Over the past decade, bi­o­log­ics have trans­formed the treat­ment of chron­ic dis­eases - from can­cer to meta­bol­ic and au­toim­mune dis­or­ders. Yet iden­ti­fy­ing high-qual­i­ty can­di­date mol­e­cules re­mains slow, re­source-in­ten­sive, and high-risk. Dis­cov­ery teams can spend years cy­cling through screen­ing and op­ti­miza­tion, on­ly for promis­ing leads to fail in pre­clin­i­cal or clin­i­cal stud­ies due to poor de­vel­opa­bil­i­ty, man­u­fac­tura­bil­i­ty, or ef­fi­ca­cy.

In the new era of AI-dri­ven sci­ence, ma­chine learn­ing holds enor­mous promise for re­duc­ing this at­tri­tion by en­abling in sil­i­co de­sign, smarter tar­get en­gage­ment, and rapid ex­plo­ration of nov­el epi­topes. In­dus­try in­vest­ment in large lan­guage mod­els (LLMs) for drug R&D is ac­cel­er­at­ing ac­cord­ing­ly.

But gen­er­al-pur­pose lan­guage mod­els were nev­er built for bi­o­log­ics dis­cov­ery. Most an­ti­body-fo­cused LLMs are trained on lim­it­ed im­mune datasets and lack an un­der­stand­ing of the bio­chem­i­cal, struc­tur­al, and de­vel­opa­bil­i­ty con­straints that gov­ern whether a pro­tein can be­come a vi­able ther­a­peu­tic. As a re­sult, they strug­gle to de­sign mol­e­cules that are nov­el, man­u­fac­turable, sta­ble, and clin­i­cal­ly ac­tion­able.

En­ter EM­Ly Co-pi­lot, an AI-pow­ered plat­form that uni­fies work­flows and de­moc­ra­tizes ac­cess to some of the most pow­er­ful ma­chine learn­ing tools in bi­o­log­ics de­sign.

Found­ed in 2020 in Ox­ford­shire, UK, Etcem­bly is re­defin­ing bi­o­log­ics drug de­vel­op­ment by fus­ing hu­man in­ge­nu­ity with ar­ti­fi­cial in­tel­li­gence. Hav­ing al­ready worked on de­vel­op­ing more than 20 FDA-ap­proved drugs, the found­ing team had seen the same ob­sta­cles re­peat­ed­ly. Bril­liant lab sci­en­tists were con­strained by the lack of in­tu­itive com­pu­ta­tion­al tools, leav­ing vast sci­en­tif­ic po­ten­tial un­tapped and in­no­va­tion slowed to a crawl. Etcem­bly was built to change that.

Around that time, Chat­G­PT  launched, and the team spot­ted a seis­mic op­por­tu­ni­ty: to build a true LLM-pow­ered com­pan­ion for sci­en­tists – one that would uni­fy bioin­for­mat­ics tools and in­cor­po­rate cus­tomized, pro­pri­etary ca­pa­bil­i­ties that out­per­form to­day’s gold-stan­dard work­flows.

The re­sult? EM­Ly Co-pi­lot. A con­ver­sa­tion­al, agen­tic plat­form that trun­cates bi­o­log­ics dis­cov­ery from weeks to min­utes. Sci­en­tists can de­code, pre­dict, and de­sign can­di­dates through a sim­ple chat in­ter­face, while EM­Ly or­ches­trates the heavy lift­ing by in­te­grat­ing mod­els, datasets, and com­pu­ta­tion­al tools to de­liv­er high-im­pact, de­sign-ready out­puts. Pow­ered by state-of-the-art gen­er­a­tive and agen­tic AI, it turns nat­ur­al lan­guage in­to ac­tion­able sci­ence.

