At the Unleash technology summit, organized by Atlassian in collaboration with YourStory, the discussion focused on transitioning from impressive artificial intelligence demonstrations to its reliable use in production settings. While demonstrations often showcase successful task completion by an agent—such as answering complex questions or generating code—the summit concentrated on a more intricate problem: how to make AI dependable in operation.
The system must ensure result accuracy, operate within established permissions, possess visibility, security, and the ability to recover actions. Furthermore, it must function stably not only on stage but continuously across various products, among users, and within complex enterprise environments.
The journey from prompt to workflow was presented through engineering masterclasses, product presentations, and practical workshops. Atlassian teams detailed systems around Rovo and its extended AI stack, including limited contexts, durable workflows, incident analysis, intelligent security, and the infrastructure required to ensure AI reliability at scale.
Kartik Chandrasekaran, Senior Software Engineer at Atlassian, demonstrated an agent system for handling reporting requests. Previously, such requests required the involvement of support engineers who would locate necessary APIs and database structures, write and execute code, and then implement changes based on customer feedback. This process could take hours or days and was estimated to cost around $500,000 in engineering hours annually.
However, the Atlassian system allows an engineer to describe a report in natural language. The agent generates executable code, which is validated before execution via a durable workflow mechanism. The final result is compiled into a report and delivered to the client. A key engineering decision was not merely using an LLM to write code, but restricting the model's access and the content it generates. Instead of giving the agent broad API specifications, the team created narrowly specialized tools in a controlled environment, applying automated checks that establish deterministic boundaries around probabilistic output. The conclusion was clear: industrial-grade agents often require stricter operational environments rather than larger prompts. A similar principle applies to failure diagnostics.
Satish Kumar, Senior Lead Engineer at Atlassian, discussed how AI can assist in root cause analysis during software incidents. Investigations may involve code changes, feature flags, metrics, logs, traces, service dependencies, and historical knowledge. Without a clearly defined scope, an agent can quickly lose signals in general noise or draw weak conclusions. The presented AI SRE approach combined the cognitive capabilities of code, observability data, and service topology to form and verify hypotheses based on available evidence. The goal was not to replace engineers in incident response but to help them reach the correct issue faster.
As software creation accelerates with AI, security teams face pressure from another side: more changes, more services, and more solutions compete for limited attention. Lohit Mehta, Senior Lead Product Security Engineer at Atlassian, and Achyuta Anti Srinivasan, Senior Product Security Engineer at Atlassian, described the company's transition from manual review queues to an AI-powered security assessment platform. Atlassian has over 6,000 engineers working on more than 1,000 microservices, and approximately 3,000 projects pass through the organization each quarter. The platform uses signals from business priorities, service health, customer exposure, and engineering activity. AI is then applied to this grounded information to generate summaries and accelerate assessment workflows. However, the model is not treated as the source of truth; deterministic service data and human judgment remain central to security-related decisions.
As AI speeds up application development and updates, another factor arises: the volume of code entering the market, which in turn increases the risk surface. Sundar Chakraborty, Senior Product Security Engineer at Atlassian, showed that the same multi-layered approach applies to the partner ecosystem and marketplace. He detailed the company's strategy to enhance security in the Data Center Marketplace through partner verification, application signing, and an automated scanning pipeline covering software composition analysis, static testing, secret detection, and malware analysis. This pipeline runs over 10,000 scans daily and has identified over 6,700 vulnerabilities, more than 5,100 of which are classified as critical or high priority. This underscored that marketplace security cannot rely on a single control; trust must be built on multiple layers—from partner identity to the origin and content of the application itself.
Several sessions were dedicated to systems that must continue functioning after visible interaction with AI concludes. Since agents begin generating code, tasks, and documentation at machine speed, the underlying infrastructure must handle a large volume of automated activity without degrading performance, correctness, or recoverability.
Priyansh Jain, Senior Software Engineer at Atlassian, and Ammi Reddy Chirla, Atlassian Lead Engineer, shared how the company addressed bottlenecks when migrating large Jira Data Center customers to the cloud. Changes in the migration pipeline led to a sixfold increase in throughput. However, this speed revealed another limitation: data errors that could force customers to clean their source systems and restart lengthy migrations. The team reworked the process to classify and correct specific errors during data transfer without altering the source data. This approach aims to reduce migration failures by 50–60% and can save an enterprise customer between 22 and 30 days.
