Chai Discovery growth playbook

Give each page and post its own title, description and share image (for example "Chai Discovery and GSK collaborate after wet-lab evaluation"), add Article schema with dates, and turn on the full sitemap so every post is listed. Make "title, description, image" part of the publishing checklist for every future announcement, starting before JPM.

Each announcement then travels with its own headline in previews, search results and AI answers. It is a one-day change in Framer settings, and it makes the next partnership land harder.

[ ] Unique page title with the news
[ ] Meta description: what, who, why it matters
[ ] Share image with partner or model name
[ ] Article schema with date
[ ] Listed in sitemap
[ ] Linked from Research & News and the homepage
[ ] Preview checked on LinkedIn and X

• Old previews are cached

  Re-scrape the top posts in LinkedIn Post Inspector.

• Your websitePage metadata (checked October 22, 2026): every page tested, including the homepage, Platform, Careers, the $400M Series C post and the GSK, Novartis, argenx and Chai-2 posts, has the same title "Chai Discovery", the same meta description ("Building frontier artificial intelligence to predict and reprogram the interactions between biochemical molecules.") and the same og:title. The sitemap lists 6 URLs (home, product, news, careers, terms, privacy); none of the 11 company posts linked from Research & News (seed, Series A, B and C, introducing Chai-1 and Chai-2, the Chai-2 antibody update, and the Novartis, argenx, BMS and GSK posts) is in it.Source
• HiringBrand and Marketing (posted September 30, 2026): "our first marketing hire", to "refine and establish the Chai brand: How we show up online and off"; audiences "range from pharma executives and scientists to AI researchers and engineers"; "making key moments count at JPM, partner events, and the AI and science conferences"; "building platforms that give our founders and researchers more reach"; "making Chai more legible and compelling to the engineers and researchers we want to hire, including those with no background in biology"; "putting a playbook in place". "This role is not about demand gen or social media maxxing."Source
• Your websiteResearch & News: the GSK collaboration (October 9, 2026) "Following Successful Wet-Lab Evaluation", reported by Endpoints as "its sixth pharma partner in nine months"; Bristol Myers Squibb (August 20), argenx (July 15), the $400M Series C (July 14), Novartis (July 13), Lilly TuneLab (June 18; TuneLab "crosses 100-member mark"), Pfizer (June 4), Eli Lilly (January 9), Series B $130M (December 2025). The homepage partner strip shows Novartis, Pfizer, Eli Lilly and argenx only.Source
• Your websiteGSK post: "Chai's models generated designs to GSK targets, zero-shot without target-specific training. GSK conducted internal wet-lab testing of the designs, finding binders to all tested targets." The Series C post: "$400 million at a $3.8 billion valuation"; "Our latest model, Chai-3, is unlocking targets that have long resisted traditional computational and laboratory methods"; partners "have started adopting our models across their organizations". The Pfizer agreement includes "Chai-3, its most advanced model".Source

Rewrite the Platform page around the current model: what Chai-3 adds, the measured results (success rates, the GSK wet-lab evaluation, time to characterization), formats supported and how to get access. Publish a short Chai-3 post with whatever technical detail is public, and fix the typos.

Pharma scientists checking the platform after a press story should find the model the press described, and accurate copy matters to a science audience.

• Revealing too much

  Share results partners have already made public.

