Applications close 28 October 2026Decisions within 21 days

Show us the postmortem.

Bring the deployment, the numbers and what went wrong. Nobody pays to speak, and you keep your IP.

Score your talk the way we will.

This is the published rubric every application is scored on, out of 100. Your self-score is a guide, not a decision. Selected speakers are invited to report their own 30-day result on the results page.

You shipped, ran or lived through something real

40

In the last twelve months: a deployment, a workplace change, an organising effort or original research.

You have real numbers and know their limits

25

Time, cost, quality, errors, hours or pay, with how you measured them.

You can say what failed and what you would change

20

The tradeoffs, the surprises and the lesson someone else can reuse.

You bring a voice the circuit skips

15

Frontline, union, public sector, early career, or outside the US.

0Answer all four to see where you stand.

What every speaker gets.

Attendees pay nothing in our first year, so we do not pay cash honoraria. This is the package instead, the same for every slot.

A booth at no cost.
For you or your organisation, on the floor from 12 November. Offered to every panellist on the same terms.
The speaker rooms.
Small live rooms on the day for speakers and exhibitors only.
A permanent talk page.
Your recording and slides, indexed. You keep full rights to the talk.
An Access Fund stipend.
Frontline, union, public-sector and non-US speakers can ask for one, paid from the sponsor-funded Speaker Access Fund as funds allow.

The stage is empty. On purpose.

Nobody is announced yet, so every slot is still open. Track curators are named first, then speakers as they are confirmed, from mid-November.

The Speaker track is eight panels on one stage, two in each of four themes. Each has four seats: someone who runs the work, someone who builds the tools, someone who studies it, and someone whose work the decision changes. Applications fill the seats on the same rubric. Pick one of the 60 open topics below, or propose your own.

Know someone who should be here? Nominate them with their name and why.

60 open topics. Eight make the stage.

Every topic is a decision every organisation now has to make about AI at work, in one of four themes. None of them points a finger: each is framed as a shared problem, opened by the moderator with three questions any team can answer, and closed with a method the room can apply the following week. Each was rewritten this week against October 2026 evidence and checked by independent readers: an executive, a team lead, a non-native reader seeing the title alone, and the programme chair. The eight marked "On the stage" are planned for 10 December; the rest are open for the Expo booths, the Skill.re rooms and the series that follows. Apply for any of them, or bring one we missed.

Evaluation

Can you trust the work?

  • AI screens your candidates. From 2027, Colorado asks every employer to explain each rejection.
  • Your insurer just excluded AI losses. What evidence gets you covered again? On the stage, 9:10am.
  • Everyone's output looks great now. Who gets promoted?
  • Your AI is talking to customers right now. Who is checking its answers? On the stage, 8:15am.
  • Paying for AI by result: the six contract terms that make 'solved' mean solved.
  • Before you change a role for AI, test it on real work. The one-week method.
  • AI made the first draft free. Checking it is the new cost. How do you staff it? Reserve panel.
  • The vendor's test says it works. The one-page test you run before you sign.
  • Measure what AI changed for the business, without surveillance of staff.
  • Candidates accept AI interviews under three conditions. Which do you meet?
  • AI flags a worker. From July 2027, California asks a manager to verify first. The five-minute check.
  • Your advisers now draft with AI. What to ask before you pay for the advice.
  • When an AI agent joins the team: how to measure the team's outcome without rating people against the machine.
  • Nurses, lawyers and teachers are now paid to grade AI. What it means for your experts, and for the tools you buy.
  • When an AI agent's decision costs a customer money: the liability position to set now.

Identity

Who did the work, and who answers for it?

  • Every AI agent needs an owner, a budget and a limit. The register that gives it one. On the stage, 10:05am.
  • Disclosing AI help should not cost credit. The team rule that makes it safe.
  • Most staff already use AI tools the organisation has not yet reviewed. The 30-day plan to bring them into the open.
  • When someone leaves, their AI agents stay. The offboarding step that catches them.
  • Approvals that count: how to make human sign-off on AI real without burning people out.
  • Training an AI on your people's work: the four terms both sides can sign.
  • Customers, citizens and patients will send AI agents to deal with you. The three checks before you serve the bot.
  • Impostor remote workers reached 70 US companies this year. The hiring check that catches them and treats real candidates fairly.
  • Your AI assistant has a company credit card. Who approves before it spends?
  • AI content and chatbots must carry a label in the EU and in several US states from 2026. One label standard for every market.
  • Help-desk resets are the new front door for fraud. The verification step that closes it.
  • Prove the vetted person is doing the work, without watching them. The presence standard for remote teams.
  • The CFO on the video call was a deepfake. What should a 'yes' to a payment now require? On the stage, 11:00am.
  • AI now pays your invoices. Who checks the supplier is real before money moves?
  • AI tools now verify your people's work history. Who holds the record, and how does a worker correct it?

