Provolut
Berlin, DE · Founded 2022 · 8 employees on LinkedIn · 2 known investors
Provolut offers an AI-based web app for protein design that predicts protein evolutions and generates candidate variants from a client's starting dataset, aiming to achieve target protein attributes with far fewer lab experiments than methods like directed evolution or deep mutational scanning. It serves pharma, biotechnology, and research organizations, with on-premises or API deployment options.
Also known as Exazyme · Exazyme GmbH
Founders & leadership
Investors · 2
Funding
SEC filings, press & company announcements- Undisclosed amountSeed RoundSep 2025Source ↗
Source: company announcements and press reports — follow each round's link for the claim.
Company profile
researched Aug 2026Provolut, formerly named Exazyme GmbH, is a Berlin-based biotech software company that develops an AI-based protein design application for biotech, pharma and chemistry organizations. Users upload a spreadsheet of protein sequence variants together with lab measurements of the property they want to optimize; the application first runs a data-sufficiency test to assess whether useful predictions can be derived from the dataset, then generates candidate sequences predicted to improve the targeted properties. Job configurations include random mutations, digital deep mutational scans and fixed candidate lists. Results are taken to the wet lab, and new measurements can be uploaded to iterate further rounds, including data from unsuccessful experiments.
The tool is positioned as a general-purpose optimizer able to address several protein properties simultaneously, including catalytic rate and enzyme activity, stability in solvents and across temperatures, binding affinity, specificity, and other measurable attributes. The company states the app can produce results from as few as roughly 20 data points, and that collaborations require at least 24 sequence variants with corresponding measurements. Deployment options include on-premises installation or connection through the customer's API; the website and app are hosted in Germany in data centers described as DIN ISO 270** certified and SOC2 compliant, and the company signs NDAs with users. Provolut was founded by PhDs in chemistry and AI, building on prior academic research, and was funded by BMWi and the European Social Fund in its Exazyme phase.
Founding story
The company was founded, under the name Exazyme GmbH, by PhDs in chemistry and AI who built the product on years of prior academic research, with the aim of turning a research-grade protein design method into a commercial software tool. Early work was supported by funding from the German BMWi and the European Social Fund.
Business model
Software product sold to biotech, pharma and research organizations as a web application, delivered as a SaaS platform with options for on-premises deployment or API integration. The company also runs scientific collaborations in which partners receive free access to the AI protein tool in exchange for data and feedback, with rights to submitted data and generated sequences remaining fully with the partner.
A one-off flat service fee covering an entire project, independent of the number of protein sequences generated.
Traction
A collaboration with the Department of Biochemistry and Synthetic Metabolism at the Max Planck Institute is presented as a case study: after the institute had tested more than 10,000 (elsewhere stated as 15,000+) enzyme variants with rational design and directed evolution, Provolut's algorithms analyzed a subset of 161 variants and proposed 10 candidates, of which 90% were active and two showed enhanced properties, including a 2.8-fold higher catalytic rate/carbon fixation efficiency and 50-60% lower energy consumption. Provolut reports top-tier rankings at the AlignBio Tournament, states it is used by partners and clients in pharma, biotechnology and research, and has joined the Ginkgo Technology Network. Its software platform progressed from MVP to a production SaaS system and, according to its development partner, contributed to a EUR 2 million seed round.
Latest developments
Provolut announced a partnership with Ginkgo Bioworks, joining the Ginkgo Technology Network to offer AI-driven protein engineering to R&D teams, and reported top-tier rankings at the AlignBio benchmarking tournament. It completed the AI NATION Accelerator program, which co-founder and CSO Jelena Ivanovska credits with helping secure investors for the company's next funding round, and CEO Ingmar Schuster appeared on a Berlin panel on AI in synthetic biology alongside Cambrium and Lunar Ventures.
▸Full profile — market position, technology, go-to-market, geography, history, risks & controversies
Market position
Positions its computational approach against established laboratory methods such as rational design, directed evolution and deep mutational scanning, claiming comparable or better outcomes with roughly 100 times fewer experiments. Ranked in the top tier among more than 30 participating teams at the AlignBio benchmarking tournament for computational protein prediction and design.