• Your websitePlatform page (/product): headline "De novo protein design that works reliably"; ">10%" success rates for antibodies and ">50%" for miniproteins; characterization "in <2 weeks"; "Learn more about the Chai-2 series of models"; "We are opening up access to Chai-2 to commercial organizations"; academic "limited non-commercial use"; one access form (name, email, organization, type, size, intended use). Chai-3 is not mentioned on the Platform page or the homepage, and no post introduces it. Copy errors on the page include "stuctures", "antigen confirmations" and "such as like".Source
• Your websiteGSK post: "Chai's models generated designs to GSK targets, zero-shot without target-specific training. GSK conducted internal wet-lab testing of the designs, finding binders to all tested targets." The Series C post: "$400 million at a $3.8 billion valuation"; "Our latest model, Chai-3, is unlocking targets that have long resisted traditional computational and laboratory methods"; partners "have started adopting our models across their organizations". The Pfizer agreement includes "Chai-3, its most advanced model".Source
• HiringBrand and Marketing (posted September 30, 2026): "our first marketing hire", to "refine and establish the Chai brand: How we show up online and off"; audiences "range from pharma executives and scientists to AI researchers and engineers"; "making key moments count at JPM, partner events, and the AI and science conferences"; "building platforms that give our founders and researchers more reach"; "making Chai more legible and compelling to the engineers and researchers we want to hire, including those with no background in biology"; "putting a playbook in place". "This role is not about demand gen or social media maxxing."Source

Update the strip to all six with links to each announcement, add one line of proof ("six pharma partners in nine months"), and set a rule that every new partner goes on the homepage the day it is announced.

Momentum is Chai's strongest message to the next partner and to candidates, and the homepage is the first place they look.

• Logo permissions

  Use names in text where logos are not approved.

• Your websiteResearch & News: the GSK collaboration (October 9, 2026) "Following Successful Wet-Lab Evaluation", reported by Endpoints as "its sixth pharma partner in nine months"; Bristol Myers Squibb (August 20), argenx (July 15), the $400M Series C (July 14), Novartis (July 13), Lilly TuneLab (June 18; TuneLab "crosses 100-member mark"), Pfizer (June 4), Eli Lilly (January 9), Series B $130M (December 2025). The homepage partner strip shows Novartis, Pfizer, Eli Lilly and argenx only.Source
• HiringSoftware Engineer, Developer & Research Platform (posted October 12, 2026): ships "the research platform where people and agents launch experiments, compare results, and decide what to try next". Every post describes Chai as "the computer-aided design suite for molecules" with "a computational workspace to iterate on molecules from idea to medicine", used by Eli Lilly, GSK, Pfizer, Bristol Myers, argenx and Novartis.Source

Correct the headline, then add a page or section that explains Chai to an ML engineer: what a binder is, why design beats screening, what the models do, what engineers work on, and quotes from engineers who joined without biology backgrounds.

Candidates apply when they understand the problem, and a clear explainer widens the pool beyond biology.

• Oversimplifying the science

  Have a scientist review it.

• Your websiteHomepage and Careers: the two main homepage statements ("Chai powers drug discovery programs at the world's leading pharmaceutical companies." and the "Our models learn..." paragraph) are split into one element per letter with no accessible label, so text tools read "C h a i p o w e r s ...". The Careers headline reads "Join and work one the most meaningful problems AI." Contact is by email only (info@ and [email protected]).Source
• HiringBrand and Marketing (posted September 30, 2026): "our first marketing hire", to "refine and establish the Chai brand: How we show up online and off"; audiences "range from pharma executives and scientists to AI researchers and engineers"; "making key moments count at JPM, partner events, and the AI and science conferences"; "building platforms that give our founders and researchers more reach"; "making Chai more legible and compelling to the engineers and researchers we want to hire, including those with no background in biology"; "putting a playbook in place". "This role is not about demand gen or social media maxxing."Source
• HiringStrategy (posted September 30, 2026): "Chai's business model has no real precedent"; open questions include "how we price, which segments and products we expand into, and what makes our position durable", plus "scaling lab capacity" and "winning the war for talent". First Finance Hire (August 31): "pharmaceutical partnerships with complex revenue recognition" and "heavy spend on compute and data acquisition".Source

Ask organization type first and route: pharma to a BD call with the evaluation offer, biotechs to the TuneLab path (and Chai's own option), academics to the research access waitlist. Tell each what happens next and when.

Clear routes stop high-value pharma requests from waiting in a general inbox and give smaller groups a fast answer.

• More admin

  Automate the academic and biotech replies.

