Know Your Agent · Virtual working session · Closed-door · By invitation
An invitation-only working session to shape how the industry trusts AI agents
AI agents already log in, open accounts, and shop for your customers. Attackers send agents too, and theirs look the same. Incode's CTO and CPO will share what we've learned and built, then spend the second half hearing how your teams see it.
- Monday, November 16, 2026
- 12:00 to 3:00 pm ET (9:00 am to 12:00 pm PT)
- Virtual · Five company seats · Bring your team
Hosted by
Jovan JovanovicChief Technology Officer, Incode
Roman KarachinskyChief Product Officer, Incode
Why now
Your login checks what an agent has. It needs to know who sent it.
KPMG found that 81% of companies have already faced agentic fraud. Block agents, and customers take their business elsewhere. Let them in blind, and fraud gets in with them.
A password, a device, and a one-time code can all be stolen or handed to an agent. The one proof an attacker can't steal is a live person. Incode verifies that person and ties each agent to them.
Source: KPMG
“You've seen the demo. Now we want your criticism, while it can still change what we build.”
How this session works
A technical discussion among peers, with three ground rules
We show the research and the build as they stand, including open proposals. If you want to keep building with us afterward, we'll make that easy.
Nothing is recorded or published. Within a week, attendees receive shared notes that capture what was said without naming who said it.
Five company seats go to the first five companies to confirm, and each company can bring several people. After that, we start a waitlist. We don't admit press.
Agenda · All times ET
We present, then you lead
In the first half, we share our research, our design choices, and what we built. In the second half, you lead: the signals you trust, the protocols you use, and what you'd change.
First half · What we have learned and built
12:00 to 1:25 pm · Incode presents
-
12:00 pm01 · 25 minutes
What the research says
How agents reach your site, which signals disappear when the session is an agent, and how fraud blends in with legitimate automation.
-
12:25 pm02 · 30 minutes
How we designed it
The design decisions behind the approach, including the ones we rejected: why detection alone falls short, how a verified human anchors every agent credential, and where your policy fits.
-
12:55 pm03 · 30 minutes
What we built
A live walkthrough of the SDK, the verdict it returns, and the step-up path to a verified human.
Second half · What you think
1:35 to 3:00 pm · Guests lead, after a 10-minute break
-
1:35 pm04 · 30 minutes
The signals you care about
Which agent-session signals your teams would trust, which they'd ignore, and which false positives you can't afford.
-
2:05 pm05 · 25 minutes
The protocols you are using or planning
Where you've landed on agent protocols and credentials, such as MCP, agent-to-agent, delegated authorization, and mandates, and what you need the industry to settle first.
-
2:30 pm06 · 30 minutes
Your feedback on what we built
What would break in your environment, what's missing before you'd integrate, and what you'd change in the specification.
We end at 3:00 pm ET. Topics within each half may shift with the conversation.
What we want to learn from you
Questions for the second half
We'll send these with the pre-read so you can prepare.
Which agent-session signals would you act on, and which would you ignore?
We'll bring our own ranked list to compare with yours.
Which agent protocols and credential formats are you building on, and which have you ruled out?
For example MCP, agent-to-agent, delegated authorization, or mandates.
What would you need before you integrate this?
Missing pieces, policy conflicts, or anything your platform team would push back on.
How do you tell a trusted agent apart from a trusted person acting through one?
We have a working answer and want to test it.
Who is in the room
The people who decide what an agent must prove before it transacts.
Fraud, identity, platform, and security leads from large US banks, carriers, fintechs, and marketplaces, plus the companies building the agents themselves. Everyone in the room is deciding what an agent has to prove before it can log in, move money, or change an account.
Your seat covers your company, so bring the people who would run this: fraud strategy, customer identity and authentication, platform or API security, and bot and agent detection. Everyone you list gets the calendar hold and a short pre-read.
What you take with you
- A path to SDK access and integration documentation
- The detection benchmark, with method
- The draft specification for agent credentials and mandates
- A direct line to the engineers who built it
- Shared notes within a week
By invitation only · agentic@incode.com
Seats go to the first five companies to confirm. Tell us who's coming.
One confirmation covers your company, so bring your team. After five companies confirm, we start a waitlist. Not sure it's right for your team? Email agentic@incode.com or reply to whoever invited you.