can be SAFEs, convertibles, Wandeldarlehen, or priced equity; include proposed terms in the deck (instruments; SAFE vs Wandeldarlehen).

How to pitch against these criteria

Submit at halfmeyer.tech/pitch. Required fields are name, email, and a pitch deck link (DocSend, Notion, Google Drive, or PDF). Company name is optional. Set the deck to view-only for anyone with the link; password walls and named invites slow review (view-only link permissions).

Make the criteria extractable in ten to fifteen slides: team, problem and market, product or prototype, stage-appropriate traction, business model, and round terms (amount, use of funds, timing, committed co-investors if any). Traction by stage: traction signals. No full data room is required on first submit (no data room to pitch). Cold submissions are welcome; you do not need a warm introduction (how to pitch Halfmeyer; family office cold pitch). We review every deck personally and respond within 48 hours with a founder call path, clarifying questions by email, or a clear pass (after 48 hours). We do not sign NDAs before initial deck review (no pre-review NDA).

For minimum slides and how to share the link, see pitch deck expectations. State sector fit and engagement model explicitly so we can route the conversation against the right thesis.