Add the odd reciprocals with alternating signs and a circle appears: π/4 = 1 − 1/3 + 1/5 − 1/7 + ⋯ The reason is arctan x = x − x³/3 + x⁵/5 − ⋯, evaluated at x = 1, where the angle is 45°. It converges painfully slowly. At N = 10 the error is 0.0227; at N = 1000 it is still 0.00025. Each extra digit costs ten times the work. Leibniz published it in 1674. Madhava of Sangamagrama had found it in Kerala some 250 years earlier.

Added the ability to share your OTC offers. Every offer now has its own share card. When you post an offer link on X, Telegram, or Discord, it expands into an image showing the token, trade volume, price, route between networks, and how much better the offer is compared to the market price. There’s also a Share on X button on every offer page. It prepares a post with the token ticker prefixed with $, making it easier for people searching for that ticker to find your offer. This is a useful way for projects to promote their tokens. Create an offer for your token and share it. Instead of just another link, people see an actual deal right in their feed, with the price and potential savings visible at a glance. Each post puts your token in front of people following or searching for its ticker, and gives them a clear reason to check out the offer.

Adding stablecoin settlement starts with understanding what changes in your stack. Our guide covers the infrastructure, integration costs, and tradeoffs payment teams need to evaluate before building:

Adding robotaxi cities can strengthen Tesla's business without immediately making its stock less sensitive to the AI trade. Fort Worth became the seventh driverless city in the same week Tesla shares fell about 10%. Geographic expansion is evidence of rollout progress, but it doesn't yet establish how much each vehicle earns after operating costs or how efficiently each new city scales. The next step in the investment case is turning a wider service footprint into repeatable economics. More cities matter. What Tesla earns from serving them will matter more over time. $TSLA

Add one new skill, and a different skill that used to be picked correctly stops being selected. That is what can happen as an agent's skill library grows 🧩 The usual fixes are retraining the router or collecting labeled routing errors for every skill. This paper shows that you can leave the router and the existing skills alone and rewrite only the incoming skill's name and description, once, at insertion. The authors call the failure routing interference: forgetting that happens at access time, not in model weights. 🔧 Skill++ rewrites the description so it stands apart from similar skills, adds a "does not cover" scope exclusion, and qualifies the name if it clashes. No training and no labels are needed. As the library grows from 27 to 731 skills, the share of queries where all required skills land in the top 5 rises from 36.4% to 76.3% (Claude Haiku 4.5). Errors caused by substitution fall from 46.0% to 6.5%. Skill++: Towards Lifelong Agents via Routing-Aware Skill Refinement Numbers come from the authors' own benchmark, and the study is text-only. #AIAgents# #LLM#

Addressing one of biggest remaining challenges in calcium modification today, NODULE-SHOCK adds a significant chapter in the nodular evidence compendium with excellent outcomes in both arms, as well as represents the first peer-reviewed, published data highlighting the benefits of the more deliverable #ShockwaveC2Aero#, which was associated with a low use of guide extension support (9%) in the calcified nodule arm. Thank you, Dr. @DoctorKini for presenting this important investigator-led research at #TCT2026#. Johnson & Johnson MedTech is committed to supporting research like NODULE-SHOCK that improves our understanding of the role of #ShockwaveIVL# in your treatment algorithm in even the most difficult-to-treat lesions. Learn more in @JACCJournals: US Rx only. Safety info: Product availability may vary by country.