Kartikeyan Murthy, Lead Engineer for Jira Problem Search at Atlassian, deeply explored the architecture serving over 1.5 billion Jira Query Language requests daily for more than 350,000 tenants. The system breaks down queries into an intermediate representation and routes them through specialized services, allowing new clauses and fields to be added without destabilizing the core search path. For the largest tenants, Atlassian's architecture provided twice the performance while continuing to support a high volume of users, issues, and concurrent queries. These examples highlighted that enterprise AI is defined not just by visible features; if agents increase the speed and volume of work, then the underlying search, system, and migration systems must also evolve.
The product track examined how the process itself must change when AI shortens the distance between idea and working prototype. Tulasi Menon, Head of AI Strategy in Atlassian Product, stated that while routine tasks in the product decrease, the responsibility for defining what is 'good' becomes more crucial. She described an AI-based product lifecycle based on perception, formation, validation, release, and continuous learning. AI can compress market research and competitive analysis, and modern tools can turn ideas into working prototypes within hours or days. However, speed alone does not make a product better. Product managers must still define the right problem, set the quality bar, and decide what evidence justifies a release or rejection of an idea. Menon also warned against using AI to create polished but empty artifacts—whether long product documents or impressive prototypes that do not aid teams in making decisions. Nevertheless, release is not the end of the lifecycle. Teams need evaluations, real traces, quality criteria, and feedback loops that identify failures and improve the product over time.
This was demonstrated by Alok Agrawal, Atlassian Product Manager, and Abhinaya Singh, Group Product Manager for Jira Service Management at Atlassian. This pair showed how product managers use Rovo and Rovo Dev to gather customer feedback, synthesize information, prepare updates, and create early prototypes. The more important question was not how many tasks AI could perform, but what product managers could do with the freed-up time. Participants used Rovo Studio to configure agents for specific product workflows, moving from one-off requests to repeatable, context-aware tools. Other sessions extended this production-oriented approach across the entire Atlassian stack, from schema-defined persistence to agent analysis of long-running business processes and AI-driven product reviews. Together, they unveiled Atlassian's broader vision for enterprise AI. The company does not view the agent as a model placed atop an existing workflow; it builds the context, permissions, validation mechanisms, orchestration, security, and recovery needed for the agent to be useful in practice. A demonstration can show what an agent is capable of doing. Production shows whether an organization can trust it.
Researchers from Austria, Canada, and Switzerland reported a clinical case where a man needing a kidney transplant successfully underwent the procedure despite having antibodies against all available donors. This operation was made possible by pretreatment with bispecific monoclonal antibodies, which are used for multiple myeloma therapy. The details of this case were published in The New England Journal of Medicine.
Sensitization to human leukocyte antigens (HLA), the major histocompatibility complex, is the process of forming and circulating antibodies against any tissue from another person. This condition can arise after pregnancy, blood transfusion, or organ transplantation itself and creates a serious, sometimes insurmountable obstacle to transplanting any solid organ, even when using desensitization methods and careful donor matching. Existing treatment methods do not always yield stable and effective results.
The drug Tecilizumab belongs to the class of bispecific monoclonal antibodies. It functions by recruiting T-lymphocytes. These antibodies are capable of simultaneously binding to the CD3 T-cell receptor and the B-cell maturation antigen (BCMA), thereby prompting T-cells to attack mature B-cells and plasma cells responsible for antibody production.
Martina Schatzl from the University of Vienna and her colleagues presented the story of a 37-year-old man who suffered from end-stage renal failure and was on hemodialysis. He had already undergone two unsuccessful kidney transplant attempts and suffered from HLA sensitization, awaiting the next transplant for 12.5 years. His panel reactive antibody (PRA) level reached 100 percent, and the proportion of donors compatible with the AB0 system was zero percent, making finding a suitable donor theoretically impossible. The researchers decided to use Tecilizumab to prepare the patient for the third operation.
The patient received Tecilizumab for 31 weeks. The treatment led to a significant reduction in HLA antibodies and rapid depletion of naive and transitional B-cells, as well as memory B-cells, plasmablasts, and plasma cells in both blood and bone marrow. This depletion process was accompanied by hypogammaglobulinemia, which was compensated for by intravenous immunoglobulin administration. A decrease in AB0 reactivity and a reduction in xenoreactive antibodies to galactose-α-1,3-galactose (alpha-gal) were noted. As a result, the frequency of compatible donors increased to 1.35 percent. Soon a donor kidney was found for the man, but the surgery was postponed for technical reasons. A month later, the patient was involved in an accident on an electric scooter, which required stopping the Tecilizumab course.
