In a bright Madison laboratory, an antibody can look less like a medicine than a logistics problem. Four protein chains must find the right partners. The resulting molecule has to bind two biological targets, survive purification, emerge at a useful yield, remain stable in a vial and, eventually, behave inside a patient. Get one pairing wrong and the elegant diagram becomes expensive soup.
Invenra has spent 15 years trying to make that soup less likely. The privately held Wisconsin biotech discovers and engineers therapeutic antibodies for drug companies, especially bispecifics that bind two targets and trispecifics that bind three. Its pitch is not that clever biology alone wins. A candidate also has to be manufacturable, testable and transferable to a partner without a fresh engineering rescue mission at every step.
That approach finally produced a distinctly non-diagram milestone in May 2025. Exelixis began a first-in-human Phase 1 study of XB628, a bispecific antibody discovered in part through its collaboration with Invenra. The drug engages NKG2A, an inhibitory receptor on natural killer cells, and PD-L1, a familiar immune checkpoint. The study is designed first to test safety and dose in people with recurrent advanced or metastatic solid tumors. It is not evidence that the drug works. It is evidence that a molecule built with Invenra's help crossed the long border between preclinical promise and a human trial.
The product is a well-behaved molecule
Bispecific antibodies are attractive because cancer is rarely polite enough to offer one clean switch. A two-headed molecule can bring an immune cell toward a tumor, block two signals at once or recognize a broader set of diseased cells. But adding specificity adds failure modes. Heavy and light chains can mispair. Expression can disappoint. Purification can demand custom steps. A concentrated formulation may become too viscous for an under-the-skin injection.
Invenra's B-Body platform attacks those problems at the scaffold level. It uses a human IgG-like structure, engineered domains and a heterodimerization scheme to encourage the intended chains to assemble. The company says the result works with conventional Protein A purification, has reached 6 to 11 grams per liter in stable cell-line studies, and combines high solubility with low viscosity. Those are company-reported platform data, not guarantees for every sequence. They matter because a molecule that needs bespoke manufacturing heroics can lose its appeal long before clinical efficacy is known.
Around that scaffold sits a menu. Rapid Bispecific Discovery advertises a four-month route from inputs to 10 B-Body sequences, using more than 30 proprietary human phage libraries, 90 parallel screens and functional testing of more than 250 combinations. A monoclonal campaign advertises 15 sequences in about two months. B-Body Express takes partner-provided sequences and returns purified, quality-checked antibodies at several expression scales in as little as four to eight weeks. The T-Body platform, launched in 2025, adds a third binding function for projects such as T-cell engagers and antibody-drug conjugates.
What broke first
The revealing failure was not an antibody. It was the company's home-grown sequencing process. It had been built for smaller runs. As Invenra's programs grew and new molecular formats arrived, each modality required more custom code. Scientists could inspect only a few dozen sequences at a time, while bioinformaticians bounced between routine analysis and one-off requests. The lab was generating possibilities faster than its software could explain them.
That changed Invenra's mind about where leverage lived. It adopted PipeBio for antibody sequence analysis and connected the results with Benchling as a shared record. Bench scientists could run configurable analyses themselves. Next-generation sequencing expanded the initial field from hundreds toward thousands of sequences; candidates could be ranked for abundance, family diversity and potential liabilities before expensive characterization.
“Our main product at the end of our process is the data.”Sara Counter / Group Leader of Data Strategy, Invenra
The reported outcome is unusually concrete: the cycle from initial selection through characterization and delivery fell from roughly 26 weeks to eight. The company says it analyzed 100 times more sequences and increased project capacity eightfold without adding headcount. Client reports that once took experts across several groups a week or longer to assemble could be generated in minutes from accumulated, structured data.
This is the copyable part. Buying laboratory software did not create the science. Invenra first defined a repeatable funnel, moved quality checks earlier and made the output legible to customers. The tools removed waiting and rework. A sloppy process would merely have become sloppy at higher speed.