• Your websitePlatform page (/product): headline "De novo protein design that works reliably"; ">10%" success rates for antibodies and ">50%" for miniproteins; characterization "in <2 weeks"; "Learn more about the Chai-2 series of models"; "We are opening up access to Chai-2 to commercial organizations"; academic "limited non-commercial use"; one access form (name, email, organization, type, size, intended use). Chai-3 is not mentioned on the Platform page or the homepage, and no post introduces it. Copy errors on the page include "stuctures", "antigen confirmations" and "such as like".Source
• Your websiteResearch & News: the GSK collaboration (October 9, 2026) "Following Successful Wet-Lab Evaluation", reported by Endpoints as "its sixth pharma partner in nine months"; Bristol Myers Squibb (August 20), argenx (July 15), the $400M Series C (July 14), Novartis (July 13), Lilly TuneLab (June 18; TuneLab "crosses 100-member mark"), Pfizer (June 4), Eli Lilly (January 9), Series B $130M (December 2025). The homepage partner strip shows Novartis, Pfizer, Eli Lilly and argenx only.Source
• HiringPartnerships founding team (posted March 2026, still open): Business Development to "set the terms and drive deal execution with the world's leading life sciences companies" ("Deal volume and demand are accelerating fast"); Engagement Management to keep partnerships healthy, "accelerating time to clinic", and to synthesize "actionable intelligence for the rest of the company"; Forward-Deployed Scientists to act as "a power user" and "advisor to some of the most important drug programs of our time".Source

Add an accessible label with the full sentence to each animated block (and hide the letter elements from assistive tech), or keep the animation on a visual layer over real text. Test with VoiceOver and a text-only fetch.

The homepage claim then reads as a sentence everywhere, at no cost to the design.

• Animation breaks

  Test on mobile and desktop.

• Your websiteHomepage and Careers: the two main homepage statements ("Chai powers drug discovery programs at the world's leading pharmaceutical companies." and the "Our models learn..." paragraph) are split into one element per letter with no accessible label, so text tools read "C h a i p o w e r s ...". The Careers headline reads "Join and work one the most meaningful problems AI." Contact is by email only (info@ and [email protected]).Source
• Google PageSpeedHomepage: real Chrome visitors pass Core Web Vitals (LCP 2.5 s, INP 138 ms, CLS 0 at the 75th percentile). In the mobile lab test the score is 35 (desktop 33) and the largest element appears at 13.6 s; 1.26 MB of the 1.9 MB page is a 3D molecule viewer library loaded from a public CDN on every page, with 7.6 s of render-blocking requests, 945 KB of unused JavaScript and five accessibility issues.Source

Publish an "evaluation program" page: a fixed scope (a set number of targets chosen by the partner), timelines, what the partner tests in its own lab, how results lead to a collaboration, and the GSK outcome as the example. Use it as the opening offer in every BD conversation.

A standard evaluation lowers the first commitment and turns Chai's results into each buyer's own data.

• Weak results on hard targets

  Agree target selection and success criteria up front.

• Your websiteGSK post: "Chai's models generated designs to GSK targets, zero-shot without target-specific training. GSK conducted internal wet-lab testing of the designs, finding binders to all tested targets." The Series C post: "$400 million at a $3.8 billion valuation"; "Our latest model, Chai-3, is unlocking targets that have long resisted traditional computational and laboratory methods"; partners "have started adopting our models across their organizations". The Pfizer agreement includes "Chai-3, its most advanced model".Source
• HiringPartnerships founding team (posted March 2026, still open): Business Development to "set the terms and drive deal execution with the world's leading life sciences companies" ("Deal volume and demand are accelerating fast"); Engagement Management to keep partnerships healthy, "accelerating time to clinic", and to synthesize "actionable intelligence for the rest of the company"; Forward-Deployed Scientists to act as "a power user" and "advisor to some of the most important drug programs of our time".Source
• Your websiteResearch & News: the GSK collaboration (October 9, 2026) "Following Successful Wet-Lab Evaluation", reported by Endpoints as "its sixth pharma partner in nine months"; Bristol Myers Squibb (August 20), argenx (July 15), the $400M Series C (July 14), Novartis (July 13), Lilly TuneLab (June 18; TuneLab "crosses 100-member mark"), Pfizer (June 4), Eli Lilly (January 9), Series B $130M (December 2025). The homepage partner strip shows Novartis, Pfizer, Eli Lilly and argenx only.Source

