Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
No official QCon AI New York 2026 event page or matching listing for a session titled “Production Patterns for Agents, Evals, and LLM Costs” has been verified. The speaker, schedule, abstract, and event date therefore remain unconfirmed. Separate conference-program entries discuss related production issues, but they are not evidence of what this QCon session will cover.
What is—and isn’t—confirmed
As of October 13, 2026, the available verification does not establish that this exact session appears on an official QCon AI New York 2026 schedule. It also does not establish who might present it, when it might run, or what its abstract might say. Those details should not be treated as confirmed event information.
This matters because the title sounds like a specific technical talk, but the evidence supports only a broader set of production topics from other conference programming. There is no verified speaker quote, session description, statistic, or recommendation that can responsibly be attributed to the named QCon session. In particular, the available material does not establish any concrete LLM-cost technique, pricing comparison, or savings figure for it.
Related production themes from other conference sessions
Separate Linux Foundation program entries provide useful context for the kinds of operational questions that can arise when agents move into production. They are from AGNTCon/MCPCon North America programming, not the unverified QCon session.
#1 Best Overall
Evaluate more than the final answer
One separate session description presents evaluation at several levels: the final outcome, the agent’s trajectory, tool accuracy, and intermediate sub-goals. It also mentions deterministic scorers, explainable LLM judges, and production A/B tests. That framing is useful because a system can reach an acceptable final result while taking an unreliable path, using tools incorrectly, or failing repeatedly on an intermediate step. Teams considering agent evaluations can use these levels to decide what they need to measure; the program entry does not establish that the QCon session recommends a particular evaluation design.
See the separate Linux Foundation program entry on agent evaluation.
Use traces to investigate changing workflows
Another program entry describes tracing and evaluation of agent workflows, including attention to context usage and output formats as models and harnesses change. The practical implication is that a score alone may not explain why an agent’s behavior shifted: trace data can help expose what happened during a workflow. The entry is a description of a different session, not a report of findings from QCon.
See the separate Linux Foundation program entry on tracing and evaluation.
Self-hosting adds operational tradeoffs
A third session summary names routing, retries, context management, quantization tradeoffs, observability, and failure recovery as issues in self-hosted agent systems. These are operational concerns rather than confirmed QCon talking points. The summary does not provide benchmark results, quantify tradeoffs, or establish which hosting approach is best for a particular workload.
See the separate Linux Foundation program entry on self-hosted agents.
What the “LLM costs” part of the title can support
The title signals an interest in LLM costs, but the available information does not substantiate what the session might say about them. It supplies no verified cost model, usage data, vendor or model comparison, price, or cost-reduction result tied to the session. Context usage appears in a separate conference description, but that alone is not enough to claim a particular cost-control method or outcome.
Readers looking for a cost recommendation should therefore wait for a verified abstract or speaker materials rather than infer one from the title. A related topic in another program can suggest a question to investigate; it cannot establish the answer or the QCon session’s intended coverage.
Best Value
How to verify the session before relying on event details
- Look for the exact title on an official QCon AI New York event page or schedule, rather than relying on a similarly named talk elsewhere.
- Confirm the speaker, date, and abstract on that official listing before quoting or sharing them.
- Keep descriptions of the Linux Foundation sessions separate from any QCon coverage unless an official source explicitly connects them.
- For any claimed cost advice, look for the speaker’s actual materials and the assumptions behind any figures, such as model, workload, and measurement period.
Until an official listing or primary session material is available, the responsible account is limited: the exact QCon session details are unverified, while related agent-evaluation, observability, and self-hosting themes are documented in separate conference-program entries.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
