How Can AI Learn with Fewer Tokens?
Two recent systems, ACE and ALTK-Evolve, enable AI agents to learn from their own trajectories, but differ in their approach to delivering these lessons, resulting in varying token costs. ALTK-Evolve achieves similar accuracy to ACE while using significantly fewer tokens.
AI agents often hit a wall when it comes to multi-step tasks - it's not that they don't know what to do, it's just that they can't seem to apply that knowledge in a reliable way. That's where systems like ACE and ALTK-Evolve come in, allowing agents to learn from their own experiences and essentially creating a kind of memory for them. What's interesting is that both systems reject the idea of boiling an agent's lessons down to a simple summary, instead opting for a detailed, itemized record of everything the agent has been through.
The main difference between ACE and ALTK-Evolve lies in how they build and deliver these memories to the agent. ACE takes a more comprehensive approach, generating a kind of playbook through a process of reflection and curation, and then delivering the whole thing to the model every step of the way. ALTK-Evolve, on the other hand, clusters similar lessons together, merges them, and extracts guidelines with causal attribution - and only delivers a select few to the model when it's actually making a decision.
This difference in approach has a pretty significant impact on efficiency - on AppWorld, for example, ALTK-Evolve manages to achieve similar accuracy to ACE using about 40% of the tokens. And when you're working with weaker models, ALTK-Evolve actually edges out ACE in terms of accuracy, all while using about one-seventh the tokens. The secret to ALTK-Evolve's efficiency is that it only serves up the context that's actually needed, rather than dumping the entire playbook on the model every step of the way - which has some big implications for developing more efficient AI systems, and making AI agents more reliable and cost-effective in real-world scenarios.
Source: Hugging Face
Related posts
Google Cloud Launches Gemini for Healthcare to Unify Clinical Data
Google Cloud launched Gemini for Healthcare on November 10, 2026, an AI platform that unifies fragmented clinical data across EHRs, billing systems, and Medicare datasets to automate multi-step workflows like prior authorization. The platform uses prebuilt connectors for existing systems like Epic and Cerner, avoiding rip-and-replace. Launch partners HCA Healthcare, Highmark Health, and Northwell Health are testing it in preview. Google aims to reclaim the 28 hours per week clinicians
Anthropic Launches AI Office Suite for Knowledge Workers
Anthropic unveiled an AI office suite Tuesday, Claude Meetings, Sheets, and Whiteboards joining Docs, Slides, Design, and Motion, letting its assistant join video calls, manage spreadsheets, and collaborate on whiteboards. Whiteboards and Sheets are in beta for all paid users; Meetings remains limited to Teams and Enterprise plans. The launch targets non-technical knowledge workers ahead of a potential record-setting IPO.
Google DeepMind Unveils Gemini Robotics 2.5 With 96% Factory Success
Google DeepMind unveiled Gemini Robotics 2.5 on November 10, 2026, a vision-language-action model that drives multiple robot bodies including Apptronik's Apollo 2 humanoid. In live Mercedes-Benz factory trials in Hungary and Germany, the system achieved a 96% success rate on automotive kitting tasks, sorting parts from bulk bins into sequenced trays. The model handles over 200 tasks with 30-100% higher success rates and 20-30% faster cycles than prior baselines.