Publish a living results page: per model (Chai-1, Chai-2, Chai-3), the benchmarks, wet-lab results, formats and dates, with links to technical reports. Update it with each release.

A single source of measured results is what scientists cite and what AI assistants quote.

• Numbers taken out of context

  State the setup next to each figure.

• Your websiteIntroducing Chai-2 (June 30, 2025): "double digit hit rates in de novo antibody design"; "When testing just 20 designs, the model routinely discovers viable hits"; 52 novel antigens targeted, with successful binders for 50% of tested targets; a technical report PDF.Source
• Your websitePlatform page (/product): headline "De novo protein design that works reliably"; ">10%" success rates for antibodies and ">50%" for miniproteins; characterization "in <2 weeks"; "Learn more about the Chai-2 series of models"; "We are opening up access to Chai-2 to commercial organizations"; academic "limited non-commercial use"; one access form (name, email, organization, type, size, intended use). Chai-3 is not mentioned on the Platform page or the homepage, and no post introduces it. Copy errors on the page include "stuctures", "antigen confirmations" and "such as like".Source
• Your websiteGSK post: "Chai's models generated designs to GSK targets, zero-shot without target-specific training. GSK conducted internal wet-lab testing of the designs, finding binders to all tested targets." The Series C post: "$400 million at a $3.8 billion valuation"; "Our latest model, Chai-3, is unlocking targets that have long resisted traditional computational and laboratory methods"; partners "have started adopting our models across their organizations". The Pfizer agreement includes "Chai-3, its most advanced model".Source
• GitHubchai-lab, the open-source Chai-1 repository ("SOTA model for biomolecular structure prediction", Apache-2.0), has 2,004 stars and 295 forks; last push June 30, 2026. The Chai-1 launch reached 270 points on Hacker News.Source

Plan JPM week now: a small technical briefing for heads of discovery, two partner dinners, founder meetings booked ahead with target pharma companies, and one announcement timed for the week with its own page and share image.

The right executives are in one city for a few days, and a planned week turns that into pipeline.

• Calendar crowded

  Book meetings weeks ahead.

• HiringBrand and Marketing (posted September 30, 2026): "our first marketing hire", to "refine and establish the Chai brand: How we show up online and off"; audiences "range from pharma executives and scientists to AI researchers and engineers"; "making key moments count at JPM, partner events, and the AI and science conferences"; "building platforms that give our founders and researchers more reach"; "making Chai more legible and compelling to the engineers and researchers we want to hire, including those with no background in biology"; "putting a playbook in place". "This role is not about demand gen or social media maxxing."Source
• Your websiteResearch & News: the GSK collaboration (October 9, 2026) "Following Successful Wet-Lab Evaluation", reported by Endpoints as "its sixth pharma partner in nine months"; Bristol Myers Squibb (August 20), argenx (July 15), the $400M Series C (July 14), Novartis (July 13), Lilly TuneLab (June 18; TuneLab "crosses 100-member mark"), Pfizer (June 4), Eli Lilly (January 9), Series B $130M (December 2025). The homepage partner strip shows Novartis, Pfizer, Eli Lilly and argenx only.Source
• HiringPartnerships founding team (posted March 2026, still open): Business Development to "set the terms and drive deal execution with the world's leading life sciences companies" ("Deal volume and demand are accelerating fast"); Engagement Management to keep partnerships healthy, "accelerating time to clinic", and to synthesize "actionable intelligence for the rest of the company"; Forward-Deployed Scientists to act as "a power user" and "advisor to some of the most important drug programs of our time".Source

With partner approval, publish short stories or co-authored talks: the problem, how the team used Chai, what changed in their workflow. Start with partners that already speak publicly (Lilly TuneLab, GSK's evaluation quotes).

