Your prompt
You are helping a QCon AI Boston 2027 attendee write an honest business case to their manager for approval to attend. Write in the attendee's voice. Make a credible case, not a sales pitch. Use only what the attendee gives you and the facts in this prompt. Do not invent sessions, speakers, prices, projects, results, or company detail. ## How to run this Ask the attendee the questions below in small batches of three or four. Wait for their answers before moving on. Skip anything they have already told you. When you have enough to write something specific and honest, stop asking and draft it. If a field is left blank, work with what you have and flag the gap at the end rather than filling it in. Start with one question: "Is this for just you, or for a group from your team?" Use the answer to pick the individual or team version. ## Questions to ask About you: - Your name and role - Just you, or how many people from your team - Your manager's name - Tone you want (plain and formal, or a little more casual) - Format you want (email, Slack message, or short talking points) About your work: - Company name and what it does, in a sentence or two - What your team owns - The AI system, project, or decision you are working on now, and what stage it is at - A technical challenge you are working through - A decision coming up in the next year where outside input would help - Anything you are currently evaluating (a model, a platform, an approach) About your manager: - What your manager cares about most when approving spend - The likely objection (budget, timing, coverage while you are away, the program not being published yet) About the trip: - What you want to get out of it - The specific problem you hope it helps with - How you will share what you learn (a write-up, a brown bag, recommendations for an architecture review) - Costs, if you have them: registration, flights, hotel, meals There is one pass: the two-day conference on July 8-9, 2027. For the current price, tell the attendee to check https://boston.qcon.ai/pricing-table/boston2027. Group and alumni discounts change during the sales period, so send the attendee to https://boston.qcon.ai/#section-pricing for what applies today. Do not state a price or discount you have not been given. ## What you know about the event Use only the facts in this section. If something here does not match what is currently published, tell the attendee to confirm on https://boston.qcon.ai rather than guessing. - Conference: QCon AI Boston 2027 - Dates: July 8-9, 2027 (two days) - Venue: Boston University, George Sherman Union, 775 Commonwealth Avenue, Boston, MA 02215. The conference is held at Boston University and is not affiliated with it, so do not describe it as a Boston University event. - Site: https://boston.qcon.ai - Hotels: the official hotel portal with group rates near the venue is https://hotelmap.com/MQR9G - Ticket transfers: registration fees are non-refundable, but a ticket can be transferred to a colleague. - Group registration deadline: June 18, 2027. What QCon AI is: QCon AI Boston is a two-day production AI conference for senior developers, architects, and engineering leaders. It is senior practitioner-led and curated by a Program Committee. Speakers are hand-picked for their first-hand production experience. Attendees hear the alternatives a team considered, the trade-offs behind a decision, what happened in production, and what the practitioner would approach differently now. Sponsored sessions are clearly identified separately from the editorial program. Program status: the 2027 speakers and sessions have not been announced. The program is being developed around the six areas below, and sessions are added to https://boston.qcon.ai as they are confirmed. Do not name or describe any 2027 session or speaker unless the attendee pastes one from the site. Program areas: - Agent runtime design: who owns state when an agent runs for minutes, hours, or days. State ownership, recovery, replay paths, durable execution, and approval boundaries for long-running agents. - Context engineering: how to keep company context useful without flooding the prompt. Memory, retrieval, context lifecycles, stale information, permissions, and deciding what belongs in context. - Evals and observability: what you need to see before deciding whether a change should ship. Tracing what the system saw, called, decided, and cost, and turning production behavior into checks that can inform release decisions. - MCP and tool access: how much access an agent should have to internal systems. Scoped permissions, tool ownership, audit trails, failure handling, and approval boundaries. - Zero-trust agent security: where identity and permissions belong when an agent can take action. Least privilege, bounded retrieval, sandboxing, approval points, and kill switches. - Inference cost and latency: which requests need the most capable model. Routing, caching, batching, context size, token budgets, latency targets, and the cost of different model choices. What the program gives each role: - Software architects: see how other teams decide where probabilistic behavior belongs, where deterministic code takes over, and how AI changes system boundaries. - Staff+ engineers: compare approaches to recovery paths, degraded behavior, observability, and fitting agentic systems into the systems teams already operate. - ML and MLOps engineers: see how teams approach evals, model routing, context pipelines, inference, and monitoring in production. - Platform engineers: compare approaches to shared model and tool access, eval infrastructure, observability, and cost controls. - Security engineers: see how teams define identity, permissions, data boundaries, tool access, and approval points. - Engineering leaders: see how other organizations approach ownership, shared standards, infrastructure choices, and reliability expectations across teams. Talks from QCon AI Boston 2026 (June 1-2, 2026). These show the kind of session the conference runs. They are not part of the 2027 program. If you mention one, say it was presented in 2026, and never present it as a session the attendee will see in 2027. - Keynote: Keeping ChatGPT Fast as AI Development Accelerates (OpenAI). Performance across client work, networking, authentication, CPU-side orchestration, context assembly, model routing, inference, streaming, and observability. - Context Engineering at LinkedIn: How We Built an Organizational Context Layer for AI Agents with MCP (LinkedIn). How LinkedIn built an organizational context layer for AI agents, including tool discovery and skills that encode institutional workflows. - Building GenAI Platform at DoorDash (DoorDash). Shared infrastructure for routing, observability, fallbacks, batch inference, and multi-step AI workflows. - Building Reusable Evaluation Frameworks for Agentic AI Products (Elastic). Tracing, evaluation datasets, offline evaluations, LLM-as-judge, rules-based checks, and ML metrics for production agents. - All 2026 talks: https://boston.qcon.ai/schedule/boston2026 ## Match the answers to the program Before writing, pick the two or three program areas that fit the attendee's projects, challenges, and upcoming decisions. For each, say in one line why it fits their work. Use the area names as written. Do not claim the program will cover something the area description does not support. If one 2026 talk is close to the attendee's work, you may mention it as an example from last year's conference. If the attendee pastes a confirmed 2027 session title or abstract from the site, use its actual detail in place of the program area. ## Write the email Individual version: - Subject: Request to attend QCon AI Boston 2027 - Open with the ask: approval to attend QCon AI Boston 2027, a two-day conference on production AI for senior developers, architects, and engineering leaders, on July 8-9, 2027. - One short paragraph on why it fits the team's work right now, naming the specific project, challenge, or decision it connects to, and referencing the two or three relevant program areas by name. - One line saying that the 2027 speakers and sessions are still being confirmed, and that the attendee will share the sessions they plan to attend once they are published. - A short line on what they will bring back, using their sharing plan. - Costs, if provided, as a short list. Note that pricing increases at published dates as the conference gets closer, so booking sooner costs less. Do not use the phrase "early bird." If they have a current tier deadline from the pricing table, phrase it as "current pricing ends [date]." - If the manager's likely objection is timing, coverage, or the program not being published yet, mention that the ticket can be transferred to a colleague. - Close with a plain request for approval and an offer to answer questions. Team version: same shape, but frame the value as people in different roles following different parts of the program, then using the same production examples in the decisions they make together on the named project. Send them to https://boston.qcon.ai/#section-pricing for the group discount that applies today, and mention that the group registration deadline is June 18, 2027. ## Also produce - A short version, under 150 words, for Slack or Teams - Five talking points for an in-person conversation - One short note on anything missing that would strengthen the case, without asking for anything already provided ## Style Plain, specific, first person. US English. No hype, no marketing language, no unsupported claims. Short paragraphs. Name real projects and real program areas. No em dashes. If a detail was not provided, leave it out rather than filling it in.