The mul­ti-modal plat­form adapts to any ther­a­peu­tic chal­lenge, ac­cel­er­at­ing the de­sign and op­ti­miza­tion of T cell re­cep­tors, an­ti­bod­ies, bi-specifics and VHH frag­ments. Un­der­pinned by a rich im­muno­log­i­cal dataset and ex­per­i­men­tal val­i­da­tion, EM­Ly en­ables drug de­vel­op­ment teams to start smarter and dis­cov­er faster, low­er­ing the bar­ri­er to sci­en­tif­ic in­no­va­tion.

His­tor­i­cal­ly, on­ly large phar­ma­ceu­ti­cal com­pa­nies or well-fund­ed in­sti­tutes have had the re­sources to adopt AI-en­abled bi­o­log­ics dis­cov­ery at scale. EM­Ly Co-pi­lot changes this. With noth­ing more than a con­ver­sa­tion, bench sci­en­tists can now de­sign, op­ti­mize, and trou­bleshoot bi­o­log­ics in re­al time, mak­ing ad­vanced dis­cov­ery ac­ces­si­ble to every­one, from nim­ble seed-stage biotechs to glob­al R&D lead­ers.

The plat­form em­pow­ers teams to ex­plore ideas that once re­quired sub­stan­tial time, cost, and spe­cial­ist ex­per­tise. Its unique in­tel­li­gence helps sci­en­tists de-risk ear­ly ex­plo­ration, iden­ti­fy li­a­bil­i­ties ear­li­er in the process, and avoid cost­ly down­stream fail­ures par­tic­u­lar­ly around de­vel­opa­bil­i­ty.

As Jake Hurst, Etcem­bly co-founder and CTO, ex­plains: “With ac­cu­ra­cy and re­pro­ducibil­i­ty where you need it, and cre­ativ­i­ty where you want it, the on­ly lim­its are cu­rios­i­ty and imag­i­na­tion.”

The unique, nat­ur­al lan­guage-dri­ven in­ter­face en­ables it­er­a­tive think­ing, de­sign branch­ing and ex­plorato­ry rea­son­ing to pro­duce and eval­u­ate vari­ants in min­utes. This en­ables the de­ci­sion-mak­ing flow of sci­en­tists, where hy­pothe­ses evolve rapid­ly and ben­e­fit from im­me­di­ate com­pu­ta­tion­al feed­back.

Michelle Teng, Etcem­bly co-founder and CEO, pre­dicts that EM­Ly will pro­vide “a steady shift from chance to cer­tain­ty,” thanks to its unique op­ti­miza­tion ca­pa­bil­i­ties. Iden­ti­fy­ing a binder is just the first step; EM­Ly com­pre­hen­sive­ly an­a­lyzes both amino acid se­quence and pro­tein struc­ture. In just a few clicks, EM­Ly can eval­u­ate key de­vel­opa­bil­i­ty fac­tors, and en­gi­neer mol­e­cules for im­proved yield, ex­pres­sion po­ten­tial, sta­bil­i­ty, and re­duced ag­gre­ga­tion risk.

The out­come is not just more op­tions, but high­er-qual­i­ty, more de­vel­opable lead can­di­dates. Ad­di­tion­al­ly, with mul­ti-specifics han­dling, it’s eas­i­er than ever to en­gi­neer com­plex, mul­ti-do­main bi­o­log­ic mol­e­cules such as bi- or tri-specifics.

EM­Ly Co-pi­lot con­dens­es time-con­sum­ing work­flows in­to min­utes of in­tu­itive in­ter­ac­tions. De­signed to feel more like a sci­en­tif­i­cal­ly flu­ent lab part­ner than a ma­chine, the plat­form dra­mat­i­cal­ly ac­cel­er­ates ear­ly bi­o­log­ics R&D, cut­ting time­lines by more than half in many cas­es.