What it cost - and who pays
Invenra operates between a contract research organization and a platform biotech. Customers can buy a defined discovery or expression service. They can license the B-Body platform for research and later negotiate commercial rights. Larger partners can fund broad collaborations that include discovery, optimization and access to future programs. Invenra can collect upfront cash and research support today, then milestones and royalties if a partner advances a drug.
The public numbers show the range better than a fictional rate card. The original 2018 Exelixis deal included $2 million upfront, plus $2 million when each discovery project began. The 2021 expansion brought a $15 million upfront payment, additional fees and funding, options for up to 20 more oncology targets, and potential milestones and tiered royalties. Orion's 2025 agreement offered commercial licenses for as many as two antibodies, but disclosed no price.
Standard service pricing remains private. One playful clue surfaced at a 2025 conference promotion: Invenra assigned an estimated retail value of $5,000 to a 50-milliliter B-Body Express run covering up to two Fab arms. That was a giveaway valuation, not a general quote. The sensible read is that expression provides the low-friction entrance; deep discovery and commercial rights are negotiated businesses.
A customer ladder, not one big bet
The customers are biotech and pharmaceutical research teams with a target, an antibody sequence or a biological hunch, but not necessarily the in-house machinery to search thousands of pairings. Exelixis uses Invenra as a source of oncology biologics. Astellas gained research access to B-Body with an option to negotiate licenses. Orion chooses targets and controls development and commercialization after Invenra discovers leads. Catalent paired its SMARTag antibody-drug conjugate expertise with B-Body; Xcellon later added another collaboration around bispecific and trispecific ADCs.
Then, in February 2026, Twist Bioscience became a co-exclusive provider of B-Body discovery services alongside Invenra. That agreement gives the platform a second storefront inside a larger antibody-discovery operation. Twist receives the service and license revenue it generates under the arrangement. For Invenra, distribution can matter as much as another assay.
The playbook you can steal
- Standardize the hard middle, not just the shiny output.
- Offer a small paid entry point with a visible turnaround time.
- Screen failure modes early, while discarding candidates is still cheap.
- Make the data package a product, not an afterthought.
- Let customers graduate from service to license to co-development.
That ladder is Invenra's differentiation. Genmab, Merus, Zymeworks and Xencor offer established multispecific platforms; AbCellera, WuXi Biologics and internal pharma teams provide other routes to discovery and development. Invenra competes by combining its scaffold with discovery libraries, functional screening, expression, data delivery and flexible licensing under one roof. The partner is buying fewer handoffs as much as a molecular format.
Where the formula does not work
A platform cannot rescue bad target biology. It cannot promise that a high-yield molecule will be safe, reach the right tissue or outperform standard care. Some programs need a different geometry, half-life, effector function or manufacturing path. Teams with mature internal antibody groups may prefer to own the stack. And for a narrow experiment, two separate monoclonals may be simpler than a multispecific molecule.
XB628's Phase 1 trial is testing safety, tolerability, dose and preliminary activity. It validates a development handoff, not clinical benefit. Invenra's platform claims still have to be proved molecule by molecule.
The company's own history makes the point. In 2015, Invenra was talking publicly about cell-free production in microscopic samples - one-thousandth the size of competing approaches, according to Green - and a fee-for-service supply model. Over time, its center of gravity moved toward engineered multispecific scaffolds, broader partnerships and internal programs. The through-line was not a single technique. It was deleting expensive, repetitive work from antibody discovery.
There is a human counterweight to all this plumbing. In 2026, Green launched 52 Views of Cancer, a deck of playing cards featuring art from patients, survivors and family members, to be handed out at scientific and medical conferences. He wrote that his mother died from metastatic kidney cancer, his sister-in-law survived breast cancer and his daughter had recently been diagnosed with ovarian cancer. It is difficult to look at a screening matrix quite the same way after that.
Invenra is still a private company with no approved medicine. Its public revenue is estimated, its service prices are mostly negotiated and its clinical evidence is early. But its arc is now visible: shrink the work, standardize the molecule, structure the data, give customers several ways in, and stay around long enough for one candidate to leave the lab. In biotechnology, the waiting is part of the product.
Continue exploring