Four months after the accident, another suitable organ was found and successfully transplanted. Vascular inflammation developed in the new organ, but it was not accompanied by donor-specific antibodies or complement activation product C4d. The inflammation was related to molecular markers and a slightly increased amount of non-cellular donor DNA (66 copies per milliliter of blood), and it was managed with daratumumab therapy targeting anti-CD38. Tecilizumab treatment caused a mild cytokine release syndrome, transient cytomegalovirus and BK virus viremia, mild urinary tract infections, and an increase in torque tenovirus (TTV) levels, indicating weakened overall immunity.
The presented case demonstrates the high efficacy of BCMA-targeted HLA desensitization for organ transplantation preparation. However, sustained destruction of B-lymphocytes and plasma cells can lead to decreased immunity, prolonged dependence on intravenous immunoglobulin, and slowed immune system recovery. These issues may be exacerbated when combining this therapy with desensitization targeting the D-cell antigen CD20. Further research is needed to determine maximum benefit, as well as to clarify optimal dosages and monitoring methods for safety.
Previously, German scientists published information on the successful experimental application of Tecilizumab for two patients with severe chronic autoimmune polyneuropathy. These patients showed disappearance of autoantibodies, reduced nerve tissue damage, and significant improvement in motor function.
A 45-year-old man was admitted to the hospital with complaints of progressive pain and swelling in his right shoulder, which had persisted for two months. The man, who worked on a construction site, had experienced intermittent pain in his right shoulder joint for a long time. Examination revealed swelling of the right shoulder and restricted mobility in that joint. Radiological examination showed complete destruction of the head and an unattached diaphysis of the right humerus. MRI revealed destructive arthropathy of the shoulder joint with the presence of voluminous effusion and freely floating calcified fragments, as well as tendon ruptures throughout the thickness of the supraspinatus, subscapularis, and infraspinatus tendons. John Waller from the Johns Hopkins School of Medicine and Ayşe Yülger from Hacettepe University in Ankara shared this information in The New England Journal of Medicine.
Japanese and Chinese researchers have created a copper-cobalt catalyst capable of autonomously changing its structure during the electrochemical reduction of nitrates. This reconstruction ability was a decisive factor in achieving high efficiency in ammonia production from nitrates, with a Faraday efficiency of 95.4 percent. The results of this work were published in the journal Advanced Science.
The team of Japanese and Chinese chemists, led by Takeshi Fujita from Kochi University of Technology, managed to solve these problems. They modified copper oxide by adding between 10 and 50 mol% cobalt oxide, resulting in a series of CuO/CoOX type catalysts. It was hypothesized that the cobalt component would promote the formation of active hydrogen species, while the copper component would ensure the binding and sequential reduction of nitrogen-containing compounds.
The catalysts were tested in an electrochemical cell for nitrate reduction, followed by a comparison of their performance against catalysts containing only copper or only cobalt. Structural changes during the reaction were monitored using in situ Raman, infrared, and electron paramagnetic spectroscopy. Before and after the reaction, the catalysts were examined using an electron microscope, and quantum chemical calculations were also performed.
The best performance was demonstrated by the catalyst with 10 mol% cobalt content (Co:Cu ratio = 1:9). Further increasing the proportion of cobalt led to an increase in the competing reaction of molecular hydrogen evolution, which slowed down ammonia formation.
It was found that the initial catalyst Co1Cu9OX transforms over time, forming a new functional structure. This structure includes metallic copper Cu(0), cobalt oxide, as well as oxidized copper compounds Cu(I) and Cu(II) saturated with hydroxyl groups on the surface. This rearrangement promoted faster nitrate binding on the copper and stabilized intermediate products. Moreover, the proximity of copper allowed the generated hydrogen particles to immediately interact with nitrogen for its reduction. Thanks to this reconstruction, the ammonia yield reached 54.68 milligrams per hour per milligram of catalyst, significantly higher than the figure for the copper catalyst, which yielded about 35 milligrams per hour per milligram under the same conditions.
Thus, Fujita and his team proved that the catalyst's efficiency depends not only on its initial composition but also on the ability of its surface to spontaneously rearrange during the reaction. The authors presented the concept of a proto-catalyst—a material that forms the most active configuration during operation. This approach may serve as a basis for developing heterostructured catalysts for various electrochemical processes.