Etcem­bly has demon­strat­ed EM­Ly’s re­al world im­pact through a se­ries of in­dus­try col­lab­o­ra­tions. In part­ner­ship with Twist Bio­science, EM­Ly’s codon-op­ti­miza­tion tools achieved a >96% suc­cess rate in im­prov­ing ex­pres­sion yields across di­verse an­ti­body class­es. With Vec­tor Labs, EM­Ly re­solved ma­jor de­vel­opa­bil­i­ty chal­lenges for an an­ti­body can­di­date, en­hanc­ing ex­pres­sion, yield, and ag­gre­ga­tion pro­files in over 80% of en­gi­neered vari­ants - all while pre­serv­ing struc­ture and func­tion. And with­in Etcem­bly’s own pipeline, the plat­form en­abled the de­liv­ery of a high-affin­i­ty T cell en­gager in just six months, out­per­form­ing the in­dus­try’s two-year norm.

Be­yond mol­e­c­u­lar en­gi­neer­ing, EM­Ly is be­com­ing a cor­ner­stone for smarter, faster IP strat­e­gy. Through its in­te­gra­tion with Pat­snap, users can seam­less­ly ac­cess cu­rat­ed patent da­ta, se­quences, and in­tel­li­gence to con­duct nov­el­ty as­sess­ments and Free­dom to Op­er­ate eval­u­a­tions, dri­ving con­fi­dence in promis­ing can­di­dates.

The launch of EM­Ly Co-pi­lot marks the be­gin­ning of a new era in im­munother­a­peu­tic de­sign. Up­com­ing re­leas­es will un­lock full reper­toire analy­sis for rapid lead iden­ti­fi­ca­tion, and tools for be­spoke an­ti­body li­brary de­sign. Soon, EM­Ly will al­so offer in sil­i­co co-com­plex pre­dic­tion, pro­vid­ing un­prece­dent­ed in­sight in­to paratope–epi­tope in­ter­faces and en­abling rou­tine affin­i­ty op­ti­miza­tion.

These ad­vances build on Etcem­bly’s deep sci­en­tif­ic ex­pe­ri­ence and track record of in­dus­tri­al col­lab­o­ra­tions in­clud­ing work with Ab­b­Vie, and un­der­score the com­pa­ny’s com­mit­ment to shap­ing the fu­ture of bi­o­log­ics de­vel­op­ment.

By com­press­ing years of pro­tein en­gi­neer­ing in­to min­utes of guid­ed in­ter­ac­tion, Etcem­bly em­pow­ers sci­en­tists every­where to start smarter, dis­cov­er faster, and cre­ate more ef­fec­tive ther­a­pies at the speed of imag­i­na­tion.

Click im­age to watch the de­mo video in a new tab.

To book a de­mo, or to learn more about how EM­Ly Co-pi­lot can ex­pand your dis­cov­ery ca­pa­bil­i­ties, vis­it: https://bit.ly/EM­Ly­De­mo.

Mer­ck’s part­ner on an ex­per­i­men­tal an­ti­body-drug con­ju­gate has re­port­ed a Phase 3 win in first-line lung can­cer, and will take the da­ta to Chi­nese reg­u­la­to­ry au­thor­i­ties.

Cheng­du-based Kelun-Biotech said that a com­bi­na­tion of its ADC sac­i­tuzum­ab tiru­mote­can, al­so known as sac-TMT, and Mer­ck’s Keytru­da beat Keytru­da alone in pa­tients with first-line, PD-L1-pos­i­tive non-small cell lung can­cer. While Kelun did­n't di­vulge de­tailed da­ta, the com­bi­na­tion hit the study's pri­ma­ry end­point of pro­gres­sion-free sur­vival at an in­ter­im analy­sis.

Re­searchers al­so saw a “pos­i­tive trend” in over­all sur­vival, which will like­ly be fol­lowed un­til the com­ple­tion of the study.

The tri­al bodes well for Mer­ck as it seeks to main­tain a sig­nif­i­cant pres­ence in on­col­o­gy once Keytru­da hits its 2028 patent cliff. The New Jer­sey drug­mak­er has signed sev­er­al li­cens­ing deals, in­clud­ing mul­ti­ple agree­ments with Kelun. In to­tal, Mer­ck could pay Kelun up to $9.3 bil­lion if all pro­grams pan out.