Peer scientists trust peer stories more than company claims, and partners gain visibility for their own teams.

• Pharma confidentiality

  Focus on workflow, not targets.

• Your websiteResearch & News: the GSK collaboration (October 9, 2026) "Following Successful Wet-Lab Evaluation", reported by Endpoints as "its sixth pharma partner in nine months"; Bristol Myers Squibb (August 20), argenx (July 15), the $400M Series C (July 14), Novartis (July 13), Lilly TuneLab (June 18; TuneLab "crosses 100-member mark"), Pfizer (June 4), Eli Lilly (January 9), Series B $130M (December 2025). The homepage partner strip shows Novartis, Pfizer, Eli Lilly and argenx only.Source
• Your websiteGSK post: "Chai's models generated designs to GSK targets, zero-shot without target-specific training. GSK conducted internal wet-lab testing of the designs, finding binders to all tested targets." The Series C post: "$400 million at a $3.8 billion valuation"; "Our latest model, Chai-3, is unlocking targets that have long resisted traditional computational and laboratory methods"; partners "have started adopting our models across their organizations". The Pfizer agreement includes "Chai-3, its most advanced model".Source
• HiringPartnerships founding team (posted March 2026, still open): Business Development to "set the terms and drive deal execution with the world's leading life sciences companies" ("Deal volume and demand are accelerating fast"); Engagement Management to keep partnerships healthy, "accelerating time to clinic", and to synthesize "actionable intelligence for the rest of the company"; Forward-Deployed Scientists to act as "a power user" and "advisor to some of the most important drug programs of our time".Source

Publish one technical post a month by a researcher (methods, lessons, failures worth sharing), give each a talk version for AI and science conferences, and track speaking slots and citations.

Researchers are the brand for both scientists and recruits, and regular technical writing compounds.

• Research time lost

  Keep posts short and edited by marketing.

• HiringBrand and Marketing (posted September 30, 2026): "our first marketing hire", to "refine and establish the Chai brand: How we show up online and off"; audiences "range from pharma executives and scientists to AI researchers and engineers"; "making key moments count at JPM, partner events, and the AI and science conferences"; "building platforms that give our founders and researchers more reach"; "making Chai more legible and compelling to the engineers and researchers we want to hire, including those with no background in biology"; "putting a playbook in place". "This role is not about demand gen or social media maxxing."Source
• Your websiteResearch & News: the GSK collaboration (October 9, 2026) "Following Successful Wet-Lab Evaluation", reported by Endpoints as "its sixth pharma partner in nine months"; Bristol Myers Squibb (August 20), argenx (July 15), the $400M Series C (July 14), Novartis (July 13), Lilly TuneLab (June 18; TuneLab "crosses 100-member mark"), Pfizer (June 4), Eli Lilly (January 9), Series B $130M (December 2025). The homepage partner strip shows Novartis, Pfizer, Eli Lilly and argenx only.Source
• Your websiteIntroducing Chai-2 (June 30, 2025): "double digit hit rates in de novo antibody design"; "When testing just 20 designs, the model routinely discovers viable hits"; 52 novel antigens targeted, with successful binders for 50% of tested targets; a technical report PDF.Source

Publish a TuneLab page: what biotechs can try, how to get the most from it, and how to work with Chai directly as they grow. Track which TuneLab users ask for direct access.

A partner channel already reaches more than 100 biotechs; a clear next step turns users into customers.

• Partner conflict

  Keep Lilly informed and aligned.