Collective Intelligence

How does a team keep learning?

  • AI does the junior work. How does anyone become senior? On the stage, 1:15pm.
  • Hiring by referral is back. How to keep it fair and open to people with no network.
  • People now ask the AI before a colleague. The weekly habit that keeps judgement moving between people.
  • Half-finished AI work is the new hidden cost between teams. The standard that stops it, and what it saves. Reserve panel.
  • When nobody does the task by hand any more, who can check the AI? The safeguard.
  • Every worker gets an AI coach. What is a manager for?
  • Every meeting is recorded by AI. The three-tier rule that turns recordings into learning, lawfully.
  • Your most experienced person retires in 18 months. Can AI keep what she knows, on her terms? On the stage, 2:10pm.
  • Nobody quits, nobody hires. How do people move in a frozen job market?
  • The one-person company run on AI agents: what works, what it costs, and what it still needs from other people.
  • Courses completed, skills missing. How peer groups teach AI faster, and how to fund them.
  • Human review helps in some steps and hurts in others. The map that tells you which.
  • Your team's best AI methods live in 40 private chats. The routine that makes them shared practice.
  • Everyone brainstorms with the same AI models. What makes your organisation's judgement different?
  • AI can question a thousand experts overnight. What expert knowledge is worth now, inside and outside your organisation.

AI-Ready Work

How does the job change?

  • AI tools arrived. The work didn't leave. Who owns the redesign, and what gets removed? On the stage, 3:05pm.
  • Every extra AI tool costs attention. The stop rule that protects your team. Reserve panel.
  • AI adoption moves at the speed of managers. The objective and the hours that let them lead it.
  • When your best expert says no to AI: keep the judgement, change the practice.
  • 71% of workers expect fewer jobs. What can leaders promise, and keep? On the stage, 4:00pm.
  • AI skills pay 62% more in job ads. How to pay for them fairly inside your company.
  • When a competitor is rebuilt around AI: what to protect in your own quality, and how to compete.
  • Employers want fewer prompt-writers and more AI builders. What to retrain for in 2027.
  • AI usage is now a cost line per person. The budget rule for what each role may spend.
  • Most of your frontline workers have no company email. Where is their AI? Reserve panel.
  • 'AI fluent' is in 11 times more job ads. How to define it and test it fairly.
  • Handing an AI agent high-stakes work: the three numbers to watch before and during, at any size.
  • Buying services delivered mostly by AI: who checks the work, and how to decide between suppliers.
  • The vendor's people run your AI rollout. What your own team must own before they leave.
  • Who certifies AI skills for the 97% who do not write code? Which credential earns trust, and a raise.

The organiser's family includes eval.qa (an early-stage testing project), manav.id (an early-stage project on proving a person is human) and Skill.re (the workshop partner). No one from any of them sits on a panel; their topics are disclosed where they come up.

Apply to speak.

About ten minutes. Short answers are fine. A person reads every application.

Read the speaker guide: what we look for and how talks are chosen.

When do I hear back?

Applications close 28 October. Every applicant gets a decision with written feedback within 21 days.

Are speakers paid?

We do not pay cash honoraria in our first year. Every speaker gets the same package: an expo booth at no cost, the speaker rooms, a permanent talk page and full rights to their talk. Frontline, union, public-sector and non-US speakers can ask for a stipend from the sponsor-funded Speaker Access Fund.

I have not shipped an agent. Can I still apply?

Yes. Workers, researchers, public-sector practitioners and organisers can apply with firsthand experience of how the work changed. The rubric rewards evidence and honesty about limits, not job titles.

Form not loading? Open it in a new tab.

Nominate a speaker.

Know someone with firsthand evidence who should be on the stage? Tell us who and why.

Form not loading? Open it in a new tab.