Emphasis on working from very small datasets (as few as ~20-24 measured variants), an upfront data sufficiency test performed before payment, multi-property optimization in a single tool, iterative learning from failed experiments, a fixed one-off project price rather than per-sequence pricing, and data-control features including German hosting, on-premises deployment, NDAs and customer retention of rights to data and generated sequences.
Technology
Machine-learning-guided protein regression for "low-N protein engineering": algorithms learn patterns from a client's measured sequence-variant dataset and predict which sequences will improve the targeted protein properties, or assess protein quality. The system evaluates the search space computationally rather than through high-throughput screening, incorporates negative wet-lab results as further training data, and includes a built-in data sufficiency check that flags when input measurements are inadequate. The platform was built as a GDPR-compliant SaaS system with automated CI/CD DevOps pipelines and a QA framework, developed with an external remote engineering team over more than 12 months.
Go-to-market
Direct outreach via demo requests and booked conversations on the company website, a scientific collaboration program offering free tool access to researchers in exchange for joint case studies, published case studies and benchmarking results, a blog and podcast, LinkedIn presence, conference and event participation (e.g. a Berlin panel on AI in synthetic biology), participation in the AI NATION Accelerator, and a partnership placing Provolut in the Ginkgo Technology Network.
Pharmaceutical, biotechnology, chemistry and academic research organizations engaged in protein and enzyme engineering, including labs running directed evolution, rational design or deep mutational scanning programs.
Geography
Headquartered in Berlin, Germany (Max-Urich-Str. 3, 13355 Berlin), with website and application hosting located in Germany; software engineering was carried out by a partner team based in Sri Lanka. The company markets to customers in Europe and internationally.
History
Originally operating as Exazyme GmbH in Berlin, the company later changed its name to Provolut. During the Exazyme phase it engaged an external remote engineering team for more than 12 months to co-define a product roadmap and build a GDPR-compliant SaaS platform from MVP to production. Milestones since include a EUR 2 million seed round from AIX Ventures and Atlantic Labs, top-tier results at the AlignBio Tournament, a Max Planck Institute case study, participation in the AI NATION Accelerator, and a partnership with Ginkgo Bioworks.
Risks & controversies
Performance claims such as "100 times fewer experiments" and the case-study improvement figures originate from the company's own materials; the case-study numbers for variants screened before the collaboration are stated inconsistently across company pages (more than 10,000 versus 15,000+) and the energy-efficiency gain is given as both 50% and 60%. The company depends on customers supplying sufficient quality-controlled measurement data, and outsourced its core software engineering to an external partner.
Compiled by commissioned research from 8 cited public sources — announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed — not independently audited.
Timeline · 6
launches, deals, and filingsProvolut is featured as a participant testimonial of the AI NATION Accelerator, a six-month non-dilutive program for incorporated German AI startups with hubs in Berlin and Munich; co-founder and CSO Jelena Ivanovska states the program helped secure investors for the company's next funding round.
A development partner's case study identifies the company as Exazyme GmbH, now Provolut; the current website's contact forms still reference Exazyme in consent text.
Provolut reported top-tier rankings at the AlignBio benchmarking tournament for computational protein prediction and design, in which more than 30 teams participated.
Company blog announcement of EUR 2 million in funding for AI-supported protein development from AIX Ventures and Atlantic Labs; a development partner's case study also refers to a EUR 2 million seed funding round.
Joint project on the GCC enzyme in which Provolut's machine learning suggested variants that increased carbon fixation rates 2.8-fold and reduced energy demand by 60%.
Provolut partnered with Ginkgo Bioworks and joined the Ginkgo Technology Network to provide AI-driven protein engineering solutions to R&D teams.
Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.
In the news
▸Research sources · 8
primary sources listed
- Provolutprovolut.bio · web
8 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Provolut do?
- Provolut is a Berlin company offering an AI web app that designs improved protein variants from small experimental datasets.
- Who are Provolut's investors?
- Provolut's investors include AIX Ventures, Atlantic.
- Where is Provolut headquartered?
- Provolut is headquartered in Berlin, DE.