• Your websiteResearch & News: the GSK collaboration (October 9, 2026) "Following Successful Wet-Lab Evaluation", reported by Endpoints as "its sixth pharma partner in nine months"; Bristol Myers Squibb (August 20), argenx (July 15), the $400M Series C (July 14), Novartis (July 13), Lilly TuneLab (June 18; TuneLab "crosses 100-member mark"), Pfizer (June 4), Eli Lilly (January 9), Series B $130M (December 2025). The homepage partner strip shows Novartis, Pfizer, Eli Lilly and argenx only.Source
• Your websitePlatform page (/product): headline "De novo protein design that works reliably"; ">10%" success rates for antibodies and ">50%" for miniproteins; characterization "in <2 weeks"; "Learn more about the Chai-2 series of models"; "We are opening up access to Chai-2 to commercial organizations"; academic "limited non-commercial use"; one access form (name, email, organization, type, size, intended use). Chai-3 is not mentioned on the Platform page or the homepage, and no post introduces it. Copy errors on the page include "stuctures", "antigen confirmations" and "such as like".Source

Link the repository to the Platform page and the results page, add a "what's next after Chai-1" section, publish release notes, and invite active users to research access and office hours.

Open-source users are already qualified and can become champions inside their companies.

• Support load

  Keep office hours and issues triaged.

• GitHubchai-lab, the open-source Chai-1 repository ("SOTA model for biomolecular structure prediction", Apache-2.0), has 2,004 stars and 295 forks; last push June 30, 2026. The Chai-1 launch reached 270 points on Hacker News.Source
• Your websitePlatform page (/product): headline "De novo protein design that works reliably"; ">10%" success rates for antibodies and ">50%" for miniproteins; characterization "in <2 weeks"; "Learn more about the Chai-2 series of models"; "We are opening up access to Chai-2 to commercial organizations"; academic "limited non-commercial use"; one access form (name, email, organization, type, size, intended use). Chai-3 is not mentioned on the Platform page or the homepage, and no post introduces it. Copy errors on the page include "stuctures", "antigen confirmations" and "such as like".Source

Launch a selective academic program: a set number of labs per quarter, a short proposal, credits and a request to cite and share results. Highlight published work on the site.

Academic papers build scientific credibility with pharma and attract research talent.

• Compute cost

  Cap credits per lab.

• Your websitePlatform page (/product): headline "De novo protein design that works reliably"; ">10%" success rates for antibodies and ">50%" for miniproteins; characterization "in <2 weeks"; "Learn more about the Chai-2 series of models"; "We are opening up access to Chai-2 to commercial organizations"; academic "limited non-commercial use"; one access form (name, email, organization, type, size, intended use). Chai-3 is not mentioned on the Platform page or the homepage, and no post introduces it. Copy errors on the page include "stuctures", "antigen confirmations" and "such as like".Source
• Your websiteIntroducing Chai-2 (June 30, 2025): "double digit hit rates in de novo antibody design"; "When testing just 20 designs, the model routinely discovers viable hits"; 52 novel antigens targeted, with successful binders for 50% of tested targets; a technical report PDF.Source

In each partner, find the first successful team, run internal workshops with them for other therapeutic areas, share usage and results reports, and track the number of active teams per partner.

More teams using Chai means larger renewals and a stronger position against in-house models.

• Competing internal models

  Show combined results, as GSK describes.

• Your websiteGSK post: "Chai's models generated designs to GSK targets, zero-shot without target-specific training. GSK conducted internal wet-lab testing of the designs, finding binders to all tested targets." The Series C post: "$400 million at a $3.8 billion valuation"; "Our latest model, Chai-3, is unlocking targets that have long resisted traditional computational and laboratory methods"; partners "have started adopting our models across their organizations". The Pfizer agreement includes "Chai-3, its most advanced model".Source
• HiringPartnerships founding team (posted March 2026, still open): Business Development to "set the terms and drive deal execution with the world's leading life sciences companies" ("Deal volume and demand are accelerating fast"); Engagement Management to keep partnerships healthy, "accelerating time to clinic", and to synthesize "actionable intelligence for the rest of the company"; Forward-Deployed Scientists to act as "a power user" and "advisor to some of the most important drug programs of our time".Source

Interview partners and biotechs, then define three offers: enterprise collaborations, a biotech plan with usage-based access, and academic credits. Test the biotech plan with a few companies before publishing.

A clear structure widens the market while keeping large deals bespoke.

• Underpricing

  Pilot before publishing.

• HiringStrategy (posted September 30, 2026): "Chai's business model has no real precedent"; open questions include "how we price, which segments and products we expand into, and what makes our position durable", plus "scaling lab capacity" and "winning the war for talent". First Finance Hire (August 31): "pharmaceutical partnerships with complex revenue recognition" and "heavy spend on compute and data acquisition".Source
• Your websiteResearch & News: the GSK collaboration (October 9, 2026) "Following Successful Wet-Lab Evaluation", reported by Endpoints as "its sixth pharma partner in nine months"; Bristol Myers Squibb (August 20), argenx (July 15), the $400M Series C (July 14), Novartis (July 13), Lilly TuneLab (June 18; TuneLab "crosses 100-member mark"), Pfizer (June 4), Eli Lilly (January 9), Series B $130M (December 2025). The homepage partner strip shows Novartis, Pfizer, Eli Lilly and argenx only.Source
• Your websitePlatform page (/product): headline "De novo protein design that works reliably"; ">10%" success rates for antibodies and ">50%" for miniproteins; characterization "in <2 weeks"; "Learn more about the Chai-2 series of models"; "We are opening up access to Chai-2 to commercial organizations"; academic "limited non-commercial use"; one access form (name, email, organization, type, size, intended use). Chai-3 is not mentioned on the Platform page or the homepage, and no post introduces it. Copy errors on the page include "stuctures", "antigen confirmations" and "such as like".Source

Offer partners a documented API and agent interface to the design suite, with usage reporting, and publish a public overview page for developers.

Deeper integration raises usage and switching costs and suits partners' own automation.

• Data security concerns

  Point to the trust center and offer private deployment.

• HiringSoftware Engineer, Developer & Research Platform (posted October 12, 2026): ships "the research platform where people and agents launch experiments, compare results, and decide what to try next". Every post describes Chai as "the computer-aided design suite for molecules" with "a computational workspace to iterate on molecules from idea to medicine", used by Eli Lilly, GSK, Pfizer, Bristol Myers, argenx and Novartis.Source
• Your websiteGSK post: "Chai's models generated designs to GSK targets, zero-shot without target-specific training. GSK conducted internal wet-lab testing of the designs, finding binders to all tested targets." The Series C post: "$400 million at a $3.8 billion valuation"; "Our latest model, Chai-3, is unlocking targets that have long resisted traditional computational and laboratory methods"; partners "have started adopting our models across their organizations". The Pfizer agreement includes "Chai-3, its most advanced model".Source

Run an open challenge: published targets, designs from Chai and invited teams, lab validation by an independent partner, and results published openly. Present at a major conference.

Open, verified results earn scientific attention that no announcement can buy, and attract talent.

• Chai underperforms

  Report honestly; the field values it.

• Your websiteGSK post: "Chai's models generated designs to GSK targets, zero-shot without target-specific training. GSK conducted internal wet-lab testing of the designs, finding binders to all tested targets." The Series C post: "$400 million at a $3.8 billion valuation"; "Our latest model, Chai-3, is unlocking targets that have long resisted traditional computational and laboratory methods"; partners "have started adopting our models across their organizations". The Pfizer agreement includes "Chai-3, its most advanced model".Source
• Your websiteIntroducing Chai-2 (June 30, 2025): "double digit hit rates in de novo antibody design"; "When testing just 20 designs, the model routinely discovers viable hits"; 52 novel antigens targeted, with successful binders for 50% of tested targets; a technical report PDF.Source
• Competitor researchOthers in AI drug design, from their own homepages: Isomorphic Labs ("Reimagining Drug Discovery Process with AI"), Generate:Biomedicines ("a new kind of therapeutics company"), Nabla Bio and Latent Labs. Several develop their own drugs, while Chai positions itself as the design suite pharma licenses.See the research Source

Run a paid residency: a cohort of engineers works on real problems for 12 weeks with scientists, with offers for the best. Publish their projects.

It builds a pipeline of trained hires and a story that makes Chai legible to the AI community.

• Mentor load

  Small cohorts.

• HiringBrand and Marketing (posted September 30, 2026): "our first marketing hire", to "refine and establish the Chai brand: How we show up online and off"; audiences "range from pharma executives and scientists to AI researchers and engineers"; "making key moments count at JPM, partner events, and the AI and science conferences"; "building platforms that give our founders and researchers more reach"; "making Chai more legible and compelling to the engineers and researchers we want to hire, including those with no background in biology"; "putting a playbook in place". "This role is not about demand gen or social media maxxing."Source
• Your websiteHomepage and Careers: the two main homepage statements ("Chai powers drug discovery programs at the world's leading pharmaceutical companies." and the "Our models learn..." paragraph) are split into one element per letter with no accessible label, so text tools read "C h a i p o w e r s ...". The Careers headline reads "Join and work one the most meaningful problems AI." Contact is by email only (info@ and [email protected]).Source
• HiringStrategy (posted September 30, 2026): "Chai's business model has no real precedent"; open questions include "how we price, which segments and products we expand into, and what makes our position durable", plus "scaling lab capacity" and "winning the war for talent". First Finance Hire (August 31): "pharmaceutical partnerships with complex revenue recognition" and "heavy spend on compute and data acquisition".Source

Load the viewer only on pages with a 3D structure, defer it until visible, and remove its stylesheet elsewhere. Retest in PageSpeed.

Faster first views help executives opening news links on phones and improve how the site is crawled.

• Viewer breaks

  Test each page with a model.

• Google PageSpeedHomepage: real Chrome visitors pass Core Web Vitals (LCP 2.5 s, INP 138 ms, CLS 0 at the 75th percentile). In the mobile lab test the score is 35 (desktop 33) and the largest element appears at 13.6 s; 1.26 MB of the 1.9 MB page is a 3D molecule viewer library loaded from a public CDN on every page, with 7.6 s of render-blocking requests, 945 KB of unused JavaScript and five accessibility issues.Source

Publish a press page: fact sheet (founded 2024, funding, investors, partners, models and dates), founder bios and photos, logos, recent coverage and a contact.

Accurate facts in every story, and AI systems that summarize Chai correctly.

• Out-of-date facts

  Date the fact sheet.

• Your websiteHomepage and Careers: the two main homepage statements ("Chai powers drug discovery programs at the world's leading pharmaceutical companies." and the "Our models learn..." paragraph) are split into one element per letter with no accessible label, so text tools read "C h a i p o w e r s ...". The Careers headline reads "Join and work one the most meaningful problems AI." Contact is by email only (info@ and [email protected]).Source
• Your websiteResearch & News: the GSK collaboration (October 9, 2026) "Following Successful Wet-Lab Evaluation", reported by Endpoints as "its sixth pharma partner in nine months"; Bristol Myers Squibb (August 20), argenx (July 15), the $400M Series C (July 14), Novartis (July 13), Lilly TuneLab (June 18; TuneLab "crosses 100-member mark"), Pfizer (June 4), Eli Lilly (January 9), Series B $130M (December 2025). The homepage partner strip shows Novartis, Pfizer, Eli Lilly and argenx only.Source
• Your websiteGSK post: "Chai's models generated designs to GSK targets, zero-shot without target-specific training. GSK conducted internal wet-lab testing of the designs, finding binders to all tested targets." The Series C post: "$400 million at a $3.8 billion valuation"; "Our latest model, Chai-3, is unlocking targets that have long resisted traditional computational and laboratory methods"; partners "have started adopting our models across their organizations". The Pfizer agreement includes "Chai-3, its most